Themes in Kant's Metaphysics and Ethics (Studies in Philosophy and the History of Philosophy)

THEMES IN KANT’S METAPHYSICS AND ETHICS

STUDIES IN PHILOSOPHY AND THE HISTORY OF PHILOSOPHY General Editor: Jude P. Dougherty

Studies in Philosophy and the History of Philosophy

Volume 40

Themes in Kant’s Metaphysics and Ethics Arthur Melnick

THE CATHOLIC UNIVERSITY OF AMERICA PRESS Washington, D.C.

Copyright © 2004 The Catholic University of America Press All rights reserved Printed in the United States of America The paper used in this publication meets the minimum requirements of the American National Standards for Information Science—Permanence of Paper for Printed Library materials, ANSI Z39.48-1984. ∞ library of congress cataloging-in-publication data Melnick, Arthur. Themes in Kant’s metaphysics and ethics / Arthur Melnick. p. cm. Includes bibliographical references. isbn 0-8132-1371-1 (alk. paper) 1. Kant, Immanuel, 1724–1804—Contributions in metaphysics. 2. Metaphysics. 3. Kant, Immanuel, 1724–1804—Contributions in ethics. 4. Ethics. I. Title. B2799.M5M45 2004 193—dc21 2003011984

Contents

Introduction

vii

PART I. THE TRANSCENDENTAL AESTHETIC 1. The Consistency of Kant’s Theory of Space and Time 2. Kant vs. Lambert and Trendelenburg on the Ideality of Time

3 21

PART II. THE TRANSCENDENTAL DEDUCTION 3. Apperception and the Premise of Kant’s Transcendental Deduction 4. Categories, Logical Functions, and Schemata in Kant 5. A Modified Version of Kant’s Theory of Cognition PART III. THE PRINCIPLES 6. Kant’s Proofs of Substance and Causation 7. Kant’s Refutation of Idealism in the B Edition PART IV. THINGS IN THEMSELVES 8. Kant on Things in Themselves 9. Kant’s Proof of Transcendental Idealism in the First Antinomy 10. Macroscopic Facts, Quantum Mechanics, and Metaphysical Realism

35 47 69

95 125

147 164 185

PART V. FREEDOM AND MORALITY 11. Reason, Freedom, and Determinism in the Third Antinomy 12. Kant’s Formulations of the Categorical Imperative 13. The Rational Justification of Morality

229 249

Bibliography

269

Index

273

205

vi

recto

Introduction

This work is a collection of ten essays on Kant’s theoretical philosophy and three essays on his practical philosophy. All of the essays have been written expressly to be read independently of the others. Together, however, the first ten essays I believe constitute a unified and fairly comprehensive account of Kant’s views in the first Critique. Most of the essays attempt to set out Kant’s own views. Two of them (Essays 5 and 13), however, are reconstructions of Kant-like positions, while one (Essay 10) aims to reject one of Kant’s doctrines. The first two essays set out Kant’s theory of space and time. My contention is that Kant is a constructivist, believing, for example, that space itself exists entirely in constructive procedures of flowing motions, which include our moving around and about. These procedures are activities of the subject by which it obtains sensation. Thus, if one is to be affected, or to experience certain sensations, one has first to move so as to make the object present. The object’s spatial location is then given exclusively in terms of what motion makes it present (how far along in moving about the object is). A similar constructivist account applies to the nature of time. In Essay 1 I argue that this constructivism is Kant’s theory not only in the Aesthetic but throughout the Critique. In Essay 2 I claim that it explains how even the time of my own succession of conscious states is “ideal.” In Part II Essays 3 and 4 provide an overview of Kant’s theory of cognition, or of how we represent objects in a world that exists spatially and temporally way beyond our actual experience. Cognition or thought, for Kant, is a unity both of consciousness and of intuitions. Now, a rule governing what I am doing unifies my consciousness in performing the activity, and it also unifies the spatio-temporal activities that govern the attainment of intuitions. Essay 3 argues that, indeed, thoughts or cognitions for Kant are literally rules for attaining intuitions. Essay 4 then contends that the transcendental deduction is a recipe or method for how to deduce pure concepts; viz., how to show that they apply to all cognizable reality. This method is to show that these concepts are required for cognition itself; i.e., for there being vii

viii

introduction

rules that encompass a full range of spatio-temporal activity encompassing in turn all possible intuition. The Transcendental Deduction, then, sets out a specific account of what cognition is or has to be, and the method for deducing categories it opens up is specifically in terms of this theory of cognition. In Essay 4 I argue further that the B edition Deduction presents the very same method, only in the B edition the concepts whose function is to provide for rules are also characterized as providing for the applicability of propositional judgment forms. Essay 5 presents a Kantian theory of cognition as rules but goes beyond Kant, first in giving a naturalistic account of what rules are, and second in giving an explicit, real (as opposed to nominal) definition of truth for such thoughts or cognitions. In Part III Essay 6 applies the method of the Transcendental Deduction to establish the applicability of the pure concepts of substance and causation. Without substance there could be no propriety of coming in upon activities or procedures in the middle of things, and hence no rules (cognitions) pertaining to past time. Without causation there would be no rules for the necessary advance in the order of time from earlier to later. In the seventh essay the result regarding the applicability of the concept of substance is used to provide a defense of Kant’s refutation of Cartesian skepticism (the view that there might not be a material world). It is only in relation to Kant’s contentions that space and time exist in activities, and that cognitions are simply rules governing these activities, that the arguments for substance and causation and against skepticism make any sense. Essay 8, which begins Part IV, interprets and defends Kant’s idea that we cognize only phenomena; i.e., the objects we cognize have no intrinsic existence that goes beyond their being presentable in the course of allowable spatio-temporal construction. Their whole existence, then, is as appearing objects, and not as objects existing in their own right outside potential events of presentation. Essay 9 shows how the First Antinomy proves this contention that we cognize mere phenomena. Essay 10 argues, against Kant, that cognizability is not an absolute limit to theoretical reason. Mathematical equations, though not adequate for cognition, can give us a structure for reality beyond cognizability. Part V concerns Kant’s practical philosophy. Essay 11 argues that for Kant a decision is free if it is governable by reason, and that no causally determined decision can be thus free, since practical reason is incompatible with causal determination. I then give an interpretation of Kant’s claim that nevertheless free decisions are possible even with respect to a causally determined universe. Essay 12 argues that Kant’s categorical imperative is essentially a version of a social-contract theory of

introduction

ix

moral laws. This explains the equivalence of the various formulations of the imperative, as well as why respect for my own reason entails respect for all other rational agents. Essay 13 rejects Kant’s account of the motivation for obeying moral laws, and offers instead an account of moral laws distinct from social-contract theories according to which being moral is the only reasonable way to be if one wishes to be happy. These essays are designed so that each may be read independently of the others. Thus, if one were interested in Kant’s answer to Hume on causation, all the elements of Kant’s views required for understanding his answer are reviewed within Essay 6. This method leads to a good deal of repetition. For those who would read these essays consecutively as a commentary on Kant, the reviewed elements may simply be skimmed over or skipped. Five of the essays appear elsewhere in slightly modified form. Essay 2 appeared in the History of Philosophy Quarterly 18 ( January 2001). Essay 4 appeared in the Review of Metaphysics 54 (March 2001). Essay 5 appeared in the International Journal of Philosophical Studies 9 (November 2001) and is also available at http://www.tandf.co.uk. Essay 6 is to appear in The Cambridge Companion to Kant, 2d ed., edited by Paul Guyer (Cambridge: Cambridge University Press), and is reprinted with the permission of Cambridge University Press. Essay 12 appeared in KantStudien 3 (2002). I am thankful to these publishers for permission to use these essays.

PART I THE TRANSCENDENTAL AESTHETIC

1

The Consistency of Kant’s Theory of Space and Time

In the Aesthetic Kant holds that space and time are given in pure intuition. In the Transcendental Deduction he holds that they are due to a synthesis of the transcendental imagination. Since a synthesis is a putting together, this seems to contradict the view of the Aesthetic that space and time are simply given. At least, in both these sections, Kant holds that space and time are pure manifolds. However, in the Analogies he holds a “dynamical” conception of space and time, according to which objects (appearances) “determine for one another their position” (A200, B245, p. 226).1 This apparently relational account of space and time seems to contradict the account of them as pure manifolds in both the Aesthetic and the Deduction.2 In this essay I want to argue that Kant holds a constructivist theory of space and time in the Aesthetic, and that this constructivist theory is consistent with what he holds both in the Deduction and the Analogies.

1. the aesthetic Space and time, for Kant, are given in pure intuition. Now, intuition, for Kant, is a representation in which the object (that which is represented) is given to us, and an intuition is pure if it does not involve sensation (A19, B34, p. 65–A200, B34, p. 66). Thus, for space and time to be given in pure intuition means they must be given or be immediately present without sensory affection by objects. One way for something to 1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. 2. For the supposed discrepancies in Kant’s account as between the Aesthetic and the Analytic, see, for example, Hans Vaihinger, Commentar zu Kants Kritik der Reinen Vernunft, vol. 2 (Berlin: Union Deutsche Verlagsgesellschaft, 1892), 224, 227–29; Norman Kemp Smith, A Commentary to Kant’s Critique of Pure Reason, 2d ed. (London: Macmillan, 1918), 45; Peter Strawson, The Bounds of Sense (London: Methuen, 1966), 123; T. K. Swing, Kant’s Transcendental Logic (New Haven: Yale University Press, 1969), 151–52; and Robert Paul Wolff, Kant’s Theory of Mental Activity (Cambridge: Harvard University Press, 1963), 228, 244, 263.

3

4

t h e t r a n s c e n d e n ta l a e s t h e t i c

be given or present without sensation is for us to produce it and thus present it to ourselves. Thus, if I draw or construct a line, an actual individual expanse is thereby made present without sensory affection by objects. If we take expanses of space to exist in such constructions, we have one way of understanding how there can be pure intuition (viz., non-sensory presentation). A constructivist view, then, is at least consistent with Kant’s characterization of space and time as given in pure intuition, so long as we note that it is by our constructing that they are thus given. By a constructivist view of space and time, I mean a view parallel to the constructivist account of numbers in mathematics, according to which numbers exist only in procedures; say as termini of possible countings. What I wish to argue is that such an account of space and time makes sense of Kant’s arguments and conclusions in the Aesthetic.3 Kant’s first argument regarding space in the Metaphysical Exposition is that in order to represent things “as not only different, but in different places, the representation of space must be presupposed” (A23, B38, p. 68). Kant clearly means that some non-sensory representation of space is presupposed for representing objects as spatially related. But why should this be so? Why can’t I just see that one object is “outside and alongside” another? If we presume that part of what is involved in two objects having a spatial relation to each other is that they are separated from each other by a continuous expanse of space, it is no longer obvious that one can “just see” one object as outside another, for it is not obvious that one can sense continuity. On the other hand, suppose I look at the first object and then sweep my attention along to the second object. This sweep or flow of attention is something I do (something I produce). Equivalently, it is a mental act or performance “added” to sensation, not something given in sensation. It is in this flowing performance, I suggest, that Kant thinks continuity exists. This is the non-sensory representation that is presupposed in sensing one object’s being outside or alongside another (separated by an interven3. Our constructivist account contrasts with cognitive processing interpretations according to which space (and time) are cognitive processes that organize our sensory states. In these views this is either all that space is (in which case only our sensory states are in space), or else space also exists corresponding to such processing. The former makes space a secondary quality, like color (although at a perhaps deeper level of processing), which is a view Kant rejects, while the latter has space also existing objectively (presumably as either relational or substantival), which Kant also explicitly rejects. The constructivist view developed in this essay is an ontological claim as to what space is, which makes space applicable to what is outside any states of the subject, but without making space into either relational or positional characteristics of what is outside the subject. For a cognitive processing view, see Patricia Kitcher, Kant’s Transcendental Psychology (New York: Oxford University Press, 1990).

Kant’s Theory of Space and Time

5

ing continuous expanse). The separation by a continuous expanse is the separation by an intervening flowing construction such as a linear shift of attention. Consider now a nonvisual apprehension of spatial relations, say, by a blind person. First his hand feels (senses) one object. Then his finger shifts along linearly (in a tactile shift of attention) to a second object. Once again, a flowing performance (drawing out a line with his finger) intervenes between the tactile sensations, and once again it is in such a seamless construction that continuity (seamless expansiveness) exists. Thus, sensing one object as separated from another in space (and thus as being in a different place) presupposes an intervening flowing construction. It is for this reason that “The representation of space cannot, therefore, be empirically obtained from the relations of outer appearance” (A23, B38, p. 63). Continuity, that is, is not abstracted from sensed relations, but exists in the activity (the shift or flow of attention) by which spatial relations are first apprehended. Kant’s second argument regarding space is likewise directed against the idea that space is a matter of sensed relations. A relation cannot be sensed without items that are in that relation. Thus I can sense that one color patch is darker than another, but I cannot sense the darker-than relation itself in the absence of patches. On the other hand, Kant contends, I can represent space as empty of objects. Once again, suppose that from an initial object I sweep out my attention. Then I am representing a spatial expanse or distance even if I do not come upon a second object. Indeed, I can sweep out my attention whether or not there is an initial object. If space is constructed rather than sensorily apprehended, then, I can represent an individual expanse of space in the absence of objects. The first two arguments are directed against the idea that space is a matter of sensed relations, showing therefore that its representation is non-sensory, or pure. The third and fourth arguments are directed against the idea that space is a matter of conceptualized relations, showing therefore that its representation is non-conceptual, or intuitive. The third argument turns, I believe, specifically on the continuous nature of spatial expansiveness. Any spatial expanse is a whole in the sense that it embraces parts. However, spatial expansiveness is not constituted out of parts as elements that constitute its expansiveness. The whole expanse, we may say, is prior to parts. This is the seamlessness or continuity of the expanse; viz., expansiveness is so seamless that it cannot be constituted out of a multitude of elements at all. Now, for discursive conceptualization, a whole is always represented as being a multitude of elements. Its very nature as a whole is that it exists by and only by all its elements’ existing. If so, there can be no conceptual representation

6

t h e t r a n s c e n d e n ta l a e s t h e t i c

of the continuity of spatial expanse. On the other hand, in a flowing construction, such as drawing out a line, the whole is prior to the parts in the sense that the drawing out is not composed of stops and starts that would produce parts. Rather, segments or parts of the line are, as Kant says, “limitations”; viz., cuts imposed by further construction subsequent to the production of the line. Continuity, therefore, exists in and only in flowing construction.4 In a historical vein we can say that Kant is following Newton’s fluxion account of continuity as flowing quantities, rather than Leibniz’s compositional account in terms of infinitesimals. Of course Leibniz himself rejected this account, demoting it to a mere calculus. Since for Leibniz what is real is what is conceptually representable, and the conceptual representation of a whole is constituted by individual elements, and since an expanse cannot be thus constituted, Leibniz simply rejected continuous expanse. Kant, agreeing with the second two of these Leibnizian doctrines, rejected rather the idea that what is real is what is conceptually representable; viz., continuity or expansiveness is real but not conceptually representable. It is real in flowing construction. Note that this third argument of Kant’s allows him to conclude not only that space (expansiveness) is represented in pure intuition but that space exists only in pure intuition. Whatever else, space is expansive or continuous. Since continuity exists only in flowing construction, therefore space exists only in flowing construction (that is, only in a construction that non-sensorily presents the expansiveness, and so only in pure intuition). The arguments of the Metaphysical Exposition all together are directed against a relational theory of space. Space is not a matter of actual sensed relations (the first two arguments), nor a matter of possible conceptualized relations (the last two arguments). The result holds as well, I suggest, against an absolutist theory of space. The reason is that an absolutist theory, like a relational theory, is objectivist. Now, an objectively existing whole, such as an expanse, is such that it actually exists only if all its parts actually exist, since if any of its parts fail to exist, it too fails to exist. One might think that an objectively existing expanse or stretch only potentially has parts or elements that are actualized only upon our constructions. However, the constructions surely cover subexpanses that are actual (not merely potential) and so the objective expanse still has all actual components that make it up or constitute it. 4. Note that in this view motion (the flowing construction of the subject) is more fundamental than spaces “traversed by” motion. Parting or segmenting is an imposition subsequent to the motion, not a component of it. Motion, that is, isn’t expansive by traversing (all) the components of expansiveness. In this way Zeno’s paradoxes are avoided.

Kant’s Theory of Space and Time

7

Subdividing as a performance is subsequent to the original flowing or sweeping performance, and so the former can exist actually, while the latter does not yet exist (or exists only potentially as further performance). By contrast, with an objective reality there is no sequentiality to its existence (it is “all there”) and hence no mere potentiality of components. The objectively existing stretch or expanse must have some relation to elements or components, since it is complex. But this relation can only be that all the parts or elements actually exist, and are all such that without their actual existence that very stretch or expanse would not exist. In this sense any objectively existing whole is composed out of parts. But now continuity or seamlessness is exactly non-composability by parts, and so no objectively existing whole (whether relational or absolute) can be a continuous expanse.5 In sum, the contentions of the Metaphysical Exposition—first, that a pure representation is involved in representing items to be in space (that is, the first two paragraphs), and, second, that this pure representation is nonconceptual (that is, the second two paragraphs)6—are both satisfied by space being represented in a flowing construction. It is only by sensations arising in the course of a flowing shift (of attention) that they are separated by an expanse, and it is in this flowing construction that space is continuous at all. Equally, I believe, a constructivist account explains the Transcendental Exposition. Geometry, for Kant, “is the science which determines the properties of space synthetically and yet a priori” (A25, B40, p. 70). Suppose we take it that the geometer systematically determines the dimensions of triangles. Suppose further that we think of a pair of constructions, one of which is to draw out a line and mark four units on it, while the other is to rotate ninety degrees from the initial position of the first, then draw out a line marking three units, rotate sixty degrees, and draw out a line marking five units. 5. Thus Kant, in essence, retains Newton’s fluxions, but not the idea that these quantities flow through an objectively existing expanse. Indeed, to flow through such an expanse they would have to get through all the multitude of composing elements of the objective expanse, and so there would be a segmentation in the motion (leading to Zeno’s paradoxes again), which thus would really not be a flow. This implies, by the way, that space (continuous expanse) is merely a flowing motion, and not also an objective expanse (compositional whole) through which the motion takes place. The idea that space is not merely a pure intuition but also something objective is thus rejected by this third (“continuity”) argument. 6. The argument of the fourth paragraph is roughly that a conceptual representation of space would have to involve a concept with infinite comprehension. Since space is infinitely divisible, any spatial relation, such as “being one meter from,” has to “contain” all sub-relations (being one half meter from, etc.), and so the thought of the initial relation must be a concept comprehending infinitely many sub-relations. But infinite comprehension of a concept, as opposed to infinite extension, is impossible. By contrast, a flowing construction “embraces” all potential subdivisions into parts.

8

t h e t r a n s c e n d e n ta l a e s t h e t i c

The concept (rule) of these constructions does not determine whether the second one ends up coinciding with where the first one ends up, and so triangle dimensions are not determined conceptually. The constructions have to have actually been carried out. In this way geometry is synthetic. Once a geometer has carried out a basis of local constructions to conclude that space is locally Euclidean, his results, for Kant, apply a priori to all possible spatial relations among any items in global space. First, since space itself (viz., spatial construction) is homogeneous with respect to size and translation for Kant (A710–11, B741–42, p. 577), spatial constructions of any dimensions are likewise Euclidean. Second, since constructive space is that within which and by which all items have their spatial relations, Euclidean geometry is therefore likewise the geometry of all empirical phenomena (the geometry of land surveying, astronomical light triangulation, etc.). Thus the geometer by his base local constructions (a base sufficient for establishing the axioms of local Euclidean geometry) determines a priori phenomenal coincidence results without having to actually perform them. Note that in this account both pure and applied geometry are synthetic a priori for Kant, since a base of actual local constructions (constituting the synthetic nature) determine a priori the geometric character of both all possible space or spatial construction (pure geometry) and all possible relations of items in space or in the course of spatial construction (applied geometry). On the other hand, in a relational view of space, the determination of the character of the relationships in only one locale is only meager empirical evidence for this character elsewhere. Better evidence would be some “fair sample” ranging over many locales, so the “geometry” of spatial relationships would not be a priori but inductive and empirical. The difference is that the relationist has no a priori homogeneity-with-respect-to-translation principle. Kant can say that first moving off and resituating elsewhere must leave the local results of constructions the same, since after all it is the very same constructions that are there performed. Thus his geometer can take his local results as exemplary (rather than as a single case of an induction), and so anticipate (a priori) other local results. The relationist, roughly, can only measure lengths and angles elsewhere between three local objects with no a priori guarantee that he won’t wind up with different (non-Euclidean) dimensions. Kant clearly expresses the view we have attributed to him in the Transcendental Doctrine of Method (A710–11, B741–42, p. 577). He says first of all that mathematics is knowledge gained from the “construction of concepts” in “non-empirical intuition.” This construction, he further says, is of a “single object” (say, a local triangle), but it “ex-

Kant’s Theory of Space and Time

9

presses universal validity for all possible intuitions which fall under the same concept” (viz., it pertains to the construction of all local triangles at all locales, and to large triangles). The reason it has this validity is that it is the very same act of construction. As he puts it, “The single figure which we draw is empirical, without impairing its universality. For in this empirical intuition, we consider only the act whereby we construct the concept and abstract from the many determinations (for instance the magnitude of the sides and the angles) which are quite indifferent, as not altering the concept ‘triangle.’”7 Immediately following the Expositions Kant concludes that space is nothing objective and it pertains to things only in so far as they are intuited (A26, B42, p. 71). The first part of this contention, I believe, follows immediately from the fact that continuous spatial expansiveness exists only in flowing constructions. The second part of the contention needs some further argument. Why shouldn’t these flowing constructions pertain to things even apart from (empirical) intuition? Why can’t we think or conceptualize things objectively existing that our flowing constructions then pertain to? To think that our flowing constructions pertain to them, I suggest, would be to think that they are reached at various stages of our constructions. But this is to think that they themselves are “at” positions that these stages coincide with. The trouble with this is that we are then thinking that they are in space (at positions) or, equivalently, we are thinking that space is also objective. But since, as per the first part of the contention, space exists only in flowing constructions, this is impossible, and hence the second part of the contention follows from the first part. On the other hand, the empirical intuition of things is positioned by its arising in the course of spatial construction, not by its having its own position match that of a stage in the construction. Roughly, empirical intuition is something that happens in the course of construction, whereas apart from such intuition (or abstracting from the intuition) there is only the thought of the thing existing in such a way that construction reaches or matches it. Only the former is compatible with 7. I believe Kant’s point could have been made more convincingly. It isn’t clear that construction of larger triangles are the very same constructions. However, I believe that the homogeneity with respect to size follows analytically from homogeneity with respect to translation (and rotation), and these latter do seem to pertain to the very same constructions, thus allowing the single object the geometer constructs to anticipate all construction of triangles (including large ones). The defect with Kant’s argument, however, is that it begs the question as to the individuation conditions of constructive acts. If translating is part of the second construction (and not just something that intervenes between constructions), then the second construction is a different one from the first one. This allows the constructivist to reject the necessary homogeneity of space.

10

t h e t r a n s c e n d e n ta l a e s t h e t i c

space being merely or exclusively constructive. It is only then the empirical intuitions of things (not things themselves) that get positioned by spatial construction, without having to have their own spatial positions correspond to that construction. In this way both parts of Kant’s contention follow directly from the arguments of the Expositions (that space exists only in flowing construction), and so, therefore, Kant’s transcendental idealism also follows from the Expositions. An analogy, I believe, will help make clearer the nature of Kant’s contention. Consider the feature, in regard to objects on my desk, of having a linear order (of there being a “first” one, a “second” one, etc.). It should be clear that such an order is no intrinsic feature of the objects. Rather they have such an order only in relation to some ordering procedure or construction on the subject’s part. Thus I place my hand on an object while reciting “one,” place my hand on another object while reciting “two,” and so on. It is this ordered procedure that “imposes” order on objects that, by themselves (apart from the procedure) have no order. Further, if I abstract from the empirical intuition of these objects (viz., from the presentation of them by placing my hand on them, or otherwise attending to them), the ordering construction (reciting “first,” “second,” etc.) does not pertain to them. To think this construction pertains to them, even apart from their empirical intuition arising within the construction, is possible only by thinking that they themselves have some intrinsic linear order that matches or corresponds to the ordering construction. Finally, now, and perhaps most tellingly,8 our constructivist interpretation of Kant’s theory of space carries over to his theory of time. Time, like space, is expansive or continuous and so, like space, must exist only in flowing constructions. I contend now that there are indeed flowing “temporizing” constructions analogous to spatial constructions. Consider the orchestra conductor who “tempers” the playing of the orchestra by a series of slow exaggerated downbeats. Each such downbeat is a flowing construction stretching out a temporal expanse. The series of such downbeats constitutes a marking-time procedure. A traffic officer waving to move traffic along in a certain direction by a series of sweeping gestures of his hand across his body marks out a pace for traffic to follow. There are all sorts of temporizing gestures for slowing others down and speeding others up. Further, there are pure temporizing constructions where nothing is getting tempered, as when I am waiting for someone in front of a store and I just mark time by tapping my feet. 8. I say “most tellingly” because interpretations of Kant on space so often don’t carry over to time, whereas Kant’s text makes it clear that he thinks they are essentially parallel.

Kant’s Theory of Space and Time

11

Note even here that the tapping is a motion, and so a flow (from the highest point of my toe down to the ground). These temporizing constructions are also spatial motions, but Kant himself says that time itself is represented by attending to the succession in (the spatial motion of) drawing a straight line (B 154, p. 167; also A33, B50, p. 77). Note that in a series of downbeats the expanse of time marked out increases, while the expanse of space does not (viz., it is repeated). In this way, temporal construction is distinguished from spatial construction by attending to the succession. In the Aesthetic Kant essentially treats time on a par with space. Thus time is not a matter of sensed relations (A31, B46, pp. 74–75), since to represent succession presupposes the (pure) representation of time; viz., if one thing succeeds another, it does so by being separated from it by an expanse of time and, as with space, this involves an intervening flowing construction (marking-time) between the sensed items. Time is also not a matter of conceptualized relations since it, like space, is continuous (viz., the temporal expanse is a whole prior to its parts). The main difference in Kant’s treatment is that time, in being the form of inner sense, applies to all appearances whatsoever, although only “mediately” to outer appearances (A34, B50, p. 77). This latter means, I believe, that time applies to outer appearances by its applying to the spatial constructions within which outer appearances arise and so have their place. Spatial construction is an act or performance and like other acts or performances it can be tempered (viz., done at a marked pace). In this way, outer appearance arises in the course of the temporizing of spatial construction.

2 . t h e t r a n s c e n d e n ta l d e d u c t i o n In the Aesthetic Kant has abstracted from thought to consider sensibility alone. In the Deduction it is the nature of thought itself that is at issue, and, in particular, the nature of thinking space and time. Indeed, the “premise” of the Deduction is that we can think the full scope of the one space and time that there is. This capacity is a premise for any cognition of anything real, because all cognizable reality is reality that has determinate position with respect to space and time. All cognizable objects, occurrences, and even states of consciousness have their existence in the one space and time. Thus, in particular, a sensory state of consciousness, even if it isn’t self-conscious, and even if it is barely or subliminally conscious, is a real individual occurrence or state of affairs existing within (the fullness of) time. Hence any thought or cognition of that state involves thinking of it as existing in the fullness of time,

12

t h e t r a n s c e n d e n ta l a e s t h e t i c

and so involves the thought of the fullness of time within which it exists. For example, in navigating my way home I am in some sense aware of the streets and signposts, although clearly not self-consciously or conceptually. I may later think that I must have been so aware to get home. This thought or cognition of the non-self-conscious state as something real represents it as being an occurrence in time, and hence presupposes the thought or cognition of time. In the B edition Deduction Kant characterizes the premise as the “original synthetic unity of apperception,” which he calls “The first pure knowledge of understanding .l.l. upon which all the rest of its employment is based” (B137, p. 156). He says of this synthetic unity that it is “completely independent of all conditions of sensible intuition,” meaning, thereby, that this first pure knowledge does not pertain to sensible reality. The example he gives of it is synthetically unifying a space. The original synthetic unity, then, is bringing space itself (as opposed to sensible reality in space) to the unity of thought. This is in line with our contention that the premise of the Deduction is the thought or cognition of space and time themselves. In the statement of the A edition Deduction (A119, p. 143) Kant characterizes the pure understanding as the unity of apperception in relation to the transcendental synthesis of imagination. Now, this transcendental synthesis provides an a priori combination of the manifold (A118, p. 143). That surely suggests that it pertains to the pure manifold of space and time themselves. He goes on to say that this pure understanding (and hence the thought or cognition of space and time) is a “formal and synthetic principle of all experiences” (A119, p. 143). It is involved, that is, in all cognition of appearances or what is real. He makes this same point in the B Deduction, where he says that the transcendental synthesis of imagination “determines sense a priori in respect of its form in accordance with the unity of apperception” (B151, p. 165). Since it is the form of sense that is involved, he must be talking of determining space and time with respect to cognition (the unity of apperception). He goes on to say of this cognition of space and time (B152, p. 165) that it is “an action of the understanding on the sensibility, and is its first application—and thereby the ground of all its other applications—to the objects of our possible intuition” (emphasis added), once again expressing therefore that the thought or cognition of space and time is the basis of all cognition of anything real. Now, the Transcendental Deduction itself is a method for deducing categories, or a method for establishing that pure concepts apply to all cognizable reality. Because the thought or cognition of space and time is involved in all cognition of what is real, it follows that if there are

Kant’s Theory of Space and Time

13

pure concepts that are required for the cognition of space and time, then these concepts must pertain to all cognizable reality. In this manner, the method for deducing categories is to show they are required for thinking the full scope of space and time. In the statement of the Deduction at A119, p. 143, Kant says that “In the understanding there are then pure a priori modes of knowledge [categories] which contain the necessary unity of the pure synthesis of imagination”; i.e., categories are what effect (contain) the relation of thought (necessary unity) to space and time (the pure synthesis of imagination). In sum, then, we have an exceedingly simple structure to the Transcendental Deduction, or to the method for establishing the applicability of pure concepts. To show that a pure concept has applicability, it suffices to show that it is required for thinking or cognizing space and time. If so, then it must pertain to all cognizable reality, since any cognizable reality is reality determined (having position) with respect to space and time. The question now is what is it exactly to think or cognize space and time? Given the third and fourth arguments of the Metaphysical Expositions in the Aesthetic, it certainly can’t be thinking objective spatiotemporal relations. The answer, I suggest, is that cognizing space and time for Kant is a matter of bringing spatio-temporal construction to rules. To begin with, the transcendental imagination that synthesizes the pure manifold of space and time Kant calls the “productive” imagination (A118, p. 143, and B152, p. 165), which seems to indicate that the synthesis is a production. Further the figurative synthesis of the transcendental imagination is illustrated by examples of drawing a line, describing a circle, etc., each of which Kant says is a “motion, as an act of the subject” (B154, p. 167). This suggests that the production is a flowing construction (motion) on the part of the subject. If this is correct, then Kant’s view in the Deduction that space and time are due to a synthesis of the imagination is identical to his view in the Aesthetic that they are given in pure intuition. As we have just noted, the synthesis is a flowing construction or generation (not a putting together out of elements), and being given in pure intuition doesn’t signify being present as a sensorium, but rather being presented or given in a flowing construction. Thus, the two accounts are identical. The unity of apperception, now, Kant says “is impossible if the intuition cannot be generated in accordance with a rule” (A105, p. 135) and “This unity of rule determines all the manifold and limits it to conditions which make unity of apperception possible” (A105, p. 135). Since apperception is thought or cognition, for Kant, he is thus holding that we think a manifold, including a pure manifold of space and

14

t h e t r a n s c e n d e n ta l a e s t h e t i c

time, by bringing it under a rule. If so, then our cognition of space and time has the form of rules for spatio-temporal construction. In having a rule, I know what I am doing or what is to be done, and so in this sense the performance (the doing) is unified for my consciousness as what the rule requires. The rule, that is, gives unity to the act or the construction, or makes it a whole act or performance for my cognizance. This conception of the unity of thought or cognition as the unity of a rule is certainly compatible with a constructivist account of space and time, since constructions are exactly what can come under and get unity from rules. Note that in a rule for a spatial construction there are now two sorts of unity. There is the unity (non-partitiveness) of the flowing in the construction (the seamless or continuous character of the flow), and the unity for cognizance of the flowing production in the concept or rule that governs the production. The former is the unity belonging to the manifold of space and time (spatial and temporal construction itself), while the latter is the unity of the understanding (thought or cognition). This, I believe, explains Kant’s difficult footnote to B160, p. 170, where he says that “In the Aesthetic I have treated this unity [in representing space as an object] as belonging merely to sensibility.” Part of what is cognized by a rule, that is, is that it is a rule for a flowing construction, and this unity-of-flow is what “belongs to sensibility” and “precedes any concept.” Thus in this footnote Kant is not retracting the doctrine of the Aesthetic (that there is a nonconceptual unity to spatial or temporal expansiveness or continuity). He is not, that is, holding that in thinking space and time we somehow bypass pure construction in favor of some conceptual apprehension of the character of space and time. Rather, in this footnote Kant is reminding us that in the Aesthetic he has abstracted from the nature of thought that has its own unity (the unity of a rule), which governs the character of spatial and temporal construction. Rules, now, are not only rules for what we are in fact doing, but rules for what can be done or, better, for what is proper to do. In this way, rules have the spontaneity of thought, in allowing us to represent beyond what is actually presented. Thus I can have a rule for linearly constructing thirty unit segments, or a rule for marking time up to thirty units or “beats,” and so represent a space or a time without actually presenting it. It is this spontaneity of representation by rules that opens up the possibility of my cognition extending to or encompassing the full scope of space and time. If, that is, I have a repertoire of rules for spatial and temporal constructions that together or collectively encompass the full scope of spatio-temporal construction, then I have a cognition of (or a cognitive faculty that is adequate to) the fullness of space and

Kant’s Theory of Space and Time

15

time. Now, recall that the premise of the Deduction is just this cognizability of the fullness of space and time. Everything cognizable is thought of as being in space and time, and so encompassed (together with anything else cognizable) by one of these rules. It is this repertoire of rules for the full scope of spatial and temporal construction that constitutes the unity of all that is cognizable in space and time. Now, finally, we can exactly characterize the method of the Transcendental Deduction. If there are pure concepts that are required for having a repertoire of rules governing the full scope of spatio-temporal construction, then these concepts pertain to all cognizable reality (to all reality having a place or position relative to the fullness of space and time). Kant characterizes categories as “certain concepts which render possible the formal unity of experience,” or as concepts concerning “solely the form of an experience in general” (A125, p. 147). This formal unity of experience is that everything there is (all possible sensible items) exists determinately in the fullness of space and time. This is made clear by Kant in the B edition statement of the Deduction, where categories are said to effect “a combination to which everything that is to be represented as determined in space or in time must conform” (B161, p. 171). If our account is correct, then the very premise of the Deduction regarding the cognition of space and time (that we have a repertoire of rules for the full scope of spatial and temporal construction) incorporates, rather than contradicts, the constructivist doctrine of space and time (as given in pure intuition) of the Aesthetic.

3. the analogies The proof or deduction of specific categories in accord with the method set out in the Transcendental Deduction is carried out in the Principles, which includes the proof of the relational categories in the Analogies. If this is the case, then we should expect the Analogies, in accord with the premise of the method in the Deduction, specifically to concern constructive space and time. However, Kant’s statements in the Analogies make it seem instead as if he is holding a relational (and so objectivist) theory of time. He says, for example, “Since time, however, cannot itself be perceived, the determination of the existence of objects can take place only through their relations” (B219, p. 209). He says that the “unity of time-determination is altogether dynamical” (A215, B262, p. 236), and “since absolute time is not an object of perception .l.l. the appearances must determine for one another their position in time” (A200, B245, p. 226). This apparently objective and relational conception of time would be incompatible not only with the

16

t h e t r a n s c e n d e n ta l a e s t h e t i c

Aesthetic, which argues that time is not a matter of sensed or conceptualized relations, but with the Transcendental Deduction as well, whereby categories don’t constitute or effect objective relational time, but rather bring the pure manifold of the transcendental imagination (that is, the manifold apart from appearances or objects in time) to the unity of apperception. What I want to argue is that, properly understood, Kant does not abandon a constructivist theory of time in the Analogies. I try to show this by very roughly presenting an account of Kant’s arguments for the relational categories. Kant’s argument for substance, added in the B edition (B225, p. 213), begins with his saying that “All appearances are in time, and in it alone as substratum (as permanent form of intuition) can either coexistence or succession be represented” (emphasis added). The argument begins, that is, by saying that time as a form of intuition is “permanent.” If we take seriously the equation of a form of intuition with construction (where what exists in time is what arises in the course of temporizing construction), then Kant’s premise is that constructive time is “permanent.” This means, I suggest, that time does not “begin” with a rule for (or with the propriety of) a temporizing construction. Thus, it is proper for me now to go ahead and temporize, say, by producing a series of paced downbeats accompanied by a recitation of numerals (which mark or label the downbeats). But past time exists prior to this proper construction. Indeed, if, in line with the Transcendental Deduction, there is to be a repertoire of rules for temporal constructions adequate for the full scope of time, then there must somehow be rules of construction that pertain to the past. Besides its being proper to go ahead and perform a construction, it can be proper, I suggest, to now be in the middle of or in the course of a constructive procedure, without my having performed the earlier stages. As an example, if I have a procedure (order) for mixing ingredients together in baking a cake, and I presently come in upon a bowl with the first three ingredients mixed in, then it is proper for me to be up to stage 4 (and so in the middle) of the cake-baking procedure. We can say, in Kant’s terms, that the bowl with the three ingredients in it that presently exists is the “substratum” or basis of my now being up to stage 4. Similarly, to represent past time constructively is just to represent being in the middle of (rather than at the beginning of) a temporizing procedure. This, likewise, will require a basis or substratum (i.e., something presently existing that “puts me” up to a non-beginning stage of the procedure).9 Now this substratum cannot be time itself, for, 9. It cannot be proper to just be in the middle of a procedure. A procedure is exactly a

Kant’s Theory of Space and Time

17

Kant says, “time cannot by itself be perceived.” It is not the case, that is, that what makes it proper to now be up to a certain stage of temporizing is that this procedure is geared to objective time (geared to getting objective time “marked” or “covered”), and objective time is now so far along in its progression. The reason is that there simply is no objective time for Kant. We have seen in the Aesthetic that time (continuous expanse) is given only in flowing construction. The time-marking procedure of a series of (flowing) downbeats doesn’t match or reflect objective (extra-constructive) time; rather, that series is time. Thus objective time (whether relational or absolute) being presently so far along cannot be the “substratum” of the “permanence” of time as a form of intuition. Rather, Kant says, “there must be found in the objects of perception .l.l. the substratum which represents time [constructive time] in general.” Something real that is present, then, must be the substratum. This cannot be a present sensation, since sensations are momentary and hence are not “so far along” in their existence as to make it proper to be in the course of, or in the middle of temporizing. However, besides sensation there is for Kant the real, which “corresponds” to sensation (A166, p. 201). Since sensation, for Kant, is consciousness that we are “affected” (B207, p. 201), the real that corresponds to sensation is that which affects us. Although the sensing is inherently momentary, that which affects us needn’t be. Hence that which affects can be and (if past time is cognizable at all) must be the substratum. We get thereby the following as the cognition of past time: (1) With regard to that which presently affects, it is proper to be up to stage k in a temporizing procedure, where the temporizing procedure in (1) is producing a series of downbeats accompanied by a recitation of numerals. But now (1) represents what is presently real (that which presently affects) as so far along in its existence that a temporizing procedure geared to it10 must be so far along (up to stage k). But this, in turn, is just to represent what is presently real as having a past temporal existence or, equivalently, as having a past duration which, in its turn, is just the permanence (lastingness) of the real, or substance.11 In sum then, the concept of subway of performing in fixed order (first one stage, then the second stage, etc.). However, if the procedure is geared to something (as the cake-baking procedure is geared to getting a cake baked), then if that thing is presently so far along (as the bowl with the mixed ingredients in it), then it is proper to be that far along as well in the procedure that is geared to it. 10. Say, by the temporizing procedure marking the tracking of what affects. 11. For substance being the permanence of the real; see B183, p. 184. In our interpretation of the First Analogy substance is required for representing any change (or

18

t h e t r a n s c e n d e n ta l a e s t h e t i c

stance is what enables rules for temporal constructions to encompass the past (as rules for being in the middle of constructions). This is exactly in accord with the method of the Transcendental Deduction for proving a category; viz., by showing it is a concept required for the representation of time (for having a full repertoire of rules for temporal construction), or for bringing the flow of past (constructive) time to the unity of apperception. In this account, then, Kant is not giving up his constructive theory of time in the First Analogy. To the contrary, his argument for substance turns explicitly on the fact that time exists only in flowing constructions. It is true that without objects of experience (substances) there is no cognition of past time, but, far from a relationist view, this is so because past time is a matter of being in the course of flowing constructions which requires objects of experience (substances). The argument of the Second Analogy is that objective succession presupposes causation. It is, I contend, a single connected argument that begins with the idea that objective succession is distinct from objective coexistence, not by being successively apprehended, but by being subject to a different rule (of apprehension), viz., a rule of irreversibility. This, however, is just the first step. After it, Kant says there must be a condition of this rule “according to which the event [the objective succession] invariably and necessarily follows” (A193, B239, p. 222), and the reason he gives is that “The advance,l.l.l. from a given time to the determinate time that follows is a necessary advance” (A193, B239, p. 222). Objective succession, that is, is not only distinct from objective coexistence, but further it is a succession that arises in time and, indeed, in a time that necessarily advances from preceding time (before the succession). As he puts it later on, “it is a necessary law of our sensibility .l.l. that the preceding time necessarily determines the succeeding time” (A199, B244, p. 225). A later time, that is, doesn’t just happen to arise upon the lapse of a preceding time; it necessarily emerges from it. Kant calls this a law of our sensibility, saying “I cannot advance to the succeeding time save through the preceding” (A199, B239, p. 225). Since Kant is talking of my advancing, he is still talking of constructive time. It is a law of actual construction that subsequent downbeats are determined or emerge from precedent ones. This law of construction, however, pertains only to actually going ahead to temposuccession) because it is required for representing the ongoing past time within which any succession must arise. In particular, substance is not required to distinguish succession from coexistence. For objections to the latter understanding of the role of substance, see Andrew Ward, “Kant’s First Analogy of Experience,” Kant-Studien 92 (2001): 387–406.

Kant’s Theory of Space and Time

19

rize, and so not to past time. The reason is that it is now proper simply to go ahead and perform downbeats, and so this propriety doesn’t emerge or advance from the propriety of any preceding performances. How can necessary advance, then, be a feature of past time if past time is constructive, and it can now be proper just on its own to go ahead and construct (without advancing from precedent construction)? The answer is that this feature must pertain to what arises in, or is encompassed by, constructive time; viz., events or appearances. Thus, if constructive time is geared to encompassing the unfolding of events within it, which events necessarily emerge from each other, then presently going ahead to encompass event e1 by a downbeat necessarily emerges out of the propriety of being in the course of a series of downbeats that encompass a series of events, each of which necessarily advances to the next event unto advancing to el. If downbeats, that is, are geared to sequences of (necessarily advancing) events then the present downbeat (as thus geared) emerges from the propriety of precedent downbeats. As in the First Analogy it cannot be an objective (extra-constructive) time series that bears the feature of the preceding necessarily advancing to the succeeding, since once again such objective time does not exist. Because of this, Kant says, “the appearances must determine for one another their position in time and make their time order a necessary order ” (A200, B245, p. 226, emphasis added). Equivalently, he says, “A series of appearances [states or occurrences] thus arises [in constructive time] which .l.l. produces and makes necessary the same order and continuous connection as is met with a priori in time—the form of intuition [as is met with a priori in actual temporal construction]” (A200, B245, p. 226). But now an event necessarily emerging from a preceding event is a causal relation. Hence, the representation of the necessary advance of (constructive) time is via its being geared to a causal series of events. Finally, now, any objective succession arises in a time series where each time necessarily advances to the next, and hence it follows that any objective succession is a member of a causal series of events or occurrences that lead up to it. In this way, objective succession (in advancing time) presupposes causation. In our account Kant is holding a version of a causal theory of time, viz., a theory according to which temporal relations are “reduced” to (replaced by) real (causal, dynamical) relations. However, it is only the relation of necessary succession (necessary advance) that Kant is “reducing” to a real relation. Succession separated by expanse is given in constructive time itself. It is only necessary succession that is not given in constructive time, and so must be “reduced” or constituted by a real dynamical relation. As Kant says, he is holding in this regard an “altogeth-

20

t h e t r a n s c e n d e n ta l a e s t h e t i c

er dynamical” theory (A215, B262, p. 236). Far from holding a relational theory of objective time according to which necessary advancement would be a feature of objective (extra-constructive) temporal relations, it is because there are no such relations12 that Kant’s argument forces events to bear the relation of necessary succession, and hence forces causality. Not only is the argument of the Second Analogy compatible with a constructivist account of time; it depends on it.13 In sum, then, Kant’s arguments in the Analogies are not expressions of a relational (extra-constructive) theory of time. They are, rather, as he himself says in the Transcendental Deduction, arguments that certain dynamical concepts (the relational categories) “contain the necessary unity of the pure synthesis of the imagination [i.e., the unity of spatio-temporal construction under rules] in respect of all possible appearances” (A119, p. 143). They are arguments, that is, that certain dynamical concepts are required for having a full repertoire of rules for spatio-temporal construction that encompasses the full scope and character of spatio-temporality. The First Analogy argues that only the applicability of substance enables rules (for being in the course of constructions) that encompass the past, and the Second Analogy argues that only the applicability of causation enables rules (for temporizing geared to events) that encompass the necessary advance of the time series. If I am correct, then not only is Kant’s theory of time in the Deduction and the Analogies consistent with his (constructivist) theory in the Aesthetic, but, further, the very nature of these latter sections (as a presentation and then a carrying out, respectively, of a method for establishing or deducing categories) is incomprehensible apart from the Aesthetic. There is simply no understanding the Deduction or the Analogies apart from the transcendental ideality of space and time; namely, that space and time are given merely in pure intuition, or are nothing apart from the constructive activity of the subject. 12. Such relations would have to be separated by continuous expanses, and the latter, as per the Aesthetic, exist only in constructive time. 13. The Third Analogy gives a parallel argument for coexistence in space. An objective coexistence is not only distinct (by a rule of reversibility) from objective succession, but also takes place in a full realm of coexistence (i.e., space) within which each spatial position necessarily determines and is determined by all other spatial positions (which is what Kant calls “spatial community” at A213, B260, p. 235). This relation of necessary mutual determination of spatial position (analogous to necessary advancement in the case of time) Kant again “reduces to” or finds in a universal dynamical community of mutual interaction among occupants (each mutually determining the others). See footnote (a) to A218, B265, p. 238, for a succinct statement of the structure of Kant’s argument.

2.

Kant vs. Lambert and Trendelenburg on the Ideality of Time

Kant’s thesis that space and time are transcendentally ideal can be fruitfully understood, I claim, on the model of constructivism in the philosophy of mathematics. Just as numbers are not objective (Platonic) realities but exist in and through procedures or constructions such as counting, so too space and time are not objective (absolute or relational) realities but exist in and through “flowing” procedures or constructions. In this essay I develop and use this model to answer the longstanding objections of Lambert and Trendelenburg against Kant’s views, thereby showing the fruitfulness of pursuing this model. In Section 7 of the Aesthetic Kant states and responds to Lambert’s objection that since change is real, as shown by the succession of our own representations, therefore time too must be real. As Lambert puts it, “I think though that even an idealist must grant at least that changes really exist and occur in his representations, for example, their beginning and ending. Thus time cannot be regarded as something unreal.”1 Perhaps what is outside us is not really in space since our apprehension of what is outside us is questionable, but surely, according to Lambert, we are aware of our own representations as they really are and this awareness includes their being temporally successive. In one sense this is an odd objection, since Kant holds that time is real and pertains to our representations. Indeed his response is that “Certainly time is real, namely, the real form of inner intuition.” Lambert’s objection then would have to be that we are directly aware of the succession of our 1. Lambert’s objection is made in a letter of Oct. 13, 1770, to Kant. It is reprinted in Kant’s Philosophical Correspondence, ed. and trans. Arnulf Zweig (Chicago: University of Chicago Press, 1967), 63. For a contemporary discussion of the objection, see James van Cleve, “The Ideality of Time,” in Proceedings of the Eighth International Kant Congress, vol. 1, part 2, ed. Hoke Robinson (Milwaukee: Marquette University Press, 1995). My position in this essay is that Kant takes what van Cleve calls the “radical line” that our own representations by themselves are not successive. He finds this difficult to accept and hardly imaginable. In what follows I shall claim that if we understand succession to entail separation by a continuous expanse of time, the radical line is easy to grasp.

21

22

t h e t r a n s c e n d e n ta l a e s t h e t i c

own states apart from any mediation of a form of intuition, so that in this case at least we are directly aware of time in an inner sense without any determining of inner sense by the subject. The question then becomes, how it is that the representation of our own states as successive is dependent on some determination of inner sense by the subject? To get a handle on this question, let us first turn to Kant’s corresponding views regarding space. In the first paragraph of the Metaphysical Exposition of Space (A23, B28, p. 68).2 Kant contends that in order for us to represent things as outside or alongside one another, the representation of space is presupposed. Thus for Kant we cannot simply sense that one thing is alongside another apart from our form of intuition. Spatial relationships such as outside or alongside seem analogous to temporal relationships such as succession, and so if we can understand why the former presuppose space as a form of intuition, then we can understand why the latter presuppose time as a form of intuition (or why the latter are real only in relation to a determining by the subject). If one thing is outside another then the two things are separated by a span or expanse of space. Further, this span or stretch is seamless or continuous. Without such an intervening expanse there is no relation of one thing being outside or apart from another. Now, it is this seamless or continuous expanse, I suggest, that is not a matter of sensation for Kant but is contributed by the subject. Kant thinks that continuity can only signify that the whole is prior to the parts (A25, B39, p. 69). Following Leibniz he holds that a continuous expanse cannot be composed out of elements or components. But if an expanse of space were an objective whole it would have to be composed of elements, since a whole can only exist objectively if all its parts or components do. Take away the parts of an objective whole and there is nothing left, so the whole indeed is determined by the totality of its parts or elements. Since continuity requires rather that the whole is not made up of parts, a continuous expanse of space cannot be something existing objectively. Leibniz concluded from all this that continuous space was a sensory confusion, somewhat like hearing the ocean without hearing the individual component waves. Kant could not abide that geometry, the paragon of science, was based on or pertained to a confusion.3 He needed an account of continuity, or of a whole that was not made up 2. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. 3. See Kant’s Inaugural Dissertation of 1770, trans. William J. Eckoff (New York: Ams Press, 1970), 53 (sec. 2, para. 7).

Kant vs. Lambert and Trendelenburg

23

out of parts. Newton’s notion of fluxions or flowing quantities may have provided the model Kant needed. Consider drawing a line, which is a construction that sweeps out from a point and continues in a flow until cessation. In constructing this whole I do not construct any parts since construction of a part of the line would have to involve a second terminus, and in my construction there is no terminus until the end. In this fashion my construction of the whole is not composed of a construction of any parts. Once the whole is constructed I can go back and mark off parts, but then the whole is prior to the parts. What I am suggesting is that Kant’s model of continuity, and so his model of spatial expanse, is construction or, equivalently, the motion of the subject (B154–55, p. 167). Sensation, then, is not sufficient for representing one thing as outside another. Rather, upon the sensation of the one thing a flowing construction (such as a sweeping out of a line with one’s finger) must intervene at the terminus of which there is sensation of the other thing. It is this and only this that represents the things as separated by a continuous stretch or expanse, and which is the further determination added by the subject that is involved in sensing one thing outside another. This construction, for Kant, can as well be in the sheer movement or shift of mental attention (as one turns one’s head and eyes from one thing to the other), but even here it is the active sweep or flow of attention, not sensation, that represents an intervening continuous spatial stretch. One might object that one can stand back and simply see the separation between one thing and another. What one sees, however, cannot be an objective intervening continuous expanse, since, as before, an objective whole cannot be prior to its parts and so cannot be continuous. What one simply sees that is at all relevant to continuous space is what construction or what intervening delineation can be performed. As an analogy, when one simply sees that there are four peanuts, what one sees is that a certain pairing construction (a certain counting) can be performed. Let us return now to the succession of our inner states. In analogy with the space case, one state is separated from the other by intervening continuous time. Indeed, these states are apart from one another in time. As with space, the continuity of time is understood by Kant in terms of its being a whole stretch that is not composed out of parts or elements, which once again makes sense only for a construction, not for an objectively existing time span. As an example of a temporal construction, consider a conductor who paces the orchestra’s playing of notes by downbeat gestures (each note to be played at the bottom of the downbeat). Intervening between the successive notes is again a

24

t h e t r a n s c e n d e n ta l a e s t h e t i c

flowing construction (from top to bottom of the downbeat) that is not composed of parts or subintervals, since there is no delay or marking of any terminus until the end of the downbeat. All sorts of gestures of pacing, of delaying, or marking time are temporal constructions. Sensation, then, is not sufficient for representing one state of the subject as before another, since it has to be before the other in time, which means it has to be separated from the other by an intervening seamless or continuous stretch of time. Thus upon the apprehension of the first state a “temporizing” construction must intervene, at the terminus of which there is apprehension of the second state. It is only by adding this determination to inner sense, or to the apprehension of our own states, that one state can be represented as preceding the other in time. Thus Kant says that we represent time by determining inner sense by an act of synthesis that he also calls “motion,” and that this act “first produces the concept of succession” (B154–55, p. 167). Again, it is the intervening temporizing construction (motion or act of the subject) that is required for representing the succession of our own states in time. As Kant says, we do not “find in inner sense such combination of the manifold,” but rather we “produce” it (B154–55, p. 167); viz., no objective passage of time can be found in inner sense, since only constructions are such that the whole is prior to the parts, and so only constructions can be continuous. Kant’s answer to Lambert, then, turns on the very notion of succession itself. First, succession is succession in time; i.e., an order of states separated by an intervening continuous time passage. Second, a continuous passage has to be one that is not composed out of discrete parts or components, and so one in which the whole is prior to the parts. Third, only in a flowing or sweeping construction is a whole prior to parts, since the flow is not composed of stops and starts marking off any parts. The succession of inner states, then, cannot be represented apart from this active determination (construction) of the subject. As Kant himself says, the dispute with Lambert is not over whether time is real but what the nature of real time is (A37, B54, p. 79). Is it something to be found in what is apprehended or detected in our cognizance of our own states. Or is it rather something that exists in the mode or way in which we represent these states (viz., by representing them within the course of a flowing construction determining an intervening expanse)? The continuity of time forces the latter alternative. Thus, as Kant holds, there is a complete parity between inner and outer sense (“both are in the same position” [A38, B55, p. 80]), for the issue is not the certainty or indubitability of our representations, but rather the topological nature of space and time. If we accept Kant’s

Kant vs. Lambert and Trendelenburg

25

contention that intervening spatial and temporal separation is continuous, and that only flowing construction can be thus continuous, then Lambert’s mistake is that he isolates the concept of succession from succession in (separated by) time. He is left with some sheer concept of asymmetric order perhaps, but not a concept of temporal order. In order to see what is going on, let us concoct a Lambert-like challenge to the ideality of square-shaped. We detect square shapes in outer sense, and so such shapes must be real (recall that the issue does not turn on any doubts regarding outer sense). Now, for Kant again, square-shaped is a constructive notion; indeed, it is a rule of construction to proceed straightly (produce lines) making 90-degree turns and proceeding just as far each time. Apart from this, for Kant, there is no precise notion of square shape, and, more important, no notion that a square shape has continuous sides. It is true that without construction, or even the ability to construct, an animal might be able to sense the difference between a square and a circle that is presented to it. What it is representing at best, however, by its sensory detection is square-like shape or a “quality” of square-like-ness, in isolation from any continuity of sides (or quantitative equality of sides). Far from such detection showing that squares are real, it doesn’t even pertain to squares but rather to square-likeness. Similarly, sheer apprehension in inner sense does not show that succession is real, since it pertains rather to succession-like-ness (order apart from, or isolated from, separation by an intervening seamless stretch of time). As a second analogy, suppose an animal without the ability to count can discern three from four pellets. This by itself does not show that numbers are real apart from being places or positions in counting constructions. What the animal is representing at best is three-like-ness or four-like-ness, a quality apart from both three and four, having positions in a simply ordered series. There is nothing in Kant’s view, as we have presented it, to suggest that square-like, say, isn’t a sensory feature, which would be the view that the matter of sensation is restricted exclusively to “intensive” qualities (perhaps feels, non-extended color qualities, etc.), and that any extensive-like aspects such as square-like-ness are contributed by the subject. Just as a constructivist regarding numbers, for whom numbers exist only as termini of countings, need not deny a sensory quality of three-like-ness, so too a constructivist like Kant, for whom space and time exist only in flowing constructions, need not deny a sensory quality of square-like-ness or successive-like-ness. In each of two ways, then, Kant allows for a detection of successiveness in inner sense. First, without any ability to construct an intervening temporizing flow, one may detect successive-like-ness. Second, with

26

t h e t r a n s c e n d e n ta l a e s t h e t i c

the ability to construct, that detection of successive-like-ness may also be a detection of the possibility of performing an intervening temporizing flow. Neither of these, however, challenges the ideality of succession in time, namely, that such succession is representable only via a flowing construction contributed by the representing subject. Kant’s claim, that a representation of the succession of our own states is thus dependent on a contributing determination by the subject, is a sufficient response to Lambert’s direct challenge. The response, however, opens up a second challenge to the ideality of time, namely, doesn’t the flowing construction or constructive act of the subject itself take time, or take place in time? When the conductor performs a downbeat gesture doesn’t this performance take place within a span of time extrinsic to the performance itself? If so, then time cannot be a matter of a flowing construction (something that exists only in the acts of the subject), but must also be something objectively real as the time in which the flowing construction takes place. The challenge, then, is that time cannot merely be a form of intuition but must also be something objective. This challenge is akin to Trendelenburg’s objection,4 that even if Kant’s arguments show that space and time are the manner in which we intuit things, this does not imply that what we intuit isn’t also really in space and time. What Trendelenburg calls the “neglected alternative” is that space and time may also be objectively real. Our present challenge, in holding that time must be (and not merely may be) also objectively real, is a stronger version of this Trendelenburg objection. To begin with, this issue of whether time is also something objective is certainly not a neglected alternative in Kant. The Mathematical Antinomies are concerned to show that space and time are mere forms of intuition that are not also objective realities with existence outside the regress of the subject. The issue in these Antinomies, that is, is not whether space and time are forms of intuition or objective existences (thus neglecting the alternative that they are both), but whether our forms of intuition also correspond to objective realities. I discuss Kant’s arguments in the Antinomies elsewhere.5 Here it suffices to say that 4. The Trendelenburg objection is discussed by Hans Vaihinger, Kommentar zu Kants Kritik der Reinen Vernunft, vol. 2 (Berlin: Union Deutsche Verlagsgesellschaft, 1892), 268. See also Norman Kemp Smith, A Commentary to Kant’s Critique of Pure Reason (London: Macmillan, 1918), 113–14. For a more recent discussion, see Henry Allison, Idealism and Freedom (Cambridge: Cambridge University Press, 1996), chap. 1. Allison’s position (following Lorne Falkenstein) is that since things in themselves can only have a comparative order of internal relations (which is not a spatial order), they cannot be in space. This, I believe, still effectively begs the question against Trendelenburg since it doesn’t explain why things in themselves cannot have extrinsic order. 5. See Arthur Melnick, Space, Time, and Thought in Kant (Dordrecht: D. Reidel, 1989), part 4, chap. 2.

Kant vs. Lambert and Trendelenburg

27

Kant considers his arguments in the Antinomies to be an indirect proof of what is directly proved already in the Aesthetic (A491, B519, p. 439). What then is the direct proof in the Aesthetic against Trendelenburg? Suppose time were also something objective. Suppose, that is, that a temporizing construction itself took place in an objective stretch or expanse of time. Such an objectively existing expanse would be a whole expanse that is real (all “there”) only if all its parts or elements were real (all “there”), and so something that is real in virtue of all its parts or components being real. But now, once again, for Kant as for Leibniz, a continuous whole cannot have its being or existence determined by elements or components, and so an objective span of continuous seamless time is impossible. Thus the same argument from continuity, that the whole precedes the parts, that shows time is a form of intuition or a flowing motion of the subject, shows that it is not also something existing objectively. A flowing temporizing construction, then, cannot take place in a seamless continuous expanse of time extrinsic to it. Although this much shows that Trendelenburg was quite wrong as to there being some alternative that Kant neglected, it doesn’t quite fully answer the challenge that the act of the subject can take place only because there is, objectively, time for it to happen in. To respond that there cannot be such time leaves us still with the position that there must be such time. Kant’s own thinking on this matter is that the idea that space and time must also be something objective is a “transcendental illusion” (A504, B532, p. 448). Kant’s arguments expose it as an illusion, but even according to Kant they do not dispel the illusion. This I believe is unsatisfactory, and so in the rest of this essay I shall try to dispel the illusion that the temporizing act of the subject must take place in time. Again we consider the issue first in relation to space, where equally it seems that a spatial construction such as sweeping out a line has to take place in objective space. Our question concerns the source of this idea that a construction must flow through a space that is already there. Part of the answer, I suspect, derives from visual perception, for it seems that I can visually detect the space upon which my construction can be carried out prior to carrying it out. I simply see the expanse laid out before me within or through which I can trace out a line. I do not deny that this is the case with vision phenomenologically, so dispelling the illusion will have to mean dispelling the phenomenology of vision as a guide to what is so. Note that a blind person can carry out spatial constructions and, more generally, can have a fully adequate representation of space. Something is in space if it is possible to reach it or get to

28

t h e t r a n s c e n d e n ta l a e s t h e t i c

it, or if there is a motion that ends in contact with it, whether or not it is possible to see it from afar. If so, then the visual aspect of an expanse already laid out is not a component of the representation of space, but more like a secondary quality. As an analogy, consider that the visual detection of shape involves an aspect of color contrast between the shape and its background. One wouldn’t conclude from this that color contrast is a feature of the representation of being square, since once again a blind person can perfectly well represent shape without it. Besides visual perception of an expanse before one, another deeper source of the illusion, I suggest, is the idea that motion is a change of relation to space outside it. If one holds a substantivalist theory of space, then motion is a change of relation to portions of that space where these portions maintain their relations to each other. If one holds a relationist theory of space, motion is a change of relation to elements of a system of objects that maintain their relation to each other. Kant, of course, rejects both accounts of space as objectivist, but if motion is extrinsic (requires a constant system of external places, or of other objects in relationships, to vary against) then it would seem that at least one of these accounts is presupposed by motion. Granted now that spatial construction is a motion, specifically of the subject, it still seems that this motion too is extrinsic, or has to be understood as a change in relation to (absolute or relational) space. The ideality of space, however, requires rejecting such an extrinsic space. Dispelling this source of the illusion requires dispelling the idea that motion is extrinsic or relative. Once again there is a purely visual component to this idea, in that perhaps one cannot see motion except against some relatively stable background. However, as before, blind people can very well represent motion, so this visual feature seems irrelevant. Beyond the visual component, however, there is the idea that the very essence of motion itself, apart from its detectability, is extrinsic. What else can motion be but a variation in relation to what is outside or extrinsic to the moving thing? Just as an object by itself, or intrinsically, cannot be taller than, so an object by itself, or intrinsically, cannot be in motion (varying its spatial relationships). Note that in a constructivist account motion is intrinsic, since I do not draw a line by that construction varying in relation to other constructions. How then are we to understand this idea of intrinsic motion? In relativistic physics the geodesic motion of an object is intrinsic motion. That an object is “following” a geodesic is a matter of there being no mechanical or electromagnetic forces acting on it locally. This dynamical conception of motion is indeed intrinsic, since it does not depend on how the object is varying spatially with respect to other ob-

Kant vs. Lambert and Trendelenburg

29

jects. Further, in relativistic physics space is built up out of, or is nothing more than, the full possibilities of geodesic motion. Rather than motion having to take place through (in relation to) space, space itself is just the structure of the full possibilities of geodesic motion. In this account, then, motion does not presuppose space. Now, it may be empirically that certain bodily reference frames exist cosmically that allow us to talk somewhat of a fixed space or space-time system (in analogy to relationist or substantavilist accounts), and so allow us to introduce a notion of extrinsic motion. This, however, is secondary to the existence of space or space-time itself, which is just the structure of the full possibility of canonical intrinsic motion (geodesic motion). It is then the purely dynamical conception of motion that dispels the illusion that motion presupposes space. The character of Kant’s theory of space is, I believe, parallel to the above characterization of relativistic space-time. There are canonical basic motions (constructions of the subject) that are intrinsic, in that they are governed only by (the internal “force” of) rules for constructing apart from extrinsic mechanical forces. Thus, when I draw a line, my hand is governed by a rule of construction undisturbed by extrinsic pushes or pulls. The structure of space itself then is just the structure of the full possibility of such canonical intrinsic motions. Secondarily 6 and empirically, there may be relatively cosmic reference frames or spatial systems that arise from how matter is situated in terms of the canonical motions, which frames “simulate” relational space and allow for attributions of extrinsic motion (change of relations relative to such frames).7 The situation with respect to time now is essentially parallel to space. The “illusion” is that activity or performance (including temporizing construction) takes time extrinsically; viz., various elements of the activity relate to an external system of events unfolding (as per relational time), or moments unfolding (as per substantival time), and it is only via these relations that activity can take place or progress in time. The dispelling of the illusion, once again, is simply the existence of intrinsic time-exhaustive processes. In relativistic physics the unfolding of local clocks (processes), so long as there are no interfering external forces, is local time, and the structure of time is just the structure of the full possibility of intrinsic local clocks. What we have, then, is a dynamical conception of activity or process (in analogy to a dynamical conception 6. Immanuel Kant, Metaphysical Foundations of Natural Science, trans. James Ellington (Indianapolis: Bobbs-Merrill, 1970), 120–21 (prop. 1 of chap. 4). 7. Kant’s distinction of motion of the subject as transcendental vs. “motion of an object in space” as empirical (see footnote [a] to B155) may indeed be coextensive with intrinsic vs. extrinsic motion, only the former of which is fundamental to space.

30

t h e t r a n s c e n d e n ta l a e s t h e t i c

of motion) that is intrinsic, in that it doesn’t take up time by being related to any external flow of events or moments. Once again it may be possible, based on the empirical (accidental) distribution of matter, to secondarily define a cosmic time, and thus to reintroduce extrinsic attributions of time-taking relative to cosmic time. Fundamentally, however, unfolding in time (everywhere locally) exists whether or not there are such temporal systems. Similarly, again, in Kant’s view there are basic canonical activities (temporizing constructions) unfolding only according to the rule of construction without dynamical disturbances. The structure of the full possibility of such activities is time. Indeed, a temporizing construction, such as the conductor’s marking time, can be regarded as the subject’s being a local clock. It is thus again a dynamical (force-free) conception of local process that dispels the illusion that taking up time can mean nothing other than varying in relation to something (absolute or relational time) that is extrinsic to the process. Much of the force of this illusion probably derives from thinking of time on the model of space (both perceptually and kinematically), so that the dispelling of the illusion in the case of space really is the heart of dispelling it in the case of time. In neither case do we need any extrinsic background or goings on relative to which motion or process takes place. The response, then, to the Trendelenburg objection is that the temporizing activity or construction (motion of the subject) does not take place in a further objective time because, for Kant, time is continuous and only constructions (flowing motions) can be continuous. The idea that nevertheless these activities must take place in objective time (or in objective space) is an illusion due in part to projecting qualities of visual perception onto things, and in part to the idea that motion in its nature is inherently relative and extrinsic. The fact that the temporizing activity, or motion of the subject, does not take place in objective time does not imply that it is, therefore, a timeless activity, or an activity that is a-temporal (apart from time altogether). The supposed objection, that the transcendental act of the subject (viz., temporizing construction) has to be either in objective time or else has to be a-temporal,8 is no more cogent than the objection to the constructivist regarding numbers, that the procedure that constructs numbers (say, reciting or counting) has either to take place in relation to objective (Platonic) numbers or else has to be a-numerical. This latter is not an objection to the constructivist regarding num8. For this type of objection, see Peter Strawson, The Bounds of Sense (London: Methuen, 1966), 247–49.

Kant vs. Lambert and Trendelenburg

31

bers but a bald statement of objectivism (Platonism) regarding numbers. Similarly, once the visual perceptual illusion and the illusion of the extrinsic nature of motion is removed, the idea that a construction which is not related extrinsically to objective time is thereby timeless is not an objection to a constructivist regarding time, but a bald statement of temporal objectivism. The fundamental transcendental act for Kant is the production (synthesis) of the pure manifold of space and time. What we have now seen is that this notion of such a transcendental act is perfectly coherent in just the way that constructivism generally is a coherent position. Indeed, my contention is that Kant’s idealism regarding space and time is simply a version of constructivism, on a par with mathematical constructivism (constructivism regarding numbers). Hence, the idea that the transcendental acts determining space and time must themselves be outside of space and time (noumenal and mysterious) is no more convincing than the idea that the acts of counting must be beyond all number (numberless and mysterious).9 Just as the counting procedure itself is serially ordered, nonrepetitive, etc. (viz., is numerical or has numerical character within it), so too the temporizing flowing procedure itself is extensive, divisible, etc. (viz., is temporal or has temporal character within it). Just as for the numerical constructivist, everything else (like beans in a bag) has numerical character by being subject to counting, or to the procedure that originally constructs numericality, so too, for Kant, everything else (including our own inner states) has temporal character (being a succession in time) by its being subject to temporizing, or to the procedure that originally constructs temporality. 9. Although for Kant there certainly is this transcendental act of the productive imagination producing the manifold of space and time, this is not a component of some transcendental psychology. First, conceptualization is not a matter of further acts (syntheses) but a matter of governing acts of the imagination by rules (which are the unity of syntheses). Second, a constructivist account of the representation of space and time is a semantical theory, not a psychological one, just as constructivism in the philosophy of mathematics is not a psychological doctrine. Having said this, it still seems to me that the objection against a transcendental psychology—that it has to involve some mysterious noumenal (nontemporal) agency—is not an objection to the ideality of time but a sheer denial of it (viz., if acts don’t take place in extrinsic objective time then they must be timeless).

PART II THE TRANSCENDENTAL DEDUCTION

3

Apperception and the Premise of Kant’s Transcendental Deduction

Kant’s Transcendental Deduction is supposed to provide a method for establishing the applicability of pure concepts to all cognizable reality. The premise of this method, then, must itself be such as to pertain to all cognizable reality and so, a fortiori, to all varieties of experience and consciousness as well as all varieties of objects and states of affairs. Now, a principle that is the basis of all cognition of reality whatsoever will pertain to all cognizable reality, and so be a suitable premise. My contention is that this basis is that the unity of apperception relates to the transcendental synthesis of the imagination. This relation is the original synthetic unity of apperception. In the first part of this essay I claim that apperception is our present intellectual cognizance, or the present capacity for thought that I possess, and that all cognition requires that this cognizance relate to the full extent of space and time. In the second part I claim that this relation is just the unity of apperception in relation to the transcendental synthesis of imagination.

1. the synthetic unity of apperception Apperception as a faculty is identified by Kant with the understanding (A119, p. 143), where the understanding is the power of thought (A126, p. 147).1 The exercise of this faculty, then, is having thoughts, and so this faculty defines me as an intelligence or a thinking subject (B155, p. 167).2 For Kant, as for Leibniz, thinking is self-conscious. In thinking, that is, I am aware of myself as a thinker (an intelligence).3 This self-conscious aspect of thought, however, plays no key role in the Deduction. What does play a key role is that apperception, as Kant un1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. 2. See also B159, p. 169. “I exist as an intelligence which is conscious solely of its power of combination.” 3. Apperception is “that self-consciousness” involved in “generating the representation ‘I think’” (B132, p. 153).

35

36

t h e t r a n s c e n d e n ta l d e d u c t i o n

derstands it, pertains specifically to the present. As a faculty it is my present power of thought or cognition (what thoughts I am presently capable of). As an exercise of the faculty it pertains to my present thinking (what thoughts I now have). The textual evidence for this being Kant’s intent is, first, that although apperceptive self-consciousness must be able to accompany all other representations, it “cannot itself be accompanied by any further representation” (B132, p. 153). Since past states of mine can be presently thought of, and so accompanied, it is only my present thinking that is unaccompanied (that is unthought of in having the thought). Equivalently, perhaps, apperception pertains to the determining rather than the determinable self (B107, p. 268 and also footnote (a) to B158, p. 169). Since past states can be determined with respect to my present thinking, it is only that present thinking that belongs to the determining self. Second, Kant talks of the necessary identity of apperception (A107, p. 136), which seems to signify what he says in the Second Paralogism, that the “subjective ‘I’ can never be divided or distributed” (A354, p. 336). This necessary identity then would be the unity or simplicity of my present cognizance throughout my present thought or cognition, or what Kant characterizes as “the absolute unity of the subject of a thought” (A354, p. 336). Now it is true that apperception can pertain to my past states, but in this case my necessary identity is the same as in any other case; viz., the undivided or undistributed subject of the thought of my past states. What distinguishes thoughts of my past states from other thoughts is that what is brought to apperception is inner sense rather than outer sense. Thus, if I now have the thought that I saw Mary, what I think is just having seen her, not having spatially observed seeing her. If apperception were anything more than the present necessary identity of the thinker throughout its thought, it would be hard to see what apperception could be in regard to nonsubjective states. An object in space that existed at a previous time and that I have never experienced still belongs to the faculty of apperception for Kant. But now, in thinking of such an object, there would be no diachronic necessary identity of my reaching back beyond my whole history of subjective states. Rather, I would be the necessarily identical subject through the present thought of that object (the content of which is outside my personal history). Now, Kant thinks that inner sense is brought to apperception in a way similar to outer sense (B156, p. 168, and B153, p. 166). By parity, then, the identity of apperception in regard to past states of mine is likewise my identity as the subject of a present thought of them (the content of which, in this case, is not outside my personal history), not any diachronic identity as an empirical subject through them.

Apperception and Transcendental Deduction

37

I contend next that the scope of apperception for Kant is all cognizable reality whatsoever. My necessary identity as a thinker encompasses, for example, Julius Caesar, dinosaurs, etc. Kant says that “our entire sensibility and with it all possible appearances” stand in relation to “original apperception” (A111, p. 139). This doesn’t mean that I must presently have a single thought that encompasses, or is of, all there is. Apperception, recall, is a faculty or capacity. What it means, I suggest, is that my present capacity for thought extends to all that is intuitable within the full reach of space and time. This, in turn, signifies that I presently am capable of a repertoire of individual thoughts that together cover or represent the full scope of what is intuitable in space and time (that is, our entire sensibility). It is this which constitutes my present existence as an intelligence or a thinker in regard to the world in space and time. Kant says that the understanding (the faculty of apperception) “relates to all objects of the senses” (A119, p. 143), not just to actually experienced objects or objects that it is personally possible to experience. Although it is true for Kant that the purport or scope of apperception is possible appearance or perception or experience, this signifies anything that is sensible in the full scope of space and time, which clearly goes beyond what is possible for me personally to have experienced or perceived.4 As an analogy, my “arithmetical cognizance” extends to possible recitations that are impossible personally, in the sense that I can have thoughts of (I can represent) recitations of numerals that are proper to do, although impossible for me to do. I still represent them as recitations (things to do). Likewise, my worldly cognizance can extend to possible perceptions that are impossible personally, in the sense that I can have thoughts of perceptions that are proper to have, although impossible for me to have. We can now give our first preliminary statement of the premise of the Transcendental Deduction. For Kant, all the manifold must be unified in relation to pure apperception (A116, p. 142), but this presupposes an “a priori combination of the manifold” in “relation to the original unity of apperception” (A118, p. 143). The “manifold” signifies what is real in space and time, while the “a priori combination of the manifold” signifies space and time themselves. Apperception or my present intellectual cognizance, then, pertains to anything real in space and time by pertaining to the full scope and character of space and time themselves. The premise, thus, is that I must presently have a ca4. To think any less is to wreak havoc with the Analogies, the Mathematical Antinomies, and the Postulates of Empirical Thought, all of which have global character pertaining to the full scope of reality in space and time.

38

t h e t r a n s c e n d e n ta l d e d u c t i o n

pacity for thinking or representing that contains a repertoire of thoughts that together cover the full extent of space and time. Any cognizable reality for Kant is spatio-temporal in the strong sense that it is reality that has position in, or with respect to, the fullness of space and time (or at least time). If so, then the basis of any cognition of reality is the cognizability of space and time, and so this latter is a proper premise of the Deduction. Note that on this view the premise of the Deduction is not a minimal one, such as that I can represent having experience, or I can represent myself as having perceptions over time. Nevertheless, I claim that our premise is more basic than these apparently more minimal ones. The reason is that anything real that is cognizable or representable (including present experience or a series of perceptions) is real and cognizable as being in time (or in both space and time). Thus to represent a series of my own perceptions is, or involves, the representation of them as existing in a time that extends far back before these perceptions. Even to represent my present inner experience is to represent it as existing in the present, and hence as existing at a moment that emerges from past time extending far back before the present. As far as cognition of individuals is concerned (whether these be individual objects, states, occurrences, experiences), nothing is more basic than representing space and time, since an individual is precisely what exists in the fullness of space and time. The alternatives are either that what we represent is not represented as being in space and time at all, or else that it is represented as being spatio-temporal-like, but not fixed or positioned in global space and time. The former, for Kant, is clearly beyond the limits of cognizable reality, while the latter, for Kant, is incoherent according to the nature of space and time. Space and time are continuous expanses that have no existence (as regions or positions or moments) separable from the full extents that they are. As an example, if I think of my own present conscious state as one of having a toothache, the present is either what emerges from the past or else it is a “boundary” at the “edge” of time itself. The latter for Kant is incoherent since for him points of time are limits separating expanses of time. Thus the cognition of even my present consciousness, as an individual occurrence or state in time, presupposes that my apperception pertains to the full scope of time. Roughly, if my present conscious state is real (that is, a real particular occurrence or state), it is so only by being real in time, and so the cognition or representation of it as real presupposes the cognition or representation of (the fullness of) time. Note further that on our understanding of the premise of the Deduction, there is no implication that all conscious states are appercep-

Apperception and Transcendental Deduction

39

tive states, or states the subject ascribes to himself, or states belonging to the personal identity of the subject. Indeed, there is no implication regarding the having of conscious states at all. There is only an implication regarding my present capacity to represent any kind of conscious state. As a conscious state it is a real event or occurrence in time that is cognizable as such even if it wasn’t and couldn’t have been “directly” cognized. Thus I may represent that I must have been aware of signposts in order to navigate my way home yesterday. Here, my present apperception pertains to conscious (but not self-conscious or self-ascribed) states. My cognition of them is “indirect,” somewhat in the sense in which my cognition of magnetism may be indirect. Nevertheless, as real events in time they are cognizable and only cognizable in terms of all of space and time being within the compass of my present apperception. All conscious states of mine, like all real individuals in space and time, belong to my present intellectual capacity; viz., I can presently have thoughts of them or represent them, even if in the having there was no thought or representation.5 The actual deduction of the categories, as stated at A119, p. 142, goes as follows. Kant says, “The unity of apperception in relation to the transcendental synthesis of the imagination is the pure understanding.” Granting for now that this unity in reference to the transcendental synthesis of the imagination is apperception in relation to the fullness of space and time, in saying that this is the pure understanding Kant is saying that this is the aspect of cognition (of the present capacity for thought or representation that I am) that goes beyond what is given to actual experience. He is going to go on to say that this pure under5. Kant is doing semantics (of thoughts or representations), not cognitive psychology, and semantically all conscious states are on a par with each other (and with objects in space) as being particulars or individuals with position in time. Semantically, then, they all belong to apperception in being cognizable realities. For Strawson the premise of the Deduction is that diverse experiences belong to one consciousness (Peter Strawson, The Bounds of Sense [London: Methuen, 1966], 97). My contention is that cognizing the fullness of time is more fundamental than cognizing the diverse experiences. For Wolff the premise is consciousness itself (Robert Paul Wolff, Kant’s Theory of Mental Activity [Cambridge: Harvard University Press, 1963], 94). I contend that cognizing the fullness of time is more fundamental than cognizing consciousness. Both of these views start from a premise weaker than objectivity and are called the “received interpretation” by Ameriks, who holds that the premise of the deduction is that there is objectivity (an external world) that we cognize (Karl Ameriks, “Kant’s Transcendental Deduction as a Regressive Argument,” in Kant’s Critique of Pure Reason: Critical Essays, ed. Patricia Kitcher [Lanham, Md.: Rowman & Littlefield, 1998], 89). This seems incorrect because, first, the categories do apply to merely subjective experiences and, second, an external world (substance and our immediate consciousness of it) is proved in the B Refutation, so it is unlikely to be the premise of the Deduction. I suggest that, for Kant, cognizing the external world is part of what is required for cognizing the fullness of time, and so part of what can be deduced—not the premise.

40

t h e t r a n s c e n d e n ta l d e d u c t i o n

standing must be contained in our empirical faculty, meaning roughly that even actual experience (and whatever empirical syntheses it involves) is subject to this pure understanding. In sum, then, Kant is stating our premise of the Deduction that the basis of all cognition whatsoever is the cognition of all space and time. He next says, “In the understanding there are then pure a priori modes of knowledge which contain the necessary unity of the pure synthesis of imagination in respect of all possible appearances. These are the categories, that is, the pure concepts of the understanding.” Pure concepts, then, are those that are required for (that “contain”) the present cognizance of the fullness of space and time. The Transcendental Deduction is not a proof of any specific categories but a method according to which categories could be proved or deduced. That method then, Kant is saying, is to show that the pure concept contains (effects or is required for) the unity of space and time in relation to my present cognizance. The concept then would pertain to all possible reality in space and time, since, whatever else, all such reality has its being (and cognizability) in existing in the fullness of space and time. A fortiori the concept would pertain to all conscious states (whether self-conscious or not, whether integrated into my personal identity or not) and to all perceptual experiences (whether of spatial objects or not); viz., it would pertain to all varieties of actual experience.

2. rules for constructions So far we have characterized the premise of the Deduction as our present ability to have thoughts that pertain to all spatio-temporal reality. This leaves it completely open how such thoughts are possible, or what the nature of such thoughts is. So, for example, it is compatible with what we have said so far that apperception is the capacity to have conceptual descriptive thoughts that match (are uniquely satisfied by) various entities existing in objective space and time.6 This, however, is surely not Kant’s conception of apperception, or cognition of space 6. An example of this would be Peter Strawson’s conception of reference in Individuals (London: Methuen, 1959). It is true that Strawson, to eliminate massive duplication (multiple satisfiers), has demonstrative reference as a component of uniquely identifying descriptions (so that his thoughts aren’t purely descriptive or conceptual). Still, the massive duplication of fictional or purely possible satisfiers is not thereby eliminated. Until the range of potential satisfiers is fixed to what is real, there is no unique reference specifically to what is real. But then the whole issue is how to fix this range, which indeed is just the whole issue of how cognition of reality is possible. Strawson’s demonstratively enhanced descriptive thoughts are thus as empty as Leibniz’s purely conceptual descriptive thoughts.

Apperception and Transcendental Deduction

41

and time, since conceptual descriptive thought is empty (without an object) for Kant,7 and space and time are not objective. We have to characterize the premise of the Deduction, then, in terms of Kant’s specific account of what it is to have thoughts pertaining to the full scope of space and time. To begin with, I claim that thoughts or intellectual cognitions for Kant are not descriptions of states of affairs, but rules. The unity of apperception is exactly the unity of a rule. In discussing objectivity in both the Transcendental Deduction (A104-105, pp. 134–35) and the Second Analogy (A191, B236, p. 220), Kant equates objectivity of representations with their being subject to necessity, rather than being haphazard or arbitrary. Instead of just perceiving ship upstream and ship downstream, my perceptions are objective if that is how it is necessary to perceive or, equivalently, if that is how it is proper or required to perceive. Our perceptions are objective, not in that we have them, but in that they correspond to or agree with the rule of how it is proper (required or necessary) to perceive.8 But now this necessary unity, or this unity of how it is required to perceive, Kant says “can be nothing else than the formal unity of consciousness in the synthesis of the manifold of representations” (A105, p. 135). The unity of consciousness in regard to my perceptions, that is, is just my rule of how it is proper to be perceiving. Kant goes on to say that “this unity of rule determines all the manifold and [in so doing] limits it to conditions which make the unity of apperception possible” (A105, p. 135). In other words, my actual perceptions are brought to the unity of apperception by being brought under a rule for perceiving. Presently having a rule for what I am doing is what first brings my actions or states to my present unified cognizance. It is in knowing, or having an idea of, what I am up to in a rule, that I presently know, or have an idea of what I am doing. Later on Kant says, “this unity of consciousness would be impossible if the mind in knowledge of the manifold could not become conscious of the identity of function whereby it synthetically combines in one knowledge” (A108, p. 136). The “identity of function,” I suggest, is cognition of what it is I am supposed to be doing in the combining of the manifold. Thus, if I simply recite numerals 1, 2, 3, etc., it is by understanding what it is I am supposed to be doing when I recite 4 that the previ7. For further discussion of descriptive thought, see Essay 5, section 4, in this volume. 8. He says at A191, B236, p. 220, that appearance “can be represented as an object distinct from them [i.e., from our actual perceptions or apprehension] only if it stands under a rule which distinguishes it from every other apprehension.” Thus, actually perceiving in a certain way is objective because it, unlike other apprehensions, is the one that conforms to the rule of how it is proper or necessary to be perceiving.

42

t h e t r a n s c e n d e n ta l d e d u c t i o n

ous steps signify anything to my present intellectual cognizance. For a manifold of representations (perceptions, recitings, etc.) to be something to my present cognizance is for me to have a rule for those representations, or a rule that encompasses those representations.9 There is evidence that not only concepts, including concepts of objectivity, are rules for Kant. He equates the unity of apperception (in relation to the synthesis of imagination) with the understanding (A119, p. 143), and defines the understanding, in a way fruitful to its essential nature, as the faculty of rules (A126, p. 147). But since he here characterizes the understanding not only as the faculty of concepts but of judgments as well, he must think not only concepts but also judgments (complete thoughts) are rules. If so, and if the unity of apperception is the unity of (effected by) a thought, then once again the unity of apperception is the unity of a rule. As far as the premise of the Deduction is concerned now, the key unity is the unity of apperception in relation to the transcendental synthesis of imagination. The latter for Kant is the synthesis of the pure manifold (the manifold of space and time). To understand this synthesis, and how rules might constitute its unity for thought, we need to understand that Kant is a constructivist regarding space and time. Just as a constructivist regarding numbers holds that numbers exist only in constructions (say, as termini of counting procedures), so too a constructivist regarding space and time holds that they exist only in constructions or procedures. In the Aesthetic Kant says that space and time are given in pure intuition. Intuition is that kind of representation which involves the presence of what is represented, and pure intuition is intuition without sensible affection by objects. But now, in performing a construction such as sweeping out a line, I produce the very object (the expanse of space), and so make it present without sensible affection by objects. It is in construction, then, that the space is given in pure intuition. The reason Kant is a constructivist, I believe, is that he holds that only in flowing construction (such as sweeping out a line) can there be continuity, and space and time, whatever else for Kant, are continuous magnitudes or expanses. In a flowing construction, the whole of the construction is not composed of stops and starts. Hence the whole is prior to the parts (which are segmentings or cuts subsequently imposed), and so, in this way, the construction is seamless or continuous. 9. I believe Kant makes the same point at A156–57, B195–96, p. 193: “Experience, however, rests on the synthetic unity of appearances.l.l.l. Apart from such synthesis, it would not be knowledge [cognition], but a rhapsody of perceptions that would not fit into any context according to rules of a completely interconnected possible consciousness” (emphasis added).

Apperception and Transcendental Deduction

43

A temporal construction would be something like a conductor’s series of downbeats, each of which is a flowing construction, that temper or pace the orchestra’s playing. Now, in the Deduction Kant ascribes space and time to a synthesis of the imagination. This is not only consistent with their being given in pure intuition by a construction, but is equivalent to it. The synthesis is the production of a spatial or temporal expanse, or a series of such expanses. Indeed, Kant attributes the synthesis to the productive imagination (A118, pp. 142–43; also B152, p. 165). His examples of it includes drawing a line and describing a circle (B154, p. 167), and he characterizes it as involving “Motion, as an act of the subject” (B154–55, p. 167). Flowing constructions, such as drawing a line and describing a circle, are motions or acts of the subject and these, Kant says, belong “not only to geometry, but even to transcendental philosophy” (footnote (a) to B154, p. 167); viz., it isn’t only figures that are constructed, but the space and time in which reality arises. The synthesis, then, of the productive imagination is equivalent to flowing construction which generates (synthesizes) a whole expanse or stretch, but not, of course, out of elementary parts. If this is correct, we have further evidence that the unity of apperception is the unity of a rule, for to think space and time, or to bring them to the unity of apperception, is to bring construction to the unity of apperception, and the unity of a construction (or a synthesis) is just the rule governing it. What brings a construction together for my intellectual cognizance, that is, is just the rule that governs it, giving it a beginning and end and making it something to my present cognizance. For example, I can produce an expanse of space by a construction without knowing what it is I am doing. It is the rule (say, to construct linearly in three-unit segments) that unifies what I am doing in relation to my intellectual cognizance (although the flowing nature and so the continuity belongs to the construction itself). Note that rules enable me not only to cognize what I am doing but to cognize constructions that go beyond anything I am doing or could personally do. I may presently have a rule to construct linearly in 10 million segments, or a rule for the propriety of being up to stage 40 million in a temporizing procedure. Rules, in accordance with Kant’s characterization of thought as spontaneous, enable us to represent beyond actual construction (beyond the presentation of spatial or temporal expanses in actual constructions), and so it makes sense to talk of a repertoire of rules for construction that encompass the full extent of spatio-temporal construction. With this in mind, we can restate the premise of the Deduction as follows: I must presently have a capacity for thinking or representing

44

t h e t r a n s c e n d e n ta l d e d u c t i o n

that encompasses a repertoire of rules for spatio-temporal constructions that together encompass the full scope of space and time. This I claim is the unity of apperception in relation to the transcendental synthesis of imagination by which we have “an understanding [a power of thought or cognition] that relates to all objects of the senses [to everything real that exists in space and time]” (A119, p. 143). Equivalently, this is the premise of the Deduction stripped of any purely descriptive notion of thought and any objectivist notion of space and time. The categories now are just concepts “which contain the necessary [= rule] unity of the pure synthesis of imagination in respect of all possible appearances” (A119, p. 143). The method, that is, for showing the universal applicability of a pure concept is to show that employing that concept is required for having rules for spatio-temporal construction that together encompass the full scope of space and time (the full scope of spatio-temporal construction). Such categories, then, will pertain to all cognizable reality, since all cognizable reality is what has determinate existence with respect to (the fullness of) space and time. In effect, the premise of the Transcendental Deduction is Kant’s new theory or account of the (semantical) nature or content of thought as it pertains to spatio-temporal reality.10 I claim that it is the unity of apperception in relation to the transcendental synthesis of imagination, not the unity of apperception by itself, that is the premise of the Deduction. In the A edition Kant says that the unity of apperception presupposes the synthesis of imagination, which might make it seem as if he is “deriving” the latter from the former. This, I believe, is not the case. In the B edition he says, “This principle of the necessary unity of apperception is, itself, indeed, an identical and therefore analytic proposition” (B135, p. 155). Surely an analytic proposition can not be the full premise of a deduction of synthetic a priori knowledge (of the applicability of the categories). In the B edition Kant equates the first or primary or basic nature of the understanding with the unity of apperception in relation to the synthesis of imagination. He says, “The first pure knowledge of understanding, then, upon which all the rest of its employment is based .l.l. is the principle of the original synthetic unity of apperception” (B137, p. 156). The example of this that he gives is the unity of consciousness in rela10. The “method” for deducing categories simply will not work in an account of cognition such as Strawson’s, according to which thoughts are descriptive and space and time are objective. For example, the arguments of the Analogies for the relational categories will not go through simply from the requirement that we be able to cognize things in relation to space and time, for the arguments depend essentially on cognition being rules for spatio-temporal construction. See Essay 6 in this volume.

Apperception and Transcendental Deduction

45

tion to drawing a line (that is, a pure spatial manifold). The further premise that is added to the unity of apperception, then, to give the original synthetic unity of apperception, is that space and time provide a “manifold of a priori intuition for a possible knowledge” (B137, p. 156). Further, Kant quite clearly holds here that this manifold is produced or constructed (drawn), and so due to the synthesis of the productive imagination. Thus the original synthetic unity of apperception, and so the premise of the Deduction, is the unity of apperception in relation to the transcendental synthesis of imagination. It should be clear that the pure synthesis of imagination (involving the constructive nature of space and time) is part of Kant’s conception of the nature of the pertinence of thought to reality (and so part of Kant’s “semantics”). Since it is part of the premise of the deduction of the categories, it clearly doesn’t belong to whatever Kant is signifying by the “Subjective Deduction.”11 Furthermore, it is a component of what it is for possible perceptions or appearances to be cognizable, not a way of “forcing” them to fit or conform to conditions of cognition, or a way of “guaranteeing” that they do. The only reality that might not fit the synthesis of imagination is reality that is not spatio-temporal at all. What is true, however, is that space and time are “imposed” by the subject, but exactly in the sense that they are constructive, and so therefore any reality that is in space and time is in the compass of the subject’s construction. Nor does the synthesis of imagination belong to any transcendental psychology as some sort of nonempirical mental operating that is best excised from Kant’s more “analytic” argumentation. Kant is telling us what he thinks the capability of thought in relation to sensible reality comes to, not what sorts of mind-processing mechanisms produce or underlie such thoughts. His talk of pure synthesis doesn’t signify some nonempirical mental processing but rather the constructive nature of space and time. Kant is no more doing transcendental psychology than a mathematical constructivist regarding numbers is doing transcendental psychology. The terminology of constructivism will include notions such as “acts,” “rules for acts,” etc., but this terminology pertains to what numbers are, not to mental mechanisms for cognizing (real, nonconstructive) numbers. For the same exact reasons, the synthesis of imagination does not belong to empirical cognitive psychology either. It doesn’t refer to present mental processing but to what the very content of thoughts is (viz., constructions or procedures). Having rules for constructions, that is, is part of the nature or 11. See Axvi, p. 12. Only the empirical threefold synthesis belongs to the subjective deduction.

46

t h e t r a n s c e n d e n ta l d e d u c t i o n

character of what it is to think, not part of the mechanisms or processing by which we are able to think. Just as how the pieces are supposed to move in chess is no part of the cognitive psychology of chess but part of what it is to play chess, so too that thought pertains to constructions is no part of the cognitive psychology of thinking but part of what it is to think (what it is to have thoughts that bear on spatio-temporal reality). One might as well read mathematical constructivism as a cognitive psychology as read Kant’s synthesis of imagination that way. Both those who would excise the transcendental synthesis of imagination from Kant as being some sort of bogus transcendental psychology and those who would insist on retaining it as a component of processing in a legitimate cognitive psychology misunderstand this synthesis, and so the premise of Kant’s Deduction, and so his whole understanding of how thought pertains to reality.

4

Categories, Logical Functions, and Schemata in Kant

In the first edition Transcendental Deduction of the categories Kant does not mention the logical functions of judgment. In the second edition (the B edition), the Deduction can be said to be dominated by the logical functions of judgment. A transcendental deduction supplies a method for showing that pure concepts can have applicability. My contention is that the two deductions constitute exactly the same method, and so are the exact same deduction.1 The difference between them, rather, is in the characterization of the pure concepts that the method is supposed to be a method for. The categories of the A edition become the logical functions together with their schemata in the B edition. This doesn’t mean that Kant has split the A edition notion of categories, since the A edition categories are equivalent to just the schemata themselves. The B edition simply adds the logical functions to the A characterization of the pure concepts. The rationale for this addition is that Kant’s radically new theory of cognition had so changed the notion of judgment or thought that the issue of the relation of judgment, thus newly understood, to logical reasoning was called into question. I believe that the picture I shall present clarifies not only the structure of the B edition Deduction but the nature of the Metaphysical Deduction and the Schematism as well.

1 . t h e t r a n s c e n d e n ta l d e d u c t i o n in the a edition In the first of what Kant calls the preparatory sections of the Deduction, he characterizes objective cognition or objective representation as cognition that involves a constraint that “prevents our modes of knowl1. For references to a pervasive counterview that the B Deduction constitutes a significant departure, see Peter Baumanns, “Kants transzendentale Deduktion der reinen Verstandesbegriffe (B),” Kant-Studien 82 (1991): 329–48, and 83 (1992): 60–83.

47

48

t h e t r a n s c e n d e n ta l d e d u c t i o n

edge from being haphazard or arbitrary” (A104, p. 134).2 Kant holds that this constraint cannot be from an object outside our sensible representations. I believe that such objects, for Kant, would have to be represented purely conceptually or descriptively,3 a kind of representation Kant had allowed in the Inaugural Dissertation but soon after came to reject. In any case, Kant locates the constraint, rather, in rules for sensible representations (A105, p. 135). My actual sensible representations or reactions may be constrained by a rule of how it is proper or legitimate to react. Indeed, how I actually react may agree (correspond) or not with a rule of how it is proper to be reacting. This “unity” of reactions (sensible representations) under a rule is equally a necessary unity, since a rule unifies according to how it is required or necessary to proceed. Objective unity, in thus being identified with rule unity, is said by Kant to be “nothing other than the formal unity of consciousness” (A105, p. 135), or “nothing but the necessary unity of consciousness” (A109, p. 137). The unity of a rule, I suggest, is the unity of apperception.4 Our cognitions or thoughts, that is, are rules, so that the unity of our intellectual or cognitive consciousness in regard to the sensible is in terms of rules for proper sensible reactions. Thus, without going outside sensible representations (appearances), Kant has imported intellectual or objective representation into the sensible realm, by equating it with rules for reacting. In the second of the preparatory sections of the Deduction (A111– 14, pp. 138–40) Kant introduces the idea of one single experience, or one and the same general experience (A110, p. 138), to which all possible perception belongs. Rules enable us not only to constrain our actual reactions but to extend cognition beyond actual experience altogether. Thus it may have been proper to react in such and such a way a long time ago (before my birth), even though such reaction is completely beyond my actual experience. Kant is saying here that not only do we cognize objectively but we cognize a world extending way beyond the course of actual experience. All possible appearances, Kant says, must stand in relation to apperception (A111, p. 139). My present cognitive ability,5 that is, must encompass a set or repertoire of rules that 2. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. 3. For the inadequacy of descriptive representation, see Essay 10 in this volume. 4. At A119, p. 143, Kant identifies the unity of apperception with the understanding, and at A126, p. 147, he in turn identifies the understanding (i.e., the power of thought) as the faculty of rules. Kant’s repeated contention that the unity of apperception is a necessary unity (viz., necessity is literally its unity) can thus be understood, as well, as the contention that the form of thought is rule-form (how it is necessary or required to react). 5. Since apperception stands in relation to proper reactions beyond what can possibly

Categories, Logical Functions, and Schemata

49

together cover the full scope of all proper reactions ranging over our “entire sensibility” (A111, p. 139), or over the full reach of space and time. This is an utterly central characterization of our cognitive power for Kant. I shall be suggesting that there is no understanding of Kant’s proofs of substance and causation if one thinks that the “functions of synthesis,” as Kant puts it (A112, p. 139), pertain basically to other possible aspects of one’s actual experience (such as the back side of a perceived object), as opposed to possible experience completely beyond actual experience. For Kant we always cognize even our actual experience as “situated” in a world that extends completely beyond it. It is in the A Deduction proper6 that Kant introduces the productive synthesis of imagination (A118, p. 142). If the unity of apperception (rule unity) is to encompass all possible (that is, proper) reactions, it must relate to the pure synthesis of imagination. Indeed, it must combine or unify that synthesis. Now, the synthesis of imagination produces or constructs the pure manifold of space and time. Already in the Aesthetic Kant has argued, I believe, that space and time are constructions. Just as for a constructivist in mathematics numbers exist only in constructions (say, as termini of possible counting procedures), so too for Kant space and time exist only in flowing constructions. His argument, I believe, is that continuity is that kind of utter seamlessness of a whole (such as a spatial region) that precludes the whole being made up out of elements. In a flowing construction, such as the producing of a line segment, the construction of the whole sweep or flow is prior to the construction of parts, which, rather, are cuts made subsequently to the flow. Indeed, I believe that what Kant means by space and time being given in pure intuition is that they are, or exist in, flowing constructions.7 An example of a temporal construction would be the downbeat gestures (flowing constructions) of an orchestra conductor, which temper or mark time for the orchestra’s playing. I suggest now that the synthesis of the productive imagination is exactly such flowing construction; viz., the synthesis is not putting an extent of space together out of basic elements, but rather generating a spatial extent in a flow.8 This is belong to my personal history, apperception does not refer to the idea of a unified subject through various (actual) experiences. Apperception, I suggest, is more closely to be understood as the present cognitive capacity that I am (that identifies me as an intellect). 6. A116–19, pp. 141–43. This is the deduction “from above” (starting with apperception). I shall not directly consider the deduction “from below” that follows, since I believe it is the same deduction. 7. Thus in producing a line I am in immediate singular relation to the very object (the production); viz., I intuit or exhibit it. Further, I do so independent of sensation (affection by an object), and so the producing is given in pure intuition. 8. In the Aesthetic Kant says that space and time are given in pure intuition. However

50

t h e t r a n s c e n d e n ta l d e d u c t i o n

made clear in the B edition Deduction, where Kant characterizes the transcendental act of imagination as a “motion .l.l. of the subject” (B155, p. 167), viz., a flowing construction of the subject. The relation of the unity of apperception to the transcendental synthesis of imagination is that apperception brings construction under rules for proper constructing. Finally, the way apperception has rules encompassing all possible appearances (that is, the full scope of proper reactions), is by having rules for all possible spatio-temporal constructions, since all proper reactions take place within the compass of proper constructions.9 We now have all the elements of Kant’s account of cognition. Cognition is the capacity to form rules for the full propriety of spatio-temporal construction, and thereby for the full propriety of empirical reaction. If there are concepts “which contain the necessary unity of the synthesis of imagination in respect of all possible appearances” (A119, p. 143)—i.e., concepts that are required for bringing the full scope of spatio-temporal construction to rules—then these concepts apply to all possible appearances (that is, proper reactions). This, then, is the method for deducing categories, or for showing that categories have applicability.

2 . s u b s ta n c e a n d c au s at i o n The Transcendental Deduction basically sets out Kant’s theory of cognition. Any concepts, then, that are required to effect or realize cognition, so understood, are concepts that apply to any cognizable reality. It is in the Principles that Kant applies this method. In order to see how this method works, and in order to have examples to refer to in our later discussion of logical functions, I shall consider how Kant establishes the applicability of the concepts of substance and causation.10 Rules for temporally constructing or marking time have to encompass past time. Indeed, the propriety of now going ahead to mark time, as with a series of downbeat gestures, does not “begin” a new time but must be a “continuation” of already ongoing time. Thus Kant says that (see the preceding footnote), he means by this that expanses of space and time are “given” by flowing constructions. This understanding of their being given in pure intuition is by no means inconsistent with their being produced by a synthesis (i.e., a flowing generation of an expanse) of the imagination. 9. Spatio-temporal construction is not only given in pure intuition but is the form of all empirical intuition as well. Proper reactions have their place and time by being proper at certain stages of spatio-temporal construction. 10. What follows are sketches of Kant’s arguments in the first two Analogies. For further discussion, see Essay 6 in this volume.

Categories, Logical Functions, and Schemata

51

time is the “permanent form of intuition” (B224, p. 213). Note that it is constructive time (time as a form of intuition) that must be permanent (viz., that must not begin anew with a present construction). It must then be proper to be presently in the course of temporizing procedures, which then one’s present constructions continue. If, for example, the procedure is to mark time by a series of downbeat gestures accompanied by a recitation of numerals in order, then there must be rules for being in the middle of such a procedure, or up to a stage k in the procedure. But it cannot be proper to be in the middle of a procedure without having carried out earlier steps, since a procedure is exactly a construction to be carried out in order. Since I haven’t been performing earlier stages, the question arises of how it can now be proper for me to be in the course of the procedure. Suppose there is a procedure for baking a cake according to which one first puts in certain ingredients, mixes them, puts in certain further ingredients, etc. Suppose I have this procedure, and that I walk into a room and find the cake already partially prepared in the mixing bowl, so that it only needs brown sugar added to be ready for the oven. Then it is proper for me to be up to (as far along as) the brown sugar stage of the procedure. Equivalently, it is proper for me to be past putting in oil or eggs, despite the fact that I never put them in. Something in my present circumstances to which my procedure is geared makes it proper to be in the course of the procedure, without my having performed the initial stages. Likewise, if I am to be in the course of a temporizing procedure, then something in my present circumstances to which it is geared must set me ahead or put me at that stage.11 This something will not be objective time itself12 but will have to be something “in the objects of perception” (B225, p. 213). It must be the case, that is, that something presently real (something that presently affects me)13 is so far along in its continuing existence or reality that to “keep up with it,” it is proper to be so far along in a temporizing procedure geared to its existence ( just as the present mixing bowl of ingredients is so far along in its “cake-baking existence” that, in order to keep up with it, it is proper to 11. Whatever this something is functions as the “substratum” that represents time in general (B225, p. 213), or as the basis of representing the permanence (ongoingness) of time as a form of intuition (i.e., constructive time). 12. B225, p. 213. It can’t be, that is, that it is in order to keep up with how far objective time has progressed that I have to be so far along in my temporizing. The reason is that there is no objective time. Recall that the continuity or expansiveness of time is incompatible with any existence of it other than in a flowing construction. 13. Kant says that the real “is an object of sensation,” where (the matter of) sensation “contains the consciousness that the subject is affected” (B207, p. 201). The real, then, is that which affects.

52

t h e t r a n s c e n d e n ta l d e d u c t i o n

be so far along in the cake-baking procedure). What gears a temporizing procedure to continued existence is tracking, or keeping track of what exists. If existence pertains to that which affects us, then continued existence pertains to continued affection in keeping track of what affects. It is not then the temporizing procedure itself (the time-marking procedure of a series of downbeats accompanied by recitation of numerals), but that procedure geared to, or imposed on, keeping track, which keeps up with the existence of what is presently real (or keeps up with how far along what is present is in its continued existence). We can summarize this in the following representation: With respect to what presently affects (that is, the real), it is proper to be so far along (say, up to stage k) in a temporizing-cum-tracking procedure. This represents what presently affects as that which is proper to have been tracking, and so as what is now so far along in its existence. Equivalently, it represents the permanence of the real (at least through k stages of temporizing). Thus permanence of the real, or substance, is the substratum of the permanence of time as a form of intuition; i.e., the lasting existence of what presently exists is the basis of rules for temporal construction that pertains to the past. Note then that in accord with the method set out in the Transcendental Deduction the concept of substance is shown to be required to bring the pure synthesis of imagination (time-construction pertaining to the past) to the unity of apperception (to a present rule of propriety regarding such construction). Without the concept of substance, there are simply no rules for being in the course of temporizing constructions. Furthermore, and still in accord with the method of the Deduction, proper reactions or appearances in past time are thereby also brought to present apperception. Thus, if the procedure is augmented to one in which reacting “red” accompanies the initial stage of temporizing, we get the following representation: (1) With respect to what presently affects, it is proper to be up to stage k in temporizing-cum-tracking from first reacting “red.” In (1) the reaction is proper with respect to what presently affects me, only not now but at an initial stage of keeping track of it. Because it is thus proper with respect to the permanence of the real (that is, the trackable existence through a temporizing construction), it is, in that sense, simply a determination or mode of substance (of what is permanent). Thus, in sum, past proper reactions can be brought to present rule (present apperception), or can be part of one single experience in which all perceptions (reactions) have their place, only by the applica-

Categories, Logical Functions, and Schemata

53

bility of the concept of a substance of which they are determinations (that is, in regard to which they are proper). Although the purported scope of cognition is just the full propriety of reactions, it is only by those proper reactions’ being determinations of (proper with respect to) substances that they can be cognized. This applicability of the category of substance to appearances (proper reactions) is a refutation of phenomenalism, since without substance the proper reactions of the phenomenalist cannot be represented or cognized.14 In the Second Analogy it isn’t time itself (as a magnitude or expanse) that is at issue, but the nature of the time series. That nature is that each time in the series advances necessarily to, or forces, the later time. It doesn’t just happen to be that the succeeding time comes after the preceding; rather it emerges from it or is determined by it to arise (A194, B239, p. 222).15 It is this character of necessary advance that must be brought to present rule. Neither time construction itself 16 nor an objective (extra-constructive) temporal series17 can represent the necessary advance. Hence it is only the appearances themselves (proper reactions) that can represent the necessary advance. It must be, that is, that “the appearances of past time determine all existences in the succeeding time” (A199, B244, p. 225). The necessary advance of appearances, that is, represents the necessary advance of the time series as follows: (2) With respect to what presently affects, it is proper to be in the course of a series of reactions, each of which necessarily advances to, or determines, the succeeding reaction.18 14. Note further that the basis or authority of the rule in (1) that includes reacting is the real that is permanent (viz., substance). This entails that the rules for reaction are not arbitrary or invented but are due to phenomenal affection. This ensures that these rules express real objectivity or constraint (an arbitrary or invented rule being no real constraint at all). 15. See also A199, B244, p. 225, where Kant says that it is a law of our sensibility (time) “that the preceding time necessarily determines the succeeding.” 16. Nothing “determines” the propriety of now going ahead to temporize or produce a series of downbeats. It just is, all on its own, proper. This propriety may come after (may continue) the propriety of being in the course of a producing a series of downbeats, but it doesn’t necessarily advance from that propriety. 17. All temporality is in relation to the continuous expanse that time is, and, as per the Aesthetic, only within construction is there continuity. There simply is no objective temporal series (whether understood relationally or absolutely). 18. In (2) the series is with respect to the real (what presently affects). Once again, because what affects makes it proper to be in the course of a series of reactions, it is not only the real but the permanence of the real (i.e., substance). I am thus following Kant here in relativizing the necessary series (and so causation) to determinations of a single substance.

54

t h e t r a n s c e n d e n ta l d e d u c t i o n

But now to say that the propriety of reacting a certain way determines (or necessarily advances to, or forces) the propriety of subsequently reacting another way, is to express a causal relation between the proper reactions. Thus, in order to bring the character of the time series (as one in which the preceding necessarily determines the succeeding member) to apperception, the concept of causation is required, where the concept of causation consists in “the succession of the manifold, in so far as that succession is subject to a rule” (A144, B183, p. 185), viz., a rule that each necessarily advances to the next. Only by representing any proper reaction as emerging within a series of proper reactions where each determines the next can I represent the reaction within the (necessary advance of the) time series, and so represent it as determined with respect to the unity of time under a present rule.19

3. logical functions Neither the categorical nor the hypothetical function of judgment played any role in our account of the A edition Deduction of substance (permanence of the real) or causation (necessary determination of the succeeding by the preceding appearance). It isn’t until the B edition that Kant incorporates logical functions of judgment into the Deduction. The logical functions are introduced in the A edition in Section 9 (A70, B95–A76, B101). They are the forms of judgment so far as judgment is involved in reasoning (general logic), and perhaps in abstract thinking generally. Kant goes on in the next section (the Metaphysical Deduction) to derive the list of categories. The significance of this deduction, I believe, is that Kant has set out an entirely new theory or account of intellectual cognition or thought in the Transcendental Deduction, according to which cognitions are rules for the propriety of reacting. The two preparatory sections of the A edition Deduction make clear that it is indeed a theory of intellectual cognition, since the rules account both for objective cognition and for cognition beyond actual experience to the full scope of proper or possible reaction. Further, these rules pertain to what is “outside” the understanding (viz., the sensible, or proper reacting), and these rules may be true or false (viz., a rule that imputes the propriety of reacting in such and such a way may be an incorrect rule, since it may not be proper to react in 19. An objective succession (such as ship upstream, ship downstream) is only objective because it has a place in the necessary advance of the time series as represented in (2). Irreversibility may be the rule distinguishing succession from coexistence, but necessary determination in the time series (i.e., causation) is what makes that succession objective in time.

Categories, Logical Functions, and Schemata

55

such and such a way). In sum, these rules are complete thoughts (that is, cognitions capable of truth and falsity) pertaining to sensible reality. Not only, then, is Kant’s account an account of intellectual cognition, but it is an account of judgment, since a judgment is exactly what pertains to reality in such a way as to be capable of truth or falsity. Thus Kant has a whole new conception of judgments as rules for proper reacting. Now, we also reason and think abstractly in judgments. If the notion of judgment is to be a unitary notion (viz., if there is to be a single phenomenon of judgment involved both in reasoning and in cognizing sensible reality), then it is incumbent upon Kant to show how judgments in reasoning are related to judgments in real cognition. After all, so far, on his new theory of judgment as cognition, there is nothing that “looks like” categorical or hypothetical form. In this sense it is the very unity or integrity of the understanding as the faculty of judgment that is at stake in the Metaphysical Deduction. What I am suggesting, then, is that the issue Kant characterizes as the issue of the “origin” of the categories is just the issue of the understanding being a unitary capacity of judgment, operative both in reasoning and in cognizing the sensible—an issue made quite severe by Kant’s radically new account of the nature of judgment in cognition of the sensible.20 Besides the issue of the origin of the categories, Kant sees the derivation from logical functions as providing systematicity and completeness to the categories. I shall consider these issues briefly, but the main issue is the issue of judgment as a unitary capacity. The list of logical functions themselves can perhaps be faulted over their systematicity and completeness, which, of course, would call into question how the categories then could inherit these features from them. First, there can be alternative lists, each providing sufficient forms for reasoning (as in modern quantificational logic).21 This is no problem, as far as the issue of there being a unitary judgmental capacity is concerned, so long as at least one complete list of logical functions can be paired with categories. Second, one may wonder over the completeness of the logical functions, but from a modern point of view model theoretic completeness provides a standard for a complete proof theory and hence a standard for adequate logical form. Of greater concern is Kant’s apparent “doctoring” of the list of logical 20. The origins issue, then, seems to me to be a real one. Theoretically, Kant had two options. Give a new theory of reasoning directly in terms of rules for proper reacting (hence, abandoning the traditional logical forms), or show how the logical functions are operative even in his new theory of cognition. He clearly saw only the latter as an alternative. 21. See Peter Strawson, The Bounds of Sense (London: Methuen, 1966).

56

t h e t r a n s c e n d e n ta l d e d u c t i o n

functions—for example, his distinguishing the singular from the universal and the affirmative from the infinite.22 As long as the doctoring is not done in terms of his new rule theory of cognition, however, there is no circularity in the derivation of categories that relate to that new theory. Kant’s ground for distinguishing the singular from the universal is that they express different quantities of knowledge. His ground for distinguishing the affirmative from the infinite is that they express or have a different content of knowledge. Whatever he means by these remarks, it is clear that he thinks that they differ in these ways in a context of purely abstract thinking.23 If judgment is involved in purely abstract thinking, such as thinking purely conceptual connections, and some of its forms (singular, limitative) in this context go beyond mere logical form, they are still forms of judgment. Hence the unitary character of judgment then becomes that forms of judgment in relation to Kant’s new account of judgment as cognition should cohere with the forms of judgment in two other contexts (viz., logical reasoning and abstract thinking), which is no circularity. Finally, the supposed idea that general logic depends on transcendental logic,24 so that the former cannot be the source of the latter, is also, I believe, harmless. As far as the unitary nature of judgment goes, all that matters is that the forms cohere between both contexts (reasoning and Kant’s new theory of cognition by rules). Kant states the supposed priority of transcendental logic in footnote a to B133–34. Even here, all he says is that general concepts presuppose a synthesis, and (supposedly) general logic requires general concepts (since syllogistic logic is a logic of concepts). This much by no means implies that judgment forms are involved in this presupposed synthesis.25 If they aren’t, then the synthesis presupposed by general logic does not imply that general logic (and its forms) presuppose the categories (forms of judgment involved in Kant’s new account).26 22. See Henry Allison, Kant’s Transcendental Idealism (New Haven: Yale University Press, 1983), 128. 23. Nowadays the distinction between singular and universal is purely logical, and the distinction between negation (rather than affirmation as per Kant) and limitation can be made out logically, perhaps, in multivalued logics. 24. See Norman Kemp Smith, A Commentary to Kant’s Critique of Pure Reason (London: Macmillan, 1918), 184–85, 196. 25. For example, I can represent different reactions of “red” in (1) above (different for being proper at different stages of temporizing-cum-tracking), and so come to a “conceptus communis” or general concept of “red,” without any employment of logical functions. 26. As to the problem of the supposed lack of fit between the disjunctive judgment and the category of community to follow, it seems replacing disjunction with the bi-conditional would suffice.

Categories, Logical Functions, and Schemata

57

We turn now to the Metaphysical Deduction proper in Section 10 (A72, B102–A83, B109). In this section Kant first rehearses or outlines his theory of cognition from the Transcendental Deduction. The reason for this is simply that he is leading up to the idea of logical forms of judgment being also forms of judgment according to this new theory of cognition or judgment. The key passage is the paragraph at A79, B105, pp. 112–13. In this paragraph Kant identifies the logical functions as, indeed, functions that have a role in his new account of cognition as synthetic unity of intuitive representations. These functions of reasoning also, that is, bring sensible intuitions to the unity of apperception (or, equivalently, these functions operate to yield rules for proper reactions). I contend, however, that although they are characterized as having this intellectual cognitive role of bringing sensible intuitions to apperception, Kant is here purposely abstracting from the idea that this unity of representations is via the unity of the pure manifold of space and time. First, the concepts he comes up with in the following table of categories are exactly the concepts that he later schematizes,27 but it is only the logical functions as schematized that transcendentally determine the pure manifold of time. Second, he chides Aristotle for confusing “modes of pure sensibility” with categories (A81, B107, p. 81), which suggests that he is understanding the role of the logical functions apart from pure sensibility. Third, in the B edition Deduction, where he refers to the present section and its categories, he says, “since the categories have their source in the understanding alone, independent of sensibility, I must abstract from the mode in which the manifold for an empirical intuition is given” (B144, pp. 160–61, emphasis added). The significance of the fact that Kant in the first sentence identifies these functions as giving unity to representations in an intuition, while in the second sentence he relates them to the unity of the manifold of intuition in general, I believe, is that his new theory of cognition has two components. First, as in the first preparatory section of the A edition Deduction, the unity of reactions under rules is objective unity, which pertains to constraining what is actually given (how we actually react) in an intuition. Second, as in the second preparatory section of the A edition Deduction, the unity of reactions under rules is the unity of (merely) possible reactions beyond what is actually given, by which all perceptions “belong to one and the same general experience” (A110, p. 138). For Kant, the power or nature of thought ( judgment or intellectual cognition) is that it goes beyond mere passive reception (that is, 27. Compare this table with the passage from A142, B182–A145, B184, pp. 183–85.

58

t h e t r a n s c e n d e n ta l d e d u c t i o n

actually reacting), by constraining that reception (and so representing objectivity), and also by representing it as part of the full scope of proper reactions (and so unifying possible experience in general). In this paragraph, then, Kant is identifying the logical functions as having both these roles in his new theory of cognition. In sum, then, Kant is saying that the logical functions, or the forms of judgment in reasoning, are also forms of judgment as cognition; i.e., forms for objective cognition constraining actual experience and forms for cognizing beyond actual experience. Indeed, they are forms for bringing sensible intuition to the unity of apperception, where the unity of apperception is objective unity as per the first preparatory A edition section of the Deduction (A105, p. 135), and also that unity which stands in relation to all possible appearances as per the second preparatory section (A122, p. 139). The table of categories, then, is a table of the logical functions of judgment, only conceived as bringing reactions under the unity of rule. The basis of this conception (and so this derivation) of them is simply the assertion of the unity or integrity of the notion of judgment itself; viz., judgment is both the unit of reasoning and the unit of cognition, so that if there are forms of judgment, these are both forms of judgment for reasoning and for cognition (that is, for bringing sensible intuition under the unity of apperception). This, I claim, is the Metaphysical Deduction. Of course Kant is not showing that these concepts have applicability. The concept of a form of judgment as also being cognition (as bringing sensible reactions under rule unity) is still, by itself, a “mere form of thought, without objective reality” (B148, p. 163). Of course, if they have applicability, then they apply a priori, since their origin and their very content as concepts are not in the sensible given but in the nature of cognition ( judgment) itself. However, it is only via their schemata that they could apply, or that they could be concepts that bring sensible intuition under rule unity. Not only, then, is Kant not deducing that they do or must apply; he is not even deducing that they can apply. All he is deducing is that there are purely intellectual concepts, viz., concepts of forms of judgment purportedly effecting intellectual cognition, where that cognition is a matter of bringing sensible intuitions under rules. Roughly, then, in the Metaphysical Deduction Kant is merely asserting that judgment is a unitary and integral capacity. His full proof or defense of this assertion (in relation to his new theory of judgmental cognition) takes the whole Transcendental Analytic.

Categories, Logical Functions, and Schemata

59

4 . s c h e m ata Kant says that in applying a category to appearances, we set its schema “alongside the category, as its restricting condition” (A181, B224, p. 212, emphasis added).28 Thus both the logical function of judgment and the schema are components of the cognition of appearances. In our representation of substance in (1) above, we had the permanence of the real, which is the schema of the logical function of subject-predicate. If we now put this schema alongside that logical function we get: (3) With respect to what presently affects, it is proper to be up to stage k in a temporizing-cum-tracking procedure that begins with reacting “red” and saying, “That is red.” Now, the procedure in which we are up to stage k begins not just with reacting to what is trackable, but also saying (or thinking) “That is red.” In (3) this logical function of subject-predicate is indeed functioning to express the relation of sensible intuitions (proper reactions) to the unity of apperception (to the unity of a present rule); i.e., it is functioning as a judgment or cognition according to Kant’s new conception of cognition, and according to his characterization in the Metaphysical Deduction. In (3) I am not now making the judgment “That is red.” Indeed I cannot make that judgment with that significance (of pertaining to past reality), because I am not properly situated in time to make that judgment. I can’t go back in time and say, “That is red.” Although I cannot make that judgment, I can license making it if I can presently represent the significance of making it (viz., its significance as pertaining to past reality), which in turn is a matter of representing the situation in which it can be made. This requires a determination of time; i.e., a determination of my present situation as at the kth stage of a temporizing procedure whose first stage is the situation for making the judgment. In this manner, a “formal and pure condition of sensibility” (that is, time) restricts (as it enables) the employment of the logical function (A140, B179, p. 182). This time determination enabling the logical function, further, is “in conformity with the unity of apperception” (A142, B181, p. 183) in that representing the significance of the judgment involves bringing past time to the unity of apperception or the unity of a present rule. In (3), indeed, past time is determined by a rule (the propriety of being up to stage k in temporizing). Finally, what 28. See also A181, B223, pp. 211–12, where Kant says that appearances are subsumed “not simply under the categories, but [also] under their schemata.”

60

t h e t r a n s c e n d e n ta l d e d u c t i o n

enables the sense of the logical function is not only a time determination in relation to a rule but one that is in accord with a concept (of permanence of the real). Kant says that the schema is a “pure synthesis [time-determination] determined by a rule of unity [apperception] in accordance with concepts” (A142, B181, p. 183). In (3) above, it is the concept of the permanence of the real (that what presently affects is what is proper to be in the course of tracking) by which past time is determined in relation to the unity of apperception (by which it is proper to be in the course of temporizing). Again, Kant says, “the schema contains and makes capable of representation only a determination of time” (A145, B184, p. 185, emphasis added). In (3) the permanence of the real (that is, the propriety, with respect to what presently affects, of being up to k in tracking) contains or makes capable of representation a determination of time (that is, the propriety of being up to stage k in temporizing) or, equivalently, contains the pure synthesis of time in relation to apperception. Although the transcendental determination of time is what is homogeneous as between logical functions and sensible intuitions, Kant does not identify the schema with the transcendental determination but rather with a concept that contains it or with which it is in accord. If the schema is a concept containing a determination of time or, equivalently, a pure synthesis determined by a rule of unity in accordance with concepts, then it is the entire representation in (3), other than the subject-predicate judgment itself, which is the schema of that judgment. It is this schema that “mediates the subsumption of appearances under the category” (A139, B178, p. 181), where the category is the logical function of judgment as cognition. The subject-predicate judgment, that is, is represented as pertaining to that possible appearance (that proper reaction “red”) via bringing that possible appearance to the unity of apperception by the permanence of the real—in effect, by the entire representation in (3) other than the subject-predicate judgment. In relation to the Metaphysical Deduction, we note that in (3) not only does the unity of apperception (rule unity) extend to possible appearances, but the subject-predicate form of judgment does too, as “giving expression” (A142, B182, p. 183) to the schema. Indeed, in (3) the judgment that is licensed signifies or expresses the permanence of the real since the term “that” pertains to a real that is trackable into the present. Kant’s new theory of intellectual cognition or thought as rules is now consistent with intellectual cognition as judgment having logical form. What the schema effects is not just cognitive unity as rule unity, but cognitive unity as unity for logical functions of judgment. Kant’s characterization of the derived logical functions (the categories) in the

Categories, Logical Functions, and Schemata

61

Metaphysical Deduction is now finally realized, since the subject-predicate form of judgment expresses the relation of reacting “red” (that possible experience, or that empirical intuition) to the unity of apperception. I believe what Kant says of the schemata of pure concepts applies to causation as well. If we add the logical function of the hypothetical judgment to (2) above we get: (4) With respect to what presently affects, it is proper to be in the course of a series of reactions each of which determines the next, while being in the course of saying of each pair, “If that is so and so, then that is such and such.” Since each of the reactions is with respect to what presently affects, each is a determination of substance, which gives sense to the occurrences of “that” in (4). The representation in (4), minus the hypothetical judgment, signifies “the succession of the manifold, in so far as that succession is subject to a rule” (the rule that each determines or necessarily advances to the next [A144, B183, p. 185]). Each hypothetical, then, expresses that the antecedent necessarily determines the consequent; i.e., that the substance being so-and-so is causally connected to its then being such and such. Once again, the schema “contains and makes capable of representation only a determination of time” (A145, B184, p. 185), or the schema is a “pure synthesis determined by a rule of unity in accordance with concepts” (A142, B181, p. 183). The pure synthesis or the time determination in this case is the necessary advance of the time order, and the concept that brings the necessary advance to present rule (i.e., to the unity of apperception) is causation (viz., that each proper reaction necessarily advances to the later proper reaction). Kant’s characterization of a schema, as a transcendental product of imagination that concerns bringing representations to the unity of apperception by a concept (A142, B181, p. 183), exactly duplicates his characterization of a category in the A edition Deduction, where he says, “In the understanding there are pure a priori modes of knowledge [that is, concepts] which contain the necessary unity of the pure synthesis of imagination in respect of all possible appearances,” and these are the categories (A119, p. 143, emphasis added).29 Compare this with his statement in the Schematism, where he says, “the schema is simply the pure synthesis [of imagination] determined by a rule of [that] unity in accordance with concepts” (A142, B181, p. 183). Of course in the 29. In the Schematism he even uses the same term when he says the schema “contains and makes capable of representation only a determination of time” (A145, B184, p. 185, emphasis added).

62

t h e t r a n s c e n d e n ta l d e d u c t i o n

Schematism the concept, such as the permanence of the real, that contains or determines the unity of the pure synthesis of imagination is also given expression (A142, B181, p. 183) by the category (the logical function as it pertains to cognition) of which it is a schema. In the A edition Deduction this is simply not the case. Even when Kant talks of the mediating function of imagination in the Deduction (A138, B177, p. 181), it doesn’t mediate between logical functions and sensible intuition as it does in the Schematism, but rather between the necessary unity of apperception as such (rule unity) and sensible intuition. The pure concepts of the A edition Deduction are essentially equivalent, then, to the schemata. The only difference is that in the Deduction they are not characterized as schemata (viz., as related to the logical functions of judgment). This is certainly not the case in the B edition Deduction, where the logical functions of judgment initially (up through Section 20) play the only role, and where even the unity of the synthesis of imagination (in Section 26) is characterized in relation to logical functions. What I contend is that this difference in characterization is basically the only difference between the deductions. Whereas the A edition Deduction supplies a method for deducing pure concepts, where that method is to show they are concepts containing or enabling the unity of imagination (time) in relation to apperception, the B edition Deduction supplies a method of deducing pure concepts, where that method is to show they are concepts, characterizable as the applicability of logical functions of judgment, containing or enabling the unity of imagination in relation to apperception. It is the exact same method, and so the exact same deduction, only the characterization of the concepts to which the method is applicable is augmented in the B edition to include that they are concepts of the applicability of logical functions. This claim, I now contend, makes sense of the structure and text of the B edition Deduction.

5. the b edition deduction The Deduction from sections 15 to 18 for the most part simply equates unity of apperception (that is, necessary unity, or rule unity) with objective unity. In this equation it reiterates the first preparatory section (A104–10) of the A deduction. There is some talk of space and time, but mostly to contrast these forms of intuition with intellectual unity. Section 18 clearly reiterates the second preparatory section (A111–14), with at least a hint that apperceptive unity also pertains beyond actual experience. All together, then, these sections simply review Kant’s new rule account of the two aspects of intellectual cognition;

Categories, Logical Functions, and Schemata

63

namely, that such cognition is objective, and that such cognition pertains beyond actual experience to experience in general. Section 19 explicitly connects intellectual cognition to the notion of judgment. Indeed, it says that besides the logician’s notion of a judgment, there is the notion of a judgment as cognition, and that notion, in proper Kantian terms, is “nothing but the manner in which given modes of knowledge are brought to the objective unity of apperception” (B141, p. 159). What I claim is that Kant is here asserting the integrity of the capacity of judgment, or that judgment is indeed a unit of reasoning, but it is also the unit of intellectual cognition, where that cognition is understood according to Kant’s new-rule account. Recall that this is exactly the assertion by which the Metaphysical Deduction proceeds. Given this integrity of the capacity of judgment, it follows that the forms of judgment in reasoning must be forms of judgment as cognition and, as such, must be understood in relation to the role of bringing sensible intuitions under rule unity. As we have seen, in the Metaphysical Deduction the categories are exactly the logical functions of judgment so understood as having this role. In Section 20, then, Kant can say that the method for deducing pure concepts (i.e., logical functions) is to show that their mode of bringing sensible intuition under rule unity is required. Note that Section 20 is not a deduction of all the logical functions. Indeed, Kant says that “all the manifold has to be determined in respect of one of the logical functions” (B143, p. 160). As far as cognition goes, all that is required is bringing sensible intuitions to rule unity or the unity of apperception. Even if logical functions have this role, it might be, for all that, that any one of them is sufficient. Thus the subject-predicate form of judgment might by itself be sufficient for bringing sensible intuitions to the unity of apperception. Note, second, that Section 20 is not even a complete method for deducing logical functions, since as far as these functions have been so far characterized, there seems to be no principle (such as various modes of time determination) by which we could show that more than one is required to bring sensible intuition to rule unity. Why should different functions operate so that only all together do they fully bring all sensible intuitions to the unity of apperception? Even more seriously, how could they (heterogeneous as they are from sensible intuitions) manage to have a unifying role at all? As a method, then, the Deduction up through Section 20 is, in effect, empty. It is for this reason that in Section 21 Kant says that in what we have so far only “a beginning is made of a deduction of the pure concepts of the understanding.” Kant’s point, so far, is not to provide a complete method but to ensure that the complete method will be a method for deducing (concepts charac-

64

t h e t r a n s c e n d e n ta l d e d u c t i o n

terized in relation to) logical functions. The entire B edition Deduction up through Section 21, I suggest, is just this point; i.e., it is exactly the incorporation of the Metaphysical Deduction into the Transcendental Deduction. The reason the Deduction up through section 20 is merely a beginning is not because Kant is deducing only that the manifold “so far as it is given in a single empirical intuition” (emphasis added) is subject to the categories.30 This is made clear in Section 26, where Kant says that in Sections 20–21 he has shown the possibility of categories as a priori modes of knowledge “of objects of an intuition in general” (B159, p. 170, emphasis added). In the second preparatory section in A, recall, he talked of all perception as belonging to “one and the same general experience” (A110, p. 138, emphasis added). Intellectual cognition is not only that cognition that enables us to constrain given appearances by rule (constituting objective cognition), but also that cognition that enables us to represent beyond given appearances to all possible appearances or experience in general. I contend that what Kant says in Section 26 indicates that he includes this second aspect of cognition as operative in Section 20. In other words, the beginning of the deduction is not to show that logical functions are involved only in objective cognition, so that then in Section 26 the deduction is completed by showing they are involved in cognition beyond what is given in a single empirical intuition. Rather, the beginning of the deduction is that the method for deducing pure concepts (logical functions) is to show their mode of unity is required for bringing sensible intuition to the unity of apperception (rule unity) both in the sense of that unity being objective unity for given appearances and in the sense of its being the unity of all possible sensible intuition, beyond given appearances to appearances in general.31 We do not have a complete method for either of these, of course, until we show that sensible intuition can be made homogeneous with the logical functions. What Kant is abstracting from up through Section 21 are the forms of intuition; not just our forms of intuition, but any forms of intuition 30. For this view, see Dieter Henrich, “The Proof Structure of Kant’s Transcendental Deduction,” Review of Metaphysics 22 (1969): 640–59. For the lack of textual basis for the view, see Richard Aquila, Matter in Mind (Bloomington: Indiana University Press, 1989), 126. 31. Recall that in the second preparatory section in A Kant says that the conditions of a possible experience in general are at the same time conditions of the possibility of objects of experience (viz., representing beyond constraining actual experience requires representation of objects as in [1] above). It is for this reason that in Section 26 Kant talks of having shown the possibility of categories in relation to “objects of an intuition in general”; viz., representation beyond the actual or the given always imputes objects.

Categories, Logical Functions, and Schemata

65

whatsoever (B144, pp. 160–61). What he cannot abstract from, he says, is “that the manifold to be intuited must be given prior to the synthesis of the understanding and independently of it” (B145, p. 161). Kant, in characterizing the logical functions in terms of his new account of intellectual cognition, cannot abstract from the fact that intellectual cognition can be cognition by rules only if those rules are rules for something (viz., rules for reacting). He can abstract from rules for constructing (i.e., the pure manifold of space and time), but to abstract from sensible intuition would leave the understanding, as a faculty of rules, with nothing to govern. In the A edition, it was clear enough that pure concepts that “contain the necessary unity of the pure synthesis of imagination in respect of all possible appearances” (A119, p. 143) have a role in intellectual cognition (viz., to effect the unity of apperception). It was not clear, however, that the content of such pure concepts— for example, the concept of substance—was an “intellectual” content. After all, permanence of the real seems to be a concept of time itself, or a concept of reality (what affects) enabling a concept of time, and so, in effect, a concept whose very content belongs to the form of intuition. Indeed, the whole point of the Metaphysical Deduction was to give clear intellectual content to such concepts (as also concepts, indeed schemata, of forms of judgment). It is no wonder then, that in the B Deduction, where Kant is so concerned to be deducing intellectual concepts (concepts characterizable as having intellectual content), he abstracts from forms of intuition (the very thing that seemed to swallow up the whole content of pure concepts such as substance). In Section 24 Kant reiterates the incompleteness of the Deduction so far when he says that the pure concepts (of Section 20) are “mere forms of thought through which no determinate object is known,” and that they relate only to a purely intellectual synthesis (B150, p. 164). Although the logical functions in Section 20 are characterized cognitively as a unity of sensible intuition under apperception, Kant is reiterating that it is not at all clear how it is that they can have this cognitive role (heterogeneous as they are from sensible intuition). Kant now introduces the figurative synthesis, or the transcendental synthesis of imagination, and in Section 26 he identifies the unity of this synthesis of imagination with the (cognitive) unity of the logical functions in Section 20. He says that the synthetic unity to which “everything that is to be represented as determined in space and time must conform” (that is, the unity of the figurative synthesis) can be no other than “the unity of the combination of the manifold of a given intuition in general in an original consciousness in accordance with the categories” (that is, the unity of the logical functions in Section 20, or the unity of the purely

66

t h e t r a n s c e n d e n ta l d e d u c t i o n

intellectual synthesis). As an example of the identification Kant is making, in (3) above the permanence of the real contains the unity of past reactions in time (viz., represents the reaction as determined in past time), and so constitutes the unity of the figurative synthesis. But now (3) is also the unity of the combination of sensible intuition (affection and reaction) under apperception in accordance with the subject-predicate function of judgment (the intellectual synthesis). Similar remarks apply to (4) in regard to necessary succession and the hypothetical judgment. It is this identification that completes the deduction started in Section 20. It completes it in two ways. First, recall that in Section 20 there was no idea of why the manifold had to be unified under more than one logical function. The identification allows us now to say that as many modes of unity under logical functions must be applicable as go with modes of unity required to bring all of space and time under apperception. Thus, if both permanence of the real and necessary succession of proper reactions are required modes of bringing time to apperception, then both subject-predicate and hypothetical functions, respectively, are thereby required modes of unity of sensible intuitions under logical functions. Second, the heterogeneity of logical functions from the sensible perceptions they are supposed to unify (which makes it problematic whether their supposed cognitive role is anything more than an empty characterization) is now overcome by their identification with concepts containing unity of time. Permanence of the real gives significance to the idea that a reaction is a determination of trackable reality. This unifies sensible intuition (reaction and affection), so that it is no longer heterogeneous from the subject-predicate form. Indeed, sensible intuition is now combined in such a way that the idea of that which is subject and that which is predicate is applicable. Similarly, the necessary advance of perceptions (which contains the necessary advance of time) makes perceptions homogeneous with the connectivity or dependence expressed in the hypothetical judgment.32 Note that the second unity, or the unity that determines appear32. Thus I believe that Guyer is wrong in contending that a line of argument leads from the Metaphysical Deduction to the establishment of categories that bypasses the categories as conditions of possible experience. Kant is not bypassing the categories as conditions of time-determination and so of possible experience. See Paul Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987), 99. What Guyer is correct about, however (see p. 77) is that Kant, so far as the A edition is concerned, proves the categories on some other ground (viz., conditions of the unity of time under apperception) and only “retrospectively” identifies them with logical forms of judgment. Nothing has changed in the B edition, except that Kant explicitly and “prospectively” introduces these forms in Section 20 to then identify them with categories as conditions of time determination in Section 26.

Categories, Logical Functions, and Schemata

67

ances with respect to space and time, is just the unity (and the only unity) of the A edition Deduction (A116–19). Further, once this second unity is identified with the unity of logical functions, as it is here in the B Deduction, it then constitutes with it the unity of schematized logical functions. Because of this, the pure concepts of the A edition Deduction are effectively identical to schemata (without their connection to logical functions). The B Deduction, then, can be regarded as adding the logical functions (up through Section 21), and then identifying their unity with the unity of the pure concepts of the A Deduction (in Section 26). The Transcendental Deduction, recall, is a method or recipe for deducing pure concepts. That method is to show that those concepts are required for the unity that constitutes cognition. In the A edition that unity is the unity of space and time under apperception, and so the method for establishing the applicability of a pure concept is to show it is required for some mode of the full unity of space and time under apperception. The B edition adds no new method. What it adds, rather, is a new augmented characterization of pure concepts, or a new augmented characterization of the unity to be effected by those concepts. Now, that unity is a unity of space and time under apperception, which is also a unity of the sensible under a logical form of judgment. The method, now, for establishing the applicability of a pure concept is to show that it effects a mode of unity of space and time under apperception that is also a unity of the sensible under a logical function. In order for a pure concept to effect this unity, it just has to be a pure concept in the A edition sense that also schematizes a logical function of judgment. In this way, then, the B edition is not a new method of establishing pure concepts but a new, richer characterization of the unity that constitutes cognition that is to be effected, and so a new, richer characterization of the pure concepts that, as effecting that unity of cognition, are concepts that the method is a method for. This new characterization ensures or incorporates the integrity of judgment as between the nature of judgment in reasoning and the nature of judgment as cognition; an integrity called into question by Kant’s radically new account of cognition as rules for the unity of spatio-temporal construction governing empirical reaction. That account of cognition appears fully in the A edition, so that the B edition is neither a new method (neither a new deduction) nor a new theory of cognition, but just the coherence of that method and that theory of cognition with the integrity of judgment.

68

t h e t r a n s c e n d e n ta l d e d u c t i o n

conclusion I close by presenting the following picture of the overall structure of the Analytic that emerges from our results: (1) The Metaphysical Deduction characterizes pure concepts as logical functions according to a cognitive role as unities of sensible intuition under apperception. (2) The Transcendental Deduction (in B) gives a method for establishing the applicability of such pure concepts. That method is to show that concepts required for bringing space and time to the unity of apperception are at the same time concepts by which logical functions are unities of sensible intuition under apperception (thus making the former concepts schemata of the logical functions). (3) The Schematism lists the particular schemata of the logical functions; i.e., it lists the particular modes of unity that are to be shown to be required for bringing time to apperception (each of which is a mode, also, by which a logical function is a unity of sensible intuition under apperception).33 (4) The Principles show that these particular modes of unity are, indeed, required for fully bringing time to apperception. 33. One may object textually that if the A edition categories are schemata, then the Schematism chapter (in the A edition) would make no sense. I believe the point of the Schematism chapter originally was to “list” the particular categories (permanence of the real, etc.) that were to be proved in the Principles according to the method of the Transcendental Deduction. When Kant prefixed the Metaphysical Deduction (after he had completed the A Deduction), instead of incorporating those functions into the Transcendental Deduction, he identified the A categories as schemata of these functions in the Schematism chapter. In the B edition, he properly incorporated the logical functions into the Transcendental Deduction, so that functionally the Schematism chapter still serves only to list the particular categories (i.e., schemata of logical functions) to be proved in the Principles.

5.

A Modified Version of Kant’s Theory of Cognition

Kant has a theory of cognition in which all thoughts or cognitions are rules for empirical reacting in the course of spatially and temporally constructing. These rules function as representations of our situation in relation to all the ways in which it is proper to interact with reality. Kant’s theory is fundamentally different from accounts of representation that involve reference to objects by uniquely individuating descriptions, or by causal (information) chains, or by a combination of both. In the first part of this essay I set out Kant’s theory of cognition in his own terms. In the second part I present a modified version of this account, while in the third part I extend the account in two important ways. Both the second and third parts add what Kant would have considered to be empirical elements. Finally, in part 4, I spell out the virtues of Kant’s account against descriptive and causal theories of cognition.

1 . k a n t ’ s ac c o u n t o f c o g n i t i o n For Kant, the purport or the intent of cognition is the full scope of possible perception.1 Perception, for Kant, is possible only in the course of space and time. Hence, representing the full scope of possible perception involves representing the full extent of space and time. Space and time, for Kant, are given in pure intuition. An intuition is a singular representation in which one is immediately related to what is represented. In a pure intuition, this immediate relation is not via sensation of an affecting object. Suppose that an extent of space is repre1. All possible perceptions, for Kant, “belong to one and the same general experience” (A110, p. 138) and the function of thought is to represent all perceptions as belonging to, or being of, a single world. Possible perceptions are appearances, for Kant, so that the purport of cognition is to represent all possible appearances. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition.

69

70

t h e t r a n s c e n d e n ta l d e d u c t i o n

sented by a construction, such as drawing a line. This construction immediately presents (indeed, it is) the very object (the extent of space) represented. Further, it represents it not by passively sensing but by producing it. Kant’s contention that space and time are given in pure intuition is equivalent, I hold, to the view that space and time are represented by constructions. Kant’s argument for this view in the Aesthetic is that extents of space and time are continuous, and continuity can pertain only to flowing constructions. For Kant, continuity is that sort of utter seamlessness that precludes the continuous whole from being constituted by parts or elements. With a continuous expanse, that is, the whole is prior to the parts (A25, B45, p. 69). Now, in a flowing construction, such as in drawing a line, the whole is swept out prior to any cuts that divide the line into parts. By contrast, any objective (that is, extra-constructive) expanse of space and time (whether relational or absolute) would be such that the whole actually exists only if all its parts or components exist. Therefore, an objective whole is composed of its parts or elements, and so cannot be continuous.2 An example of a temporal flowing construction would be an orchestra conductor’s marking time (pacing or tempering the orchestra) by a series of downbeat gestures (flowing constructions). Kant clearly retains this constructivist account of space and time in his theory of cognition in the Transcendental Deduction. There, the pure manifold of space and time is attributed to a synthesis of the productive imagination (A118, p. 142). This synthesis is not a combination of ultimate elements into a determinate expanse of space or time, but rather a generation of expanses by flowing constructions. The synthesis, Kant says, is “motion, as an act of the subject,” or “motion, as the describing of a space” (B154, p. 167; also footnote (a) to B154, p. 167). The understanding, which is the faculty of thought for Kant, gives unity to the pure synthesis of imagination (A79, p. 122) by bringing it under the unity of apperception (B135, p. 167). The unity of apperception is the determination of the manifold by the unity of a rule (A105, p. 135). Indeed, Kant characterizes the understanding as the “faculty of rules” (A126, p. 147). The unity of apperception, then, is the unity of rules,3 and hence our conceptual or intellectual cognition 2. This argument shows not only that we represent space and time by construction, but that in so far as they are continuous, they are mere constructions. Whether this continuity argument of Kant’s is correct is not central to his theory of cognition, which requires only that we represent space and time by construction. Our view that the very concept of continuity for Kant involves construction agrees with Friedman’s view that pure intuition is required for the very significance or meaning of mathematical concepts. See Michael Friedman, Kant and the Exact Sciences (Cambridge: Harvard University Press, 1992), 129. 3. Rule-unity is a unity involving necessity, since a rule unifies according to how it is

Kant’s Theory of Cognition

71

of space and time has the form of rules for spatio-temporal constructions. Now, space and time, for Kant, are also forms of empirical intuition or possible perception. The full scope of possible perception, then, can be brought to my present cognizance if the full scope of spatio-temporal construction can be brought to that cognizance. In other words, I can have present rules that, all together, pertain to all possible experience, if I can have present rules that, all together, pertain to the full scope of spatio-temporal construction. Since the purport or intent of cognition is all possible perception, we conclude that Kant’s account of cognition is simply a matter of having a repertoire of rules encompassing the full scope of possible perception or possible reactions in the full scope of spatio-temporal construction. Each such rule is a complete thought or a judgment. A complete thought, then, is a rule for a proper reaction within the compass of a proper construction. In this way. “All possible appearances [that is, possible reactions or perceptions] stand to original apperception” (A111, p. 139). The categories, for Kant, are concepts “which contain the necessary [i.e., rule] unity of the pure synthesis of imagination in respect of all possible appearances” (A119, p. 143). The categories, that is, are concepts that are required for bringing the full scope of possible appearances or possible perceptions to the unity of apperception (i.e., ruleunity), by bringing the full scope of spatio-temporal construction to rules. We consider now, in some detail, how this applies to the important category of substance. Kant says that “The a priori conditions of a possible experience in general are at the same time conditions of the possibility of objects of experience” (A111, p. 138). Although the purport of cognition is simply all possible perception (all possible experience), nevertheless a condition of realizing this purport is the cognition of objects of experience, or substances (enduring realities). In terms of Kant’s characterization of categories, without the application of the concept of substance to possible perception, there simply cannot be rules encompassing the full scope of spatio-temporal construction (and so the full scope of possible experience). The proof of this is in the First Analogy. Kant says that time “as permanent form of intuition” is something that “remains and does not change” (B224–25, p. 213). Time as a form of intuition is constructive time, and its “permanence” signifies that constructive time does not begin anew with a present construction. A present construction, that is, must be a continuation of an already onrequired or necessary to proceed. This accounts for Kant’s repeated assertions that the unity of apperception is necessary unity. See B135, p. 154; B135, p. 155; B142, p. 159.

72

t h e t r a n s c e n d e n ta l d e d u c t i o n

going (constructive) time. This means that it must be proper for me to be in the course of a temporal construction, so that what I now can go ahead to construct is a continuation. If the construction is, say, a series of downbeats accompanied by a recitation of numerals in order, then it must be proper for me to now be up to stage k (that is, beyond previous stages) in the construction. Only in this way can past time be represented constructively, viz., as constructions it is now proper for me to be past or beyond. But it cannot just be proper to now be up to stage k in a procedure, since a procedure is precisely a construction that has a certain fixed order. The question then arises of how it can be proper to be in the middle of a procedure when I have not performed the earlier stages.4 As an analogy, consider a cake-baking procedure of putting ingredients in a bowl in a specific order. Suppose I have this procedure and I come into a room where a bowl is present containing the first four ingredients in it. Then it is proper for me to be up to stage 5 in the cakebaking procedure, despite the fact that I haven’t been putting in any ingredients. Something in my present circumstances (the bowl containing the ingredients) is the basis or “substratum” (B225, p. 213) of my representation of my being in the course of the procedure. My procedure, that is, is geared to something (getting a cake baked), and what exists in my present circumstances (the bowl with ingredients in it) is such that, in order to “keep up with” what my procedure is geared to, it is proper to be at a mid-stage of the procedure. Similarly, the temporizing procedure must be geared to something so that what exists in my present circumstances is something I have to keep up with. It cannot be that the construction is geared to objective time (so that to keep up with how far that time has progressed by now I have to be up to stage k in the procedure).5 Therefore, as Kant says, “there must be found in the objects of perception .l.l. the substratum which represents time [constructive time] in general” (B225, p. 213). What must be kept up with, then, is the existence of something real, so that in my present circumstances that existence is so far along ( just as the cake-baking existence is so far along). Now the real for Kant is what 4. Of course, if I have been performing stages I can now be up to stage k. However, I should be able to now represent past time without having actually been constructing and, besides, past time extends back before my birth, so its representation cannot depend on my having been actually constructing. 5. Kant says, “Now time itself [i.e., objective time] cannot be perceived” (B225, p. 213). Time itself cannot be found in my present circumstances because objective time, whether conceived relationally or substantivally, simply doesn’t exist. Recall that there cannot be continuous expanses of objective (extra-constructive) time, and time is nothing without continuous expansiveness.

Kant’s Theory of Cognition

73

affects us,6 and for the real to be so far along in its existence is for its present existence to be a continuation of its existence. Continued existence is continued affection, which is a matter of affection continuing while keeping track. If temporizing, then, is geared to keeping track of, or tracking, what affects, then what presently affects may be so far along in its trackability that, to keep up with it, it is proper to be up to stage k in temporizing. We can represent this as follows: (1) With respect to what presently affects, it is proper to be up to stage k in temporizing-cum-tracking. In (1), what presently affects is represented as being so far along in its existence (trackability). Hence (1) represents the “permanence of the real” (at least as far back as k stages of temporizing) which, for Kant, is the concept of substance (A143, B183, p. 184). Hence substance is the substratum or basis of representing past time (“time in general”). Recall now that space and time are also forms of empirical intuition (that is, perception or sensible reaction). Reactions, that is, are proper in the compass of spatio-temporal constructions. If we augment the temporizing-cum-tracking procedure to one in which reacting “red,” for example, is to be done at the first stage, then (1) becomes (2) With respect to what presently affects, it is proper to be up to stage k in temporizing-cum-tracking from first reacting “red.” In (2) the reaction is proper with respect to what presently affects, only it is so k stages ago. Thus (2) represents the reaction as a determination of substance (a reaction proper in regard to the continuant that presently affects). The possible appearance (i.e., the proper reaction “red”) is thus brought to present cognition (to the unity of a present rule) via a rule for being in the course of a temporizing construction. This rule, in turn, requires cognizing that reaction as a determination of a substance (i.e., permanence of the real). Thus Kant can say that “all that belongs to existence [possible appearances] can be thought only as a determination of substance” (B225, p. 213). Although the purport of cognition, then, is exactly the full propriety of reactions or empirical intuitions, effecting this purport requires that these reactions be thought of as determinations of substances (enduring realities). Kant’s rule theory of cognition, then, pertains to the past, and also pertains to cognition of objects of experience (substances).

6. The real, Kant says, is an object of sensation, where sensation “gives us only the consciousness that the subject is affected” (B207, p. 201). The real then, which “corresponds” to sensation, is just what affects us.

74

t h e t r a n s c e n d e n ta l d e d u c t i o n

2 . a m o d i f i e d v e r s i o n o f k a n t ’ s ac c o u n t Rules, for Kant, are not formulas or linguistic expressions, but cognitive states of the subject. If we are to identify rules with thoughts, then it must be that, whatever the rule is, it doesn’t need to be “interpreted” by the organism to have its sense or content. This is so because one basic aspect of our understanding of thoughts is that they are what interpret or give sense to otherwise lifeless signs or formulas. If so, then a thought is not anything like an internal sign or formula. Now the content of a rule is not “proposition-like,” but rather behavioral. Rules, then, must pertain to their behavioral content without interpretive or cognitive mediation if they themselves are to be the units of cognition. This suggests a causal connection, where a rule becomes a state of the organism apt for causing or producing the behavior, since such a connection doesn’t require interpretation of the state for it to connect to (have) its behavioral content. I shall use the term “mechanism” generically to signify a state of an organism with causal potential for behavioral output. A rule involves not only behavior but some standardization or normativity regarding that behavior. The rule, then, must be a mechanism not only for behaving, but must also be, or go with,7 a mechanism for correcting or assessing behavior, either one’s own or another’s. Mother animals, for example, often have not only certain ways of behaving but in addition ways of correcting offspring who deviate from those ways. In our sense, they have rules for that way of behaving.8 If thoughts are to be identified with rules understood as mechanisms, then having a thought, in the occurrent sense, must be identical to a mechanism for behavior 9 being in place (so as to be operative) in the organism. Since an organism can have a thought of going to the refrigerator even if it is incapacitated motor-wise, the mechanism for behavior needn’t be fully in place. What is required only is that the mechanism be executively in place (or, equivalently, that an executive mechanism be in place). When an executive mechanism is in place, the organism’s knowing what is to be done is operative (occurrent) in him. 7. Any fine-grained individuation of mechanisms would require a theory of how the brain and body work. At the level at which we are characterizing internal mechanisms in terms of their causal potential for outputs, it is indeterminate, for example, whether there is a single mechanism for both behaving and assessing behavior, or whether there are two mechanisms—one for behaving and another for assessing. 8. Our notion of a rule does not involve its being a long-term standard. A rule can be temporarily adopted in certain circumstances, and then as quickly abandoned or altered. 9. For convenience, I shall simply refer to rules as mechanisms for behavior, rather than mechanisms for behavior that are also mechanisms for correcting or assessing behavior.

Kant’s Theory of Cognition

75

Even if I am incapacitated motor-wise, I might try to get you to go to the refrigerator to get an apple, in accordance with my knowledge of what to do being then operative in me. The identification of thoughts with rules or executive mechanisms for behavior being in place has some intuitive plausibility. A lion, upon detecting prey, may think to chase it. This thinking, given other internal factors involving wants and beliefs, leads to the lion’s being prepared or set to chase the prey. If it is to be thus prepared or set to behave, there must be some mechanism in place for the performance of the behavior. It doesn’t seem as if it is the desire or the belief that pertains to the behavior. Desire pertains to cognizance of its hunger, and belief pertains to detection of the prey and the environment. Hence, it is plausible to locate the executive source of the behavior (knowing what to do) in the thinking. Furthermore, how we are prepared to behave is something, at least intuitively, that we can be conscious of. The cat that is prepared to pounce on a mouse can know or discern (be aware of) what it is prepared to do.10 If so, then on an intuitive level, this model of having a thought as an executive mechanism being in place coheres not only with the psychological (computational) nature of thoughts but with the nature of thoughts as being conscious. Since the model of thoughts as executive mechanisms in place seems intuitively plausible at a primitive level,11 then some supposedly insuperable obstacles to extending this account to the sophisticated content of human thought must preclude it from being seriously considered. One such obstacle, surely, is that thoughts can have content pertaining to the past, but a mechanism with a causal output would seem only to pertain to behavior one can now go ahead and perform. It is here that Kant’s conception of substance enabling rules for being in the course of temporizing behavior plays a key role. In what follows I shall try to show that thoughts having the formal reality of executive mechanisms in place can have content sophisticated enough to encompass a world of objects extending through the full scope of space and time. 10. Suppose I open the door for the dog to go out and that the dog doesn’t want to go out. Still, the dog looks at the door and at me, indicating thereby that it knows what I want it to do, and hence that it is aware of what is to be done. Similarly, there seems to be such awareness when the dog is hurting and cannot get up on its legs. Thus it is aware or conscious of what is to be done even when the conative and motor components for being fully prepared to do it are not in place. In our terms, it is conscious exactly of a mechanism that is (only) executively in place. 11. Both descriptive and causal theories, I believe, are modeled more on perception than on (mechanisms for) action, where in perception our state by its internal features depicts what is around us (the model for descriptive thought as depicting what is not around us), or our state is normally caused by certain proximate external phenomena (the model for causal theories of thought pertaining to distal phenomena).

76

t h e t r a n s c e n d e n ta l d e d u c t i o n

We can characterize a Kant-like theory as having two basic kinds of singular cognitions.12 The first kind pertains to spatially shifting attention in a motion of the subject, and thereupon being affected and reacting, which we can write out as follows: (3) It is proper to take j steps, to be affected, and to react q.13 We call (3) a “future-oriented” cognition. The second kind pertains to being in the course of temporizing. This is just (2) above; i.e., (4) Upon being affected, it is proper to be up to k in temporizing-cum-tracking from first reacting r. We call (4) a “past-oriented” cognition. In our account of thoughts or rules, both (3) and (4) have to be understood as mechanisms executively in place. So, for example, the future-oriented singular thought signified by (3) is just a mechanism in place for taking j steps, being affected, and reacting q. We first consider the nature of this mechanism, and then we shall consider the nature of the mechanism pertaining to (4). In our account, the content of a thought can be identified with what the mechanism (which is the thought) is a mechanism for. Since a mechanism is just an internal or intra-organismic cause, the content is what the cause is a cause for. For example, pushing the pendulum before me is a cause for the pendulum to swing periodically, and this is true whether or not the pendulum is actually pushed. Similarly, if my mechanism is for taking j steps and being affected, that output is not, per se, an actual reality or occurrence, but the causal potential (or the potential output) of the mechanism or thought (an output that pertains to it whether or not that mechanism actually unfolds). Hence, on this view the content of a thought is neither a Fregean abstract reality nor an internal or external concrete reality, but a potential reality (a potential output). The causal potential of a mechanism is what the mechanism uniquely unfolds to, if it actually unfolds. If there is to be a mechanism pertaining to (3), it must involve a mechanism for being externally affected. Hence the unique unfolding of this (component of the) mechanism must involve a real external presence. The existence of hal12. By a “singular” cognition I mean, roughly, a cognition that purports to be of a concrete state of affairs (how it is with reality at a certain place at a certain time). See further part 4 below. 13. From the Third Analogy and the First Antinomy it is clear that Kant means his theory of spatial construction (synthesis)-cum-reaction to be more than “local.” I am representing these constructions or syntheses as bodily, involving moving off or taking steps; my attention shifting in a flow as I do.

Kant’s Theory of Cognition

77

lucinations and illusions seems to contradict this, since if, upon taking j steps, I am taken in by a hallucination (seeming to be affected) then the mechanism originally in place continues to unfold to reacting q, and hence the unfolding doesn’t uniquely involve a real external presence. This calls into question, then, how thoughts can have external content. What being affected signifies, I suggest, is gross mechanical interaction, involving an attempt at making my body mechanically interact in some holistic way with an external material body. Pushing, pulling, lifting, and so on can all be cases of such gross mechanical interaction.14 I suppose further that there are feedback components to the mechanism for detecting mechanical resistance upon the attempt to lift, push, etc.15 The part of the mechanism that is for being affected, then, is a mechanism for attempting a gross mechanical interaction and completing it only with a detection of gross mechanical resistance. I now want to argue that hallucinations and illusions do not contradict there being a mechanism whose unique unfolding involves external proximate reality. We can consider a hallucination to be an internally generated simulation. In a gross mechanical interaction, the distribution of stresses and strains throughout our bodies is so pervasive that it seems that any internal generation simulating this bodily state would have to have such further gross effects that the feedback mechanism would easily detect them, and so the mechanism would not continue to unfold. The only way to bring off the simulation so as not to be detectable would be to tamper with a part of the feedback component itself (desensitizing nonperipheral brain receptors so that the brain is not well connected to the further large-scale bodily disturbances). But now the hallucination involves changing or tampering with the mechanism itself, and from the fact that a second or altered mechanism can unfold without external proximate reality it does not follow that the original mechanism unfolds with anything other than external reality. Empirical evidence for this account of gross mechanical hallucinations would be that such hallucinations take place only in “altered states” (drugged states, sleep, 14. In other words, existence or reality (what affects) is basically mechanical existence. Kant does not explicitly incorporate this conception of existence in the Critique (although see B278, p. 246), but he does so in the Metaphysical Foundations of Natural Science. He understood mechanical existence to be, at least in part, an empirical concept and so outside the scope of the Critique. In my elaboration of Kant’s account I am freely employing empirical considerations, since my motive is to understand cognition, not to use that understanding to do metaphysics. 15. Feedback is part of executive mechanism (or mechanisms executively in place) to the extent that for knowing-what-is-to-be-done to be in place, knowing how to tell if it is being done (i.e., the feedback) must be in place, especially if the mechanism is also to function as a rule for correction and assessment.

78

t h e t r a n s c e n d e n ta l d e d u c t i o n

insanity, etc.). To the extent that these states involve holistic brain-tobody shutdowns, they are states in which the feedback component of the mechanism for being affected is tampered with (is not in place). An illusion, now, is an externally generated simulation. Since there is no such thing as external “artificial” cohesive mass (material existence), an illusion would seem to require some direct external stimulation of the brain. This, in turn, would require such elaborate hookups that they would be easily mechanically detectable by the organism. As long as the mechanism for being affected has a feedback component that includes detecting normalcy of one’s body (no electrodes being implanted, etc.), such external stimulation of the brain will not generate illusion, without also tampering with the feedback component (to block detection of the electrodes, etc.). Once again, the fact that an altered or different mechanism would then unfold without proximate external reality does not contradict the unique unfolding of the (unaltered) mechanism originally in place.16 Mechanisms for being affected, then, can have as part of their content external proximate material reality. In this sense, thoughts (and consciousness via the detection component) can have “direct” external content.17 The reacting component of the mechanism pertaining to (3) is as various as there are different sorts of reactions or detections. We simply note that it is meant to include reacting (detecting) “reflects electromagnetic waves of frequency k” as well as reacting “red.” Reactions, that is, can be with instruments, in terms of theories, and so on, as well as by unaided sensation. These latter reactions are more objective since they are reactions pertaining to how objects act on all other objects, not just how they act on our sense organs. This is true despite the fact that we detect these actions on other objects (say, on light meters) by how these other objects act on our sense organs.18 The mechanisms pertaining to the past-oriented cognition in (4) above will have to include a mechanism for being beyond a temporizing procedure. Suppose now that there is a mechanism in place for the 16. Similar considerations pertain to the spatial or step-taking component of the mechanism. 17. This does not imply that the thought of being affected involves there actually being material existence (which would make it a Russellian thought, or a thought that exists only with the existence of its object). The reason is that having the thought is only a matter of the mechanism’s being in place, not its unfolding. 18. We note that Kant similarly does not restrict the reactive component (appearance) to sheer sensory reactions, but allows, for example, that magnetism, though not directly detected, is a phenomenal feature (A226, B273, p. 246). Although Kant is an “interactionist,” in that cognition consists of rules for our commerce with the world, he is no “sensationalist.”

Kant’s Theory of Cognition

79

temporizing procedure itself (which, on its own, unfolds to a series of downbeats at pace while reciting numerals). We presume that the unfolding of this mechanism involves a series of internal states or internal actions by which the output takes place. Suppose now that there is a second mechanism in place that works to quickly inhibit each internal action of the first mechanism and then quickly set the first mechanism ahead to its next action. The second mechanism then rushes through the internal actions of the first mechanism. Thus as the embedded (first) mechanism is set to yield a slow or paced downbeat accompanied by a recitation of the numeral 1, the overriding or diverting (second) mechanism inhibits that set from yielding its output and resets the embedded mechanism to the next internal state, viz., the state that, left to itself, yields a slow downbeat accompanied by a recitation of 2, etc.19 The total mechanism in place then (that is, the embedded mechanism together with the second mechanism that diverts or overrides the actions of the embedded mechanism) can be characterized as a mechanism for quickly getting beyond the output of the embedded mechanism. Equivalently, it is a mechanism for presently getting beyond temporizing at a slow pace. But now a mechanism for presently getting beyond temporizing is also a mechanism for presently being beyond temporizing ( just as a mechanism for going to the refrigerator is also a mechanism for winding up or being at the refrigerator). Suppose now that this complex mechanism unfolds according to keeping track of how one is presently affected. Thus one is first set by the embedded mechanism to a first-slow-downbeat-while-keeping-track, but one is quickly diverted to then being set to a second-slow-downbeat-whilekeeping-track, but one is diverted, etc. Then the mechanism as a whole that is in place is a mechanism for presently being beyond temporizingcum-tracking in regard to what presently affects. If we go back to our cake-baking case, we can suppose the organism walks into the room with a mechanism in place for baking a cake (by putting ingredients in a bowl in a fixed order from the beginning). The organism then walks into the room with the mechanism set to its first internal action or state for putting in the first ingredient. The organism 19. One can devise “models” of such a complex mechanism. Thus the embedded mechanism can be modeled as an internal clock that traverses its face readings at a certain slow or tempered pace. When the clock is set at “1” it yields the downbeat gesture (which is tempered by the clock’s motion) as well as a recitation of “1.” One’s performance, then, is geared to (indeed produced by) the unfolding of the internal clock. The diverting mechanism can then be modeled as a second component that quickly takes the hand of the clock just past the position of the face at “1” and moves it quickly off the face to the position at “2,” etc., where the second slow action of the embedded mechanism is now in place.

80

t h e t r a n s c e n d e n ta l d e d u c t i o n

presently detects that the first four ingredients are in the bowl. In order to “exaggerate” what next goes on, we can say that the organism, upon detecting those ingredients, actually initiates putting in each ingredient from the first, but quickly inhibits each behavior and goes on to the next; viz., the organism quickly goes through some of the motions of putting in each ingredient until he reaches the fifth ingredient. Notice then that upon detecting the four ingredients, the organism doesn’t simply bypass or replace the mechanism originally in place for baking a cake from the beginning, but rather, as shown by his going through the motions, the original mechanism remains in place, only its actions are overridden or diverted or gone through. Whatever it is inside him by which those actions are overridden is the overriding mechanism in him. Altogether, upon detecting the four ingredients, there is a mechanism in place in the organism for presently (getting and) being beyond the first four stages of the cake baking procedure. The cognition or thought of being beyond k in temporizing-cumtracking with regard to what presently affects is a mechanism executively in place for getting past the first k actions of a mechanism executively in place for temporizing-cum-tracking what presently affects. The entire complex mechanism is an internal state of the organism in regard to what presently affects him. I want to contend now that it is only in regard to a substance that one can be in this internal state. Equivalently, for one to be in this state in regard to what presently affects is eo ipso to think of what presently affects as a substance. If so, then Quine would be mistaken when he says that representing substances is inscrutable because there is no fact of the matter,20 or because there is nothing in the head, independent of interpretation in a metalanguage, which is the representation of substance. A substance is something that exists entirely or completely at each moment of its existence. Because of this, its career or history may unfold in time, but it itself does not. Its career may be a temporal whole that has temporal parts, but it itself is not a temporal whole or a temporal series. Intuitively, now, to have the thought, with regard to what presently affects or what presently exists, of being beyond a certain temporizing stretch of tracking it, is to think of it (not some previous temporal part of it) as something which has existed. What we need to show is that this intuition pans out when the thought is explicitly identified with our mechanism for being beyond temporizing-cum-tracking. If our mechanism pertained to just a present stage or time-slice,21 20. W. V. Quine, “Ontological Relativity,” in Ontological Relativity and Other Essays (New York: Columbia University Press, 1969), 47. 21. See W. V. Quine, Word and Object (Cambridge: MIT Press, 1960), sec. 12.

Kant’s Theory of Cognition

81

then actions for slow tracking (tracking through a series of paced downbeats) would have to be in place with regard to that slice. The reason is that the mechanism as a whole is for getting beyond (diverting or inhibiting) those actions, and only if those actions can arise and be in place can they be inhibited. But there can be no slow-paced tracking of a present slice that reaches, say, stage 3 of the series of downbeats, since the slice, we may assume, doesn’t last as long as three paced downbeats. Hence, there can be no action for that behavior and so our mechanism cannot pertain to just a present slice. This leaves it that perhaps our mechanism pertains to a series of slices or stages that begins with what presently affects (each member of the series lasting, say, for one paced downbeat). Now, the actions for stages of the paced tracking can indeed arise. However, they cannot be quickly or presently diverted. The reason is that the actions have to await the existence of the “correct” slice. Since tracking at a third slow downbeat can only pertain to the third slice, there is no way that an action for such tracking can be in place until the third slice exists. Thus if our mechanism quickly diverts actions for slow tracking, it cannot pertain to what presently affects as the first of a series of slices.22 Finally, our mechanism cannot pertain to what presently affects as the last member of a series of stages or slices. Here, once again, the actions of the embedded mechanism cannot possibly arise, since initiating the tracking with a first slow downbeat would have to pertain to the initial slice, but I cannot now track the initial slice, since it no longer exists. Thus actions for initial stages of slow tracking cannot possibly presently arise, and so cannot be diverted or gotten beyond. It is only if what affects is what the whole procedure of slow-paced tracking pertains to, that the actions can all presently arise and be presently diverted. But this simply means that although what presently affects is subject to tracking with a procedure that marks time, it is not subject to it by its having different stages of existence through time; i.e., its existence is whole or complete in the present, and so it is an enduring substance.23 Thus, a mechanism in place, upon 22. Suppose the mechanism of footnote 19 above is geared to (pertains to) a slowly unfolding external phenomenon ( just as tempered tracking of a series of slices is geared to the slowly unfolding series of slices). Then action at “2” (for producing a second slow downbeat) is geared to stage 2 of the external phenomenon. This action cannot be diverted or gotten beyond until it can arise, but the action (for “covering” stage 2 of the external phenomenon) cannot arise until stage 2 of the external phenomenon does. The second diverting component of the mechanism can still quickly lift the hand off the face of the clock and put it at “2” and then at “3,” but this is no longer characterizable as getting beyond action geared to the slow unfolding phenomenon. 23. Note that our account agrees with Quine that an actual episode of keeping track does not distinguish a substance from a series of slices. There is nothing in the actual episode that settles the issue of whether one is tracking one thing or a series of several.

82

t h e t r a n s c e n d e n ta l d e d u c t i o n

being presently affected, for being beyond stages of tracking-cum-temporizing is a thought of what presently affects as being a substance. This mechanism, then, is what there is in the head (or the fact of the matter in the head) as far as representing substance is concerned.24

3 . a n e x t e n s i o n o f k a n t ’ s ac c o u n t I now want to show how the two basic forms of singular cognition in (3) and (4) above can be used to extend the account of cognition to the full scope of space and time. First, we can “nest” these singular cognitions or embed them within one another. So, for example, we can embed a (4)-like or past-oriented cognition within a (3)-like or futureoriented cognition to get: (5) It is proper to take j steps and be affected, and upon being affected it is proper to be up to k in temporizing-cum-tracking from first reacting r. Here the rule is first to spatially shift and only then (upon being affected) is it proper to be in the course of temporizing, etc. Whereas (4) effectively represents that a reaction was proper with respect to the substance before me now, (5) effectively represents that a reaction was proper with respect to a substance j steps from me. Similarly, we can embed a (3)-like or future-oriented cognition within a (4)-like or pastoriented cognition to get: (6) Upon being affected, it is proper to be up to k in temporizing-cum-tracking from its first being proper to take j steps and be affected and react q. Here the rule is to be in the course of a procedure that begins with the propriety of taking j steps, etc. Whereas (3) represents that a reaction is proper upon now taking j steps and being affected, (6) represents that a reaction was proper upon k units ago taking j steps and being affected. Although I shall not go into the details, the nesting of basic cognitions or rules carries over when these rules are understood as executive mechanisms in place. Ultimately, such nesting of mechanisms is a matter of “recombining” or “substituting” into components of mechanisms (in a way that is not fundamentally different from the case of one who Our argument turns, rather, on a present mechanism for (quickly) being beyond (slow, extended) tracking. 24. The “scrutability” of the mechanism, or evidence that the mechanism is in place, would then imply that it is determinable whether an organism is representing a substance. The “exaggeration” we described in the cake-baking case above is an outline of how there can be evidence of a mechanism in place for being beyond stages of a procedure.

Kant’s Theory of Cognition

83

knows how to frost a chocolate cake and put sprinkles on a vanilla cake then knowing how to put sprinkles on a chocolate cake). We can allow further “substitutions” in either (5) or (6). For example, for “first reacting r” in (5) we can substitute the propriety of taking m steps being affected and reacting s, and for “react q” in (6) we can substitute that upon being affected it is proper to be up to stage n in temporizingcum-tracking from first reacting t. Theoretically this gives all arbitrary finite nestings of the two basic forms of cognition as also singular cognitions. In order to get cognitions covering the full scope of space and time, we have to augment step taking to include moving off the earth and moving with vehicles. Although we do take steps “up,” as when we move up hills or climb ladders, still, once we incorporate moving off on vehicles, we shall have incorporated moving off the earth by moving with rockets, etc. Consider first the case of moving with or on a horse. I can have the thought of getting on a horse, propelling it to a certain speed, and going at that speed for a certain time (through a certain series of downbeats), and only then being affected and reacting “red.” To have this thought, recall, is just for one’s knowing what is to be done to be in place. As long as I know how to detect the horse’s reaching that speed (say, by characterizing the speed as one that is reached by so much thrusting of the horse by me), I know what is to be done to carry out this procedure. Note that this sort of rule extends the scope of cognition to proper reactions that may not be cognizable by unaided step taking if the speed is great enough. I contend that this thought or mechanism can be in place even if there are no horses around, just as a rule for cake baking can be in place even if there are no ingredients around (although, of course, the mechanism in place cannot be realized, or properly unfold, without a horse or without ingredients).25 We can add, thus, such vehicle-augmented versions of basic cognitions into our arbitrary finite nestings of cognitions. This allows that we can have thoughts of moving with geodesic vehicles (rockets in free fall, for example) for a while (that is, through a “proper” time marked by our own downbeat gestures), and only then being affected and reacting.26 It allows, further, the thought of hopping onto one geodesic vehicle g1, and then transferring to a second one g2 at a stage k of g2’s geodesic motion. There can be many such g2’s, one for each different geodesic mo25. People can rehearse or go through the motions of cake baking without equipment present, and without representing any particular eggs or cups of sugar. Roughly, “egg beating” is a kind of activity. 26. We can have such thoughts as long as we can tell we are on a vehicle moving geodesically.

84

t h e t r a n s c e n d e n ta l d e d u c t i o n

tion that coincides with g1 at the hopping-off time. This repertoire of thoughts (one for each different g2) together constitute a geodesic polar coordinate system at the hopping-off point. Further nestings give a repertoire of thoughts constituting geodesic polar coordinate systems at space-time points along any point of the original polar coordinates. Allowing arbitrary finite nestings we get, finally, a repertoire of thoughts that, collectively, are adequate for representing any spacetime event e, reachable from where we are by a finite chain of fourdimensional geodesic coordinate systems. This, I claim, gives full relativistic scope to our singular thoughts.27 In this manner we have extended Kant’s theory of cognition to encompass relativistic space-time. Kant, of course, had a conception of the topology of space-time as being flat, and because of his concern for the a priori empirical distinction he would not have included the motion of vehicles. It seems to me, however, that including this element, even if it is empirical, still leaves the account of space-time as being a constructivist account (pertaining to motions of the subject with vehicles) as opposed to a static relational or substantival account ( just as holding numbers exist in procedures of counting with or without an abacus would still be a constructivist account). Further, this constructivist account, encompassing as it does all geodesic motion, encompasses everything about relativistic spatio-temporal position that is invariant. Indeed, on this view, the full possibility of geodesic motion is the space-time manifold, with coincidence relations among geodesics constituting the geometry (local curvature) of the manifold.28 We turn now to the issue of the truth of Kantian cognitions. Kant himself accepts a “nominal” notion of truth as “agreement of knowledge with its object” (A58, B82, p. 97), but he gives no account of what truth is, nor does he show that truth (the truth predicate) is recursive with respect to the structure of his cognitions. I shall try to show how our modified version of Kant’s theory of cognition leads to a “real” definition of truth, and establishes it as recursively attributable. We can motivate our definition of truth by considering the thought an organism has of going to the refrigerator (taking certain steps to the refrigerator) to get an apple. The organism that has this thought is successful if the behavior pans out. The notion of truth should be connected to success 27. If the four-dimensional manifold from where I am to e has everywhere finite curvature and its tangent spaces can be embedded in a (higher-dimensional) flat manifold, then arbitrary finite nestings of geodesic polar coordinates will encompass e. 28. This extends to a relativistic context. See Michael Friedman’s contention that Kant views inertial motion “as definitive or constitutive of the spatio-temporal framework of Newtonian theory” (Friedman, Kant and the Exact Sciences, 143).

Kant’s Theory of Cognition

85

in order to secure its status as something positive.29 However, success presupposes purpose or desire, since if one has no desire for an apple it is hardly success that the behavior pans out. Truth, on the other hand, should pertain to thoughts independent of desires. It is the panning out of the behavior itself that directly and specifically relates to the content of the thought. Now, for us, thoughts can be mechanisms in place for behavior, but they can also be mechanisms in place for being beyond behavior, and the notion of behavior panning out does not apply to the latter. What we need is a generic notion that includes behavior panning out when the mechanism is for behavior, but also pertains to (more complex) mechanisms for being beyond behavior. This notion, I contend, is the realizability of a mechanism in place. A mechanism is realizable if conditions are such that it can fully unfold, or if it can realize its potential (realize what it is a mechanism for). However, it is not the full realizability, but what we may call the “active realizability,” that defines truth. Suppose there is some obstacle between the organism and the refrigerator that blocks or inhibits the actions of his mechanism from unfolding. This does not detract from the “correctness” or “truth” of his thought. His mechanism is only unrealizable because of what we may call a lack of “conductance” conditions. Indeed, the obstacle is analogous to a wire with high electrical resistance blocking a mechanism for electrical current through the wire from being realizable. On the other hand the “lack” of a refrigerator certainly detracts from the correctness of his thought. But now the absence of a refrigerator is the absence of an active condition (not a conductance condition), since a refrigerator, upon taking certain steps, is (together with the detective component of the mechanism) a source of action for the mechanism to continue to unfold (to continue to the point of opening it and reaching for an apple). This suggests that we equate the truth of a thought with the active realizability of a mechanism in place. Equivalently, a thought is true if all the active conditions for the mechanism in place to unfold obtain. Of course the mechanism itself’s being and remaining in place is an active condition of its unfolding, but truth pertains to active conditions other than the mechanism itself’s being in place (viz., a truth condition of a thought isn’t the existence of the thought in the organism). We get then as our analysis or real definition that the truth of a thought is the active realizability (other than the mechanism itself’s being in place) of a mechanism in place. We apply this definition now to the basic singular cognitions. 29. As Dummett puts it, truth is like winning. See Michael Dummett, “Truth,” in Truth and Other Enigmas (Cambridge: Harvard University Press, 1978), 1–24.

86

t h e t r a n s c e n d e n ta l d e d u c t i o n

In relation to the future-oriented cognition in (3) above, the lack of footing or the existence of obstacles in a path constitute a failure in conductance conditions for taking j steps. The only active condition (that is, source of actions during the unfolding) for taking j steps is the mechanism itself’s being and remaining in place. Active realizability commutes with the step-taking component. As an analogy, a mechanism, such as a missile launcher with a missile, for reaching a target and thereupon exploding is actively realizable if and only if upon hitting the target the mechanism for exploding is actively realizable. Similarly, then, the mechanism for taking-j-steps-being-affected-and-reacting-q is actively realizable if and only if upon taking j steps the mechanism for being affected and reacting q is actively realizable. The action of the mechanism for being affected, recall, involves detecting resistance as one pushes, pulls, lifts, or otherwise attempts a gross mechanical interaction. This detection, indeed, is an internal source of action for completing the attempt. But we saw above that the unique source of the detective action is real external proximate material existence. Such existence is the only active condition (other than the mechanism’s being in place) of the mechanism continuing to unfold. Allowing now that the active condition for reacting q is the suitability of the material existent for that reaction,30 we get, all together, the following: (7) The mechanism for taking j steps being affected and reacting q is actively realizable if and only if upon taking j steps there is material existence suitable for reacting q. In relation now to the past-oriented cognition in (4) above, recall that the cognition consists of an embedded mechanism plus a diverting mechanism (for getting beyond the actions of the embedded mechanism). To see how active realizability relates to diversion, suppose there is a mechanism in an object for externally discharging a certain electrical current in a direction d, and also inside the object a diverting mechanism, such as a magnetic field, that internally deflects the internal current, so that the actual discharge is in direction d*. All together, then, this is a mechanism not just for externally discharging in d* but for doing so by getting beyond (actions for) externally discharging in direction d. The whole mechanism is actively realizable if the diverting component is, which requires that, but for the diversion (other than for the diversion), the embedded mechanism for discharging in d is actively re30. The mechanism of a heat-seeking missile, we can suppose, does a final check for a minimal temperature in an object it has reached before exploding. Then the active realizability of that component of the mechanism is the suitability of the object for detection of that minimal temperature.

Kant’s Theory of Cognition

87

alizable. Without the latter an active condition for diverting (namely, something to divert) is lacking. Applying this now to our past-oriented cognition (where getting beyond k or being beyond k is the diverting component), we get that the mechanism in place (upon being affected) for being beyond k in temporizing-cum-tracking-from-first-reacting-r is actively realizable if and only if upon being affected, but for being beyond k, the mechanism for temporizing-cum-tracking-from-first-reacting-r is actively realizable. Now, temporizing, like taking steps, has no active conditions beyond the mechanism’s being in place, and time, like space, commutes with active realizability.31 The active condition for tracking is that the material existent be suitable for tracking (viz., has duration), so that all together we get the following: (8) The mechanism for being affected and, upon being affected, being beyond k in temporizing-cum-tracking from first reacting r is actively realizable if and only if there is present material existence which, other than for being beyond k, is suitable for temporizing-cum-tracking from first reacting r. In other words, the truth of the thought is not that what affects is suitable for the (embedded) procedure, but that it is suitable for it but for being beyond k (roughly, the procedure “would be” suitable except that our condition or situation is being beyond it). In (7) and (8) I believe the notion of active realizability gives the intuitively correct truth conditions for future-oriented and past-oriented thoughts, by commuting with the spatial, temporal, and diversionary elements of these thoughts. Further, active realizability commutes with the nestings of cognitions in (5) and (6) above, simply according to the principle that total mechanisms that have mechanisms nested (attached at places) within main mechanisms are actively realizable if and only if the main mechanism is actively realizable according to the nested mechanism’s being actively realizable at its place of attachment. Truth, then, understood as active realizability, commutes with the structure that determines the basic singular cognitions and the full range of complex singular cognitions.32 Finally, in regard to our “vehicle-assisted” cognitions, a mechanism 31. A mechanism of a bomb for exploding after twelve ticks of its internal clock is actively realizable if and only if after twelve ticks the mechanism for exploding is actively realizable. 32. The import of this, for us, in line with Tarski but opposed to Davidson, is that we can therefore have a recursive definition of the truth predicate, not that we can therefore ascribe content to sentences or thoughts. The content of our thoughts is already fixed by the potential of the mechanism in place, whether it is actively realizable or not, and so independent of the truth predicate. See Donald Davidson, “Truth and Meaning,” in Inquiries into Truth and Interpretation (Oxford: Clarendon Press, 1984), 17–36.

88

t h e t r a n s c e n d e n ta l d e d u c t i o n

for, say, getting on and moving with a horse for a while and then being affected and reacting q is only actively realizable with the proximate existence of a suitable horse. In this case, however, we can still say that the mechanism is actively realizable but for the existence of a horse, if there is material existence suitable for reacting q at the end of such and such a motion on a horse. Thus we can say that the thought is true but for the existence of a horse. But to say this is effectively equivalent to saying that if there had been a horse present, the mechanism in place would have been actively realizable (or if there had been a horse present, the thought of being affected and reacting q at the end of such and such a ride would have been true). Active realizability but for a condition, then, is effectively (an analysis of) counterfactual truth. Similar remarks pertain to moving on geodesic rockets. These thoughts may not be true, but they nevertheless express that certain counterfactuals (e.g., if there has been a projectile moving geodesically .l.l.) are true.33

4 . d e s c r i p t i v e a n d c au s a l t h e o r i e s of cognition and the virtues of k a n t ’ s ac c o u n t Kant abandoned purely conceptual or descriptive representation after the Inaugural Dissertation. By the time he abandoned it, he had already come to hold that space and time were forms of intuition themselves given in pure intuition, rather than conceptualizable objective aspects of reality (whether construed as relational features of things or as the independent locus of things). Thus his contention that purely conceptual representation is empty, or without connection to an object, would have applied to descriptive representation without a spatiotemporal component. However, later on he explicitly rejects conceptual representation as empty even when it involves such a component. He says regarding spatio-temporal predicates that they always belong as well to merely possible things.34 Even a full individuating description with spatio-temporal components, then, doesn’t pick out or pertain to an actual object as opposed to a purely possible one. There is nothing 33. That such counterfactually true (as opposed to flat-out true) thoughts are required for cognizing the full scope of space-time is no different from a relationist having to include possible relations (not just actual ones). Indeed, in the extended version of the Kantian account, space-time is the full scope of “possible” motions (i.e., possible positionings, including vehicle-assisted positionings or motions, whether there are vehicles or not). 34. See “The Only Possible Proof for a Demonstration of God’s Existence,” in Kants Gesammelten Schriften, Prussian Academy Edition (Berlin: Walter de Gruyter, 1929), 2:76.

Kant’s Theory of Cognition

89

within the description by which it is a representation of what is actual, as opposed to what is possible or what is fictional. Equivalently, a definite description (the x such that .l.l. x .l.l.) selects out of a range of objects the one that uniquely satisfies the description. But unless a range of objects is antecedently specifiable as, say, actual objects, the definite description no more pertains to (one of the) actual objects than to (one of the) possible objects. But then one would have to be able to represent or fix the range of actual objects before definite descriptions could refer to one of them. For this reason, definite descriptions can at best be a “derived” mode of representation, functional only when we are already able to cognize the realm of actual objects. These remarks pertain to definite descriptions per se, whatever their components. In particular they pertain to descriptions including concepts of spatiotemporal location, even when these concepts are relational concepts employing demonstratives. Thus Strawson’s35 preferred mode of uniquely individuating descriptions are of the form “the x such that x is so-and-so and x is R to this (here, now),” where R is some spatio-temporal relation. The employment of a demonstrative does not override the dependency of a definite description on a somehow given range of candidates for satisfying that description. Indeed, fictional and possible objects are characterizable as living once upon a time long ago. In Kant’s view, as in Strawson’s, spatio-temporal determination figures in essentially in cognizing concrete reality. However, for Kant it figures in not as a component of conceptual depiction but as a construction that situates me with respect to being affected and reacting. Thus the rule that it is proper to take j steps and be affected and react q situates me as a construction away from thus being affected and reacting. Neither fictional nor possible reality can be j steps away from my being affected and reacting. Nor does this rule presuppose first having to cognize the range of actual objects before the rule can function. In sum, Kant’s view has the key virtue of a view such as Strawson’s that cognition proceeds in a systematic and always available way by spatio-temporal determination, without the defect of this determination being tied to descriptive depiction that precludes it from functioning as a primary way of cognizing the world. Causal theories or information-based theories of cognition avoid the problem of indeterminacy as between actual vs. merely possible or fictional entities, since it cannot be that there is a causal chain or an informational chain from a non-actual object to me. These theories, however, lose the systematic power of representing according to spatio35. See Peter Strawson, Individuals (London: Methuen 1959).

90

t h e t r a n s c e n d e n ta l d e d u c t i o n

temporal determination. An object’s having a spatio-temporal location does not imply that it connects to me by a causal chain. Because of this, these theories suffer from a lack of full scope. Suppose intuitively, for example, that I now think (have the thought for no good reason) that a red and round object was right in front of the Parthenon two thousand years ago. My thought is purportedly of a concrete state of affairs in the sense that if true it pertains to a material existent being so-and-so at a place at a time. Kant’s view, by its very structure, accommodates this thought whether the thought is true or false, but the causal view accommodates it only on the chance that it is true (so far as its object component is concerned) and the object initiated a causal chain ending with me. Kant, in talking about Humean association as a basis or principle of cognition, makes a similar point when he says, “It would be entirely accidental that appearances should fit into a connected whole of human knowledge .l.l. [for] should they [appearances] not be associable, there might exist a multitude of perceptions .l.l. in a state of separation, and without belonging to a consciousness of myself ” (A121–22, p. 145). Hume’s account makes having an idea of Julius Caesar a matter of there being a train of association in my mind that goes back from something sensed or recalled to Caesar.36 Although this isn’t a causal theory, since the train of association is in the mind, Kant’s point against it applies equally to causal theories. One may object that a thought of a material existent being red and round is one that doesn’t involve a singular reference to an object (however such singular reference is to be understood, say, as an identifying reference enabling us to recognize the object, to relocate it through its history, etc.). Thus the lack of a causal chain is no objection to the causal theory as a theory of such singular reference. The conclusion to draw, however, is that therefore singular reference is not the key notion for cognizing what there is, or for cognizing how it is with material reality dispersed through space and time. Causal theories also fail Kant’s fundamental premise of his entire theory of cognition; namely, that “it must be possible for the ‘I think’ to accompany all my representations, for otherwise something would be represented in me which could not be thought at all, and that is equivalent to saying that the representation would be impossible, or at least would be nothing to me” (B131–32, pp. 152–53). To paraphrase Kant, it is the causal chain outside my cognizance that represents something (the source object) in me, which is not something thought by me.37 In 36. See David Hume, A Treatise on Human Nature, ed. L. A. Selby-Bigge (Oxford: Oxford University Press, 1888), book 1, part 3, sec. 2, p. 82. 37. Combining identificatory descriptive representation with causal or informational chains overcomes neither of the objections to causal theories. See Michael Devitt, Realism

Kant’s Theory of Cognition

91

Kant’s view, on the other hand, it is my real situatedness with respect to the object (that I am separated from it by such and such proprieties of construction or being beyond construction) that is the “tie” to the object, and this situatedness is part of the content of my thought. We can use the difficulties with descriptive and causal theories to motivate the Kantian account. Descriptive theories (lacking a range of potential satisfiers) essentially relate what is in the head to what is outside the head by some sort of isomorphism relation, without any real inputs or outputs.38 Causal theories relate what is in the head to what is outside by an input to the head from the object (via a causal chain). If both accounts are defective, that leaves only the alternative that the relation is an output from the head pertaining to the object. The content of thoughts, that is, must be their output. Since it is precisely rules for behavior, or mechanisms for behavior, that are in the head and have output, thoughts must be such rules or mechanisms. Given, further, that only spatio-temporal determination can ensure the full scope of thoughts, it follows that the rules or mechanisms for behavior or output must include spatio-temporal behavior or output (or, in Kant’s terms, must be rules for spatio-temporal construction given in pure intuition). Altogether, then, we have a purely cognitive argument (one that is independent of Kant’s argument from continuity) that only if space and time are cognized as constructions, and thoughts are rules for interacting with reality (being affected and reacting) in the course of spatiotemporal construction, can there be cognition of the world.39 This argument for constructivism is by no means an argument for verificationism. Although the content of thoughts must pertain to output, that output needn’t be verificatory behavior. Once again, it is Kant’s argument for substance as the substratum or basis of representing past time that is crucial. The rules (or mechanisms) for being beand Truth (Princeton: Princeton University Press, 1984). Clearly, added descriptions do not increase the scope, and although the range of candidates is now fixed to those objects connected to me by causal chains, this range is still something the causal chains “represent” in me, not something I represent, so that I still have no cognizance of the potential satisfiers of my descriptive thoughts. 38. Once the general concepts, including spatio-temporal concepts, have their significance, the object is that one whose properties or relations are “isomorphic” to the components of the description. This is essentially Leibniz’s view. 39. Our purely cognitive argument for space and time being constructions in pure intuition points up an important difference between our view of what Kant is up to and Michael Friedman’s. In our view Kant is basically and directly concerned with cognition of reality per se. He takes it as a constraint on cognitive theorizing that it be adequate to scientific cognition, but his direct interest, contrary to Friedman’s account, is not to give a theory specifically of mathematical-scientific (Newtonian) cognition. Our purely cognitive argument shows, independent of all scientific concerns, that only if space and time are forms of intuition can thought pertain to reality at all.

92

t h e t r a n s c e n d e n ta l d e d u c t i o n

yond temporizing-cum-tracking (viz., the basic past-oriented cognition as in [4] above) are independent of any present verificatory output or behavior. Representing myself as being beyond (or being past) being affected and reacting a certain way is representing what, perhaps, is to be verified, but not any procedures for verifying it. Thus Dummett is wrong when he says that constructivism is only a metaphor if understood “ontologically” (apart from a denial of the law of excluded middle).40 At least as far as cognition goes, the ontology of space and time as constructive is supported by our argument for an output account of the content of thoughts, which is quite independent of any denial of excluded middle.41 In this essay I have tried to show how Kant’s theory not only avoids the difficulties of descriptive and causal theories, but also how, in outline, it leads to a plausible account of cognition. By adding what Kant would have considered partially empirical notions (such as gross mechanical interaction, shifting spatial attention with and marking time on geodesic vehicles, etc.), and “naturalizing” the ontology of cognitions themselves (equating rules with mechanisms in place), we get a theory in which representation is determinately of substances (as opposed to Quine); a theory in which representation of the past is realist or autonomous (as opposed to verificationism, which “reduces” representation of the past to procedures for present verification); a theory with full spatio-temporal scope (even unto the complexity of relativistic topology); and a theory that leads to a real definition of truth (as active realizability) grounding a recursive truth predicate connected to positive evaluation (by the tie of active realizability to success once purposes are brought in). Further, it is a theory modeled on a very basic and primitive case of thoughts (as when a lion thinks to go after a prey), which nevertheless extends this model to such uniquely human cognitive power as to think of a world of matter in relativistic space-time. Kant himself removed the major stumbling block to an output account of cognition when he showed in his argument for substance that such as account (via rules for being beyond temporizing) can be extended to representation of the past. With this stumbling block removed, Kant’s theory of cognition is worthy of serious consideration. 40. See Michael Dummett, “What Is a Theory of Meaning?” in Mind and Language, ed. Samuel Guttenplan (Oxford: Clarendon Press, 1975), 97–122. 41. Further, Dummett’s “manifestation argument” doesn’t imply verificationism, as long as the mechanisms in place pertaining to the basic Kantian cognitions can be evidenced (and so manifested) as being in place. Once again, the exaggerated account of the cake-baking case is a rough indication of how the crucial past-oriented cognition can be evidenced.

PART III THE PRINCIPLES

6.

Kant’s Proofs of Substance and Causation

1 . p r e l i m i n a ry r e m a r k s Kant’s views on the nature of causation and substance do not depend on any compromise between or any combination of rationalism and empiricism, but on what he calls a “third thing,” the pure intuition of time, which is completely missing in both rationalism and empiricism. For Kant the empiricist position on causation fails to establish the necessary connection between events, that one event “arises out of ” or “emerges” from another. Besides constant conjunction in experience Kant grants the empiricist “empirical” universality through induction (A91, B124, pp. 124–25),1 or completely universal generalization. This universality, however, only implies that all events whatsoever of a certain type are followed by events of a second type—but not that any particular event of the first type forces, produces, or necessarily yields an event of the second type. The regularity theory that defines causation in terms of subsumption under inductively allowable universal generalization2 simply fails to account for the connection in singular causation. For Kant, the rationalist position on causation is that the causal connection is a connection of inference in the intellect, viz., that the existence of a second event can be inferred or deduced from a first event (A243, B301, p. 262).3 But for Kant this idea of inferring existence makes no sense apart from causation (one event’s producing or yielding another) and so cannot explain it.4 1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. 2. For Kant a “true” (as opposed to an inductive) universal would hold that all events whatsoever of a certain type force or produce events of a second type, and so would contain necessity. 3. For this view, see, for example, Spinoza’s Ethics, book 1, axioms 4 and 5, and proposition 3, where the effect is said to be “apprehended” or “understood” by means of the cause. 4. For example, transformations between mathematical equations have to be interpreted as signifying real causal processes before the mathematical deducibility counts as an explanation of a transition in existence.

95

96

the principles

A variant of the rationalist view is that causation is to be understood in terms of explanation. Thus, if we have an explanatory theory according to which an event explains another event, that is all there is to the first event causing the second one. If we combine this rationalism with the empiricist’s regularity theory, we simply get as an analysis of causation that events come under a universal regularity that is also explanatory (a consequence of an explanatory theory). Such a combination is not Kant’s view. If we cannot get the necessity of singular causation from regularity or explanation alone, we cannot get it by combining them. I will argue that Kant derives the nature and universal existence of causation from its function or role in constituting the necessary advance of time and therefore that Kant holds his own unique version of what later came to be called a causal theory of time. He locates the source of necessary connection, then, neither in the inferences of the intellect (rationalism), nor in the features and patterns of events (empiricism), nor in both together. Rather, he finds it in a “third thing,” which is the nature of pure time. For Kant, an empiricist conception of substance is impossible. He says that the concept of substance is what is left “if we remove from our concept of any object .l.l. all properties which experience has taught us” (B6, p. 45). For Hume, roughly, it is aspects of experiences such as uniformity of features or continuity of change that are the sole (objective) basis of the concept of substance or of identity through time. As even Hume recognizes, however, these are not sufficient since they are compatible with the existence of a series of connected but distinct momentary objects. For Kant the rationalist conception of substance is the intellectual concept of a subject that is not also a predicate.5 This concept, Kant says, is “ignorant of any conditions under which this logical pre-eminence may belong to anything” (A243, B301, pp. 261–62). In particular this rationalist conception does nothing to determine a singular use of subject term (pertaining to a substance existing through time) as opposed to a plural use of subject terms (pertaining to momentary existents). Nor will combining the rationalist’s logical concept of a subject with the empiricist’s constancy or continuity of variation determine any application of the concept of substance that neither determines on its own. Kant, I will argue, derives the determinate existence of substances 5. I believe Kant has in mind that the substantiality of something is not any property or feature of it (but rather the support or basis of all properties, as in Locke), and so the subject that pertains to the substantiality itself cannot also be any predicate (which pertains to features).

Proofs of Substance and Causation

97

from their role or function in constituting the “permanence” or the ongoing nature of time—i.e., the fact that the present doesn’t begin time but continues an already existing duration. Kant holds what can be called a “substance-theory” of the ongoingness or lastingness of time. He locates the source of substantial identity, then, neither in the individuative apparatus of the intellect (rationalism), nor in the steady qualities of experience (empiricism), nor in both together. Rather, he finds it in the “third thing” which is the nature of time itself.

2 . k a n t ’ s p r o o f o f c au s at i o n i n the second analogy Kant claims to derive causation from objective succession by showing that the representation of objective succession entails that every event has a preceding condition that necessitates the event. The argument, then, is meant to answer Hume’s skepticism regarding the nature of causation itself (of the causal tie or the necessary bond between events), and his skepticism regarding the universal applicability of causation. My contention is that Kant’s derivation turns on his holding what may be called a “partial causal theory” of time,6 so that it is this theory that is the crux of Kant’s answer to Hume. The Second Analogy begins (A190, B235–A191, B236, pp. 219–20) with a discussion of what objectivity means, if appearances alone are “what can be given us to know.” For present purposes we can understand this to mean that all we have to deal with are sensory representations. I shall call these representations reactions or responses to emphasize their passivity, but so far there is no implication that these are reactions or responses to entities outside us. Now, objectivity, for Kant, requires a distinction between our representations, on the one hand, and that which they agree with (or fail to agree with), on the other. Indeed, for Kant truth consists “in the agreement of knowledge with its object,” and so the question becomes, what is there for our actual reactions or responses to agree with? Kant’s answer is that our reactions may agree with, or fail to agree with, a rule for reacting. Thus I may in fact first react r1 (where r1 is, for example, ship upstream) and then react r2 (ship downstream). This actual sequence of reactions may or may not agree with how it is proper or legitimate to react. How it is legitimate or proper to react is a constraint on our actual reactions, since we can fault our actual reactions for not being faithful to how it is proper 6. For the causal theory of time, see Henry Mehlberg, Time, Causality, and the Quantum Theory (Dordrecht: D. Reidel, 1980).

98

the principles

to be reacting. The notion of a rule, then, takes over the function of some actual entity outside our actual representations, of being a constraint on those representations.7 As an analogy, consider making an actual move in a chess game. The move may agree or not, or may correspond or not, to how it is then legitimate or legal to move. Thus, without going outside an ontology of moves, one has a distinction between actual moves and what constrains actual moves, namely, legitimate moves. Kant next (A191, B236–A193, B238, pp. 221–22) applies this notion of objectivity to the case of succession, and finds that the rule for an objective succession is irreversibility of the order of reacting. The rule, that is, is that it is legitimate or proper to first react r1 and then r2, but not legitimate to first react r2 and then r1. To think our actual successive reactions as subject to, or governed by, such a rule is to think of them as not only being successive but also as representing what is successive. Equivalently, it is only when we think the order of reacting as necessary or required, in the sense in which it is necessary to move a bishop only diagonally, that we think of the succession as something more than the order in which we happen to react. For Kant, the objective succession is then the necessary order of proper reactions. Lovejoy and Strawson8 charge Kant with a non sequitur in arguing from the order of our reactions being necessitated by the sequence of states constituting the event outside us, to the necessity of the order of that sequence of states itself. This, however, is to miss Kant’s point that the only “states” that stand against our actual reactions are proprieties of reacting. My apprehension of the ship being upstream and then downstream is not bound down by necessity to the order of a distinct sequence of states (ship upstream, ship downstream) whose order is definable apart from necessity (as simply being the order of states that are outside my apprehension). Rather, necessity is built into the very conception of that which constrains my apprehension, namely, the rule of how it is necessary or required to react. To think of my apprehension as bound down at all is already to think of it as bound down or constrained by what has a necessary order (viz., the rule), and so there is no non sequitur over the notion of necessity. An objective succession is just a rule containing a necessary or required order of reaction. The charge of a non sequitur by Lovejoy and Strawson depends on 7. See Kant’s discussion of objectivity in the A edition Deduction at A104–5, pp. 134–35, which is of a piece with his discussion in the Second Analogy. 8. A. J. Lovejoy, “On Kant’s Reply to Hume,” in Kant: Disputed Questions, ed. M. S. Gram (Chicago: University of Chicago Press, 1967); Peter Strawson, The Bounds of Sense (London: Methuen, 1966).

Proofs of Substance and Causation

99

attributing to Kant a kind of realist view that he does not hold. Various commentators have pointed this out.9 The point remains, however, that the supposed necessity in the order of reacting that avoids the non sequitur is still not sufficient for any sort of causal connection. From the fact that it is legitimate or required to react r1 and then r2 but not vice versa, it does not follow that the legitimacy of reacting r2 is due to (or forced or determined by) the legitimacy of reacting r1. From the fact that it is legitimate to react ship upstream only before ship downstream, it does not follow that its being legitimate to react ship upstream makes it the case that it is legitimate then to react ship downstream. The rule only says that if I now have both sensible reactions, I can only have them in one order; not that the second reaction has to happen at all.10 Thus, suppose I see ship upstream. The rule as yet is not even operative, since it is a rule only for thinking a necessary order when I have both successive reactions. The rule then does not even imply that if it is proper to react ship upstream, then it is also proper to react ship downstream. Since it doesn’t even imply that the second reaction is proper after the first one, it certainly doesn’t imply that the propriety of the second reaction is caused by or forced by the propriety of the first reaction. Fortunately, however, it is not Kant’s contention that irreversibility is causation. Rather, in the paragraph at A194, B239, p. 222, Kant says not that the rule of irreversibility is causation, but that in conformity with it there must (also) be a causal connection. Recall that the rule of irreversibility is supposed to express or represent objective succession. But now there is more to the idea of objective succession than that it is distinct from objective coexistence (the distinction expressed in the rule of irreversibility). An objective succession, further, must have existence in the time-series; i.e., it must exist at a stage in the previously ongoing course of time.11 What is this time series like? Kant says in this paragraph that “The advance, on the other hand, from a given time to a determinate time that follows is a necessary advance.” The series of times, that is, is such that earlier times nec9. See, for example, James van Cleve, “Four Recent Interpretations of Kant’s Second Analogy,” Kant-Studien 64 (1973): 84. See also Henry Allison, Kant’s Transcendental Idealism (New Haven: Yale University Press, 1983), 233. 10. This point is made by both Jonathan Bennett, Kant’s Analytic (Cambridge: Cambridge University Press, 1966), 221, and van Cleve, “Four Recent Interpretations,” 80. 11. That objective succession itself involves, besides irreversibility, a connection to what precedes is made clear in the paragraph at A195, B240, where Kant says, “For mere succession in my apprehension, if there is no rule determining the succession in relation to something that precedes does not justify me in assuming any succession in the object” (emphasis added). Kant here is clearly saying that some relationship to what precedes the succession (and hence something beyond the irreversibility in regard to the apprehension) is required for objective succession.

100

the principles

essarily advance to later times. Given earlier times, the later times must happen. Later times fully emerge from earlier times. Earlier times force or determine the existence of times afterward. All this is just to say that further later times do not just happen to come after preceding times— they necessarily do. Now it is upon this character of the time series that Kant bases his conclusion that there must be something that determines the succession expressed in the rule of irreversibility. Equivalently, it is from the necessary advance of time that Kant concludes to the existence of a preceding causal or determining condition of that succession. Kant’s argument is that since time itself is not an object of perception, the necessary advance in the time series has to be represented by, or within, the series of objective occurrences themselves. Thus he says later on (A200, B245, p. 226), “Now since absolute time is not an object of perception, this determination of position [in time] cannot be derived from the relation of appearances [i.e., reactions or legitimate reactions] to it. On the contrary, the appearances must determine for one another their position in time, and make their time-order a necessary order” (emphasis added). In other words, Kant is holding that the necessary time order (that previous times force, or necessarily advance to, succeeding times) has to be represented within or between occurrences. Let us put this all together now. I must represent the succession expressed by the rule of irreversibility as occurring in the already ongoing time series. This series is one in which preceding times necessarily advance to succeeding times. But this necessary advance itself has to be represented within or between occurrences. Therefore, the succession expressed by the rule of irreversibility has to be represented as necessarily advancing from a preceding occurrence (which occurrence, likewise, to be placed in the time series has to be represented as advancing from an occurrence preceding it, etc.). Indeed, the preceding occurrence must determine or force the succession. The succession must emerge from or arise out of the preceding occurrence. In sum, it cannot be that the succession just happens to come after the preceding occurrence, if the necessary advance of times is to be represented in the series of occurrences themselves. But the idea of a preceding occurrence determining or forcing what comes after is just the idea of a causal connection. In this way it follows that if the succession expressed by the rule of irreversibility is to be an objective succession (viz., a succession determined in the time series) then “I must refer it necessarily to something which precedes it, and upon which it follows .l.l. of necessity” (A194, B239, p. 222), hence representing it as in causal connection.

Proofs of Substance and Causation

101

Let us try to formulate Kant’s conclusion in terms of his conception that occurrences or states are appearances. Recall that an occurrence, as opposed to its apprehension, is simply a proper reaction, as opposed to an actual reaction. The purported objective succession, then, that has to be placed in the time series is simply that it is legitimate to first react r1 and then to react r2. To place it in a series that represents the necessary advance of time requires a preceding condition, say r0, which forces or determines it; i.e., it must be that the propriety of reacting r0 forces or determines the propriety of reacting r1 and then r2. There is then a necessary tie or bond between proper reactions so that succeeding proper reactions emerge from or necessarily arise from preceding ones, as opposed to just happening to arise after preceding ones. The relata then of causal connections are proper reactions. These are acceptable as relata since, for example, I can say that it is proper or legitimate to react “smoke” because it was proper or legitimate to react “fire,” and in saying this I am expressing a causal connection. The same is true of the propriety of reacting r0 (the reaction preceding the original succession). It too must be placed in the time series (which then places the original succession in a longer time series), which entails that for some r–1 the propriety of reacting r–1 forces the propriety of reacting r0. In this manner not only is it shown that the original succession is caused, but that all members of the ongoing objective series are caused. In sum, from the very notion of an objective succession, we have the conclusion that all objective occurrences (occurrences in the time series) have a necessary tie or bond to preceding occurrences that determine or force them. The causal bond, that is, is universally applicable. One might think there is the following gap in Kant’s argument for the universality of the causal principle. A proper reaction, say r, might exist simultaneously, and even in the same object, as another proper reaction r*. The one reaction r could have a place in the necessary advance of the time series by arising together with r*, which latter is causally tied to a preceding series of reactions (and so has a place in the time series), without r being causally tied. In this way r inherits a place in the necessary advance without itself being caused. This gap, I believe, is closed by being careful as to what it means for proper reactions to represent the necessary advance of time. If the later moment, as opposed to one particular proper reaction at that moment, is to be represented as emerging from or being determined by the earlier moment, then every possible proper reaction at that later moment must emerge from or be determined by a reaction at the earlier moment. Otherwise, it isn’t the later moment that emerges or is determined, but

102

the principles

only some reactions at it. Roughly, properties such as emergence characterize the moment exactly by characterizing the equivalence class of all possible reactions taking place at it. As an analogy, something is a feature or property of the property wise not if some wise things have the feature, but if all possible wise things have it. Similarly, if we speak abstractly and say that “The bird comes from the dinosaur,” we mean that all birds are descended from dinosaurs. Now to say that “The present necessarily emerges from the preceding” is to speak abstractly, and means likewise that all present occurrences necessarily emerge from earlier preceding ones. If this is correct, then the scope of the causal principle is shown to be absolutely universal, extending to all possible proper reactions whatsoever. Kant seems to be making just this point at A199, B244, where he says, “If then .l.l. the preceding time necessarily determines the succeeding .l.l. it is also an indispensable law of empirical representation of the time series that the appearances of past [preceding] time determine all existences [emphasis added] in the succeeding time .l.l.”

3 . r e m a r k s o n k a n t ’ s ac c o u n t o f c au s at i o n Kant’s argument, as we have presented it, depends on his holding a version of the causal theory of time, viz., the theory that time relations are not something over and above causal relations. That Kant does hold such a theory is made completely clear in the passage just cited at A199, B244. He there says that time has the characteristic “that the preceding time necessarily determines the succeeding.” Because of this, he says, “it is also an indispensable law of empirical representation of the time series that the appearances of past time [necessarily] determine all existence in the succeeding time.” He is arguing, then, that the characteristic of preceding times determining the succeeding has to be represented in terms of (is nothing over and above) a relationship of appearances determining later existence, and so in terms of causal relationships. It is not that moments or stages of time determine one another alongside appearances that determine one another, but rather that “since absolute time is not an object of perception .l.l. the appearances must determine for one another their position in time and make their time-order a necessary order” (A200, B245, p. 226, emphasis added). In other words, it is between the appearances alone that the relation of the preceding determining the succeeding holds. Although Kant holds a causal theory of time, it is in fact a partial causal theory, since not all characteristics of time are reduced to characteristics of appearances. In particular, Kant is not defining the rela-

Proofs of Substance and Causation

103

tion of earlier to later itself in terms of occurrences. It is only the necessary-determination aspect, not the serial-order aspect, of this relation that Kant claims must be found in the appearances alone. Kant holds, indeed, that causation itself can only be understood in terms of the earlier-later relation. Thus the schema of causation involves succession (A144, B183, p. 185), and it is this notion, not some pure a-temporal notion of causation or determination, that Kant is arguing must apply to experience. He says that “in applying it [the pure category] to appearances, we substitute for it its schema as the key to its employment, or rather set it alongside the category” (A181, B224, p. 212). Succession (the schema) is applied, then, alongside or together with the pure a-temporal notion of determination (that is, the concept of ground and consequent expressed in the hypothetical judgment). Kant is certainly not defining the schema of succession in terms of the pure category. The schema, rather, is something added. This is made clear when he says at A140, B179, p. 182, that “pure a priori concepts in addition to the function of the understanding expressed in the category, must contain certain formal conditions of sensibility [viz., time relations not definable in terms of the understanding].” Kant is allowing that I can represent succession (temporal order) without thinking of causation, as when I formulate the rule of irreversibility. Here I am representing that it is only legitimate to react r1 before r2. I can further think there are other proper reactions that precede my apprehension, so that it is also legitimate (though too late for me) to react r0 before reacting r1 and r2. Kant’s point is that all of this is still not sufficient for representing succession in a time series where the preceding time determines the succeeding. This latter aspect of the relation of succession or temporal order, viz., its necessary advance, is what is representable for Kant only in terms of a relation between the occurrences themselves (and which requires, alongside the schema of succession or order, the concept of determination or of ground and consequent). To place the legitimacy of reacting r1 before r2 in a time series where the earlier determines the later, that is, I must represent not only that it is legitimate to react r0 before it is legitimate to react r1 and then r2, but that the legitimacy of reacting r0 earlier determines, or has as a consequence, the legitimacy of reacting r1 and then r2. Kant’s partial causal theory of time differs radically from Leibniz’s full causal theory. For Leibniz, the law of activity of a substance or monad, according to which its states determine one another in an a-temporal (although asymmetrical) sense of determination, well founds the phenomenal (temporal) order. In other words, some nontemporal sense of determination is the fundamental order in terms of which

104

the principles

there is temporality at all. In Kant’s terms Leibniz is applying a nonschematized concept of causation (determination of ground to consequent according to a law) to determine the temporal order. It should be clear that such a view goes against all the fundamentals of Kant’s thinking. Understanding the difference between Kant’s partial causal theory and Leibniz’s total causal theory enables us to avoid the objections Suchting makes12 to attributing the causal theory of time to Kant. Suchting recognizes the two paragraphs at A199, B244–A200, B245, as apparently expressing a causal theory of time, but he rejects them as incompatible with Kant’s thinking, since Suchting understands them as an attempt to derive the form of sensibility (i.e., time) from the form of understanding (i.e., causation). Further, he holds that such a theory is circular, since the notion of causation makes reference to the notion of succession it is trying to define. Each of these objections would be relevant if these paragraphs were expressing a Leibnizian version of the causal theory, but are irrelevant against the partial causal theory we have attributed to Kant. According to Kemp Smith and Suchting,13 the two paragraphs at A199, B244–A200, B245, which focus almost entirely on expressing a causal theory of time, constitute an argument distinct and separate from the rest of the Second Analogy. On their view Kant’s main argument for causation is quite separate from having to represent the necessary advance of time in the appearances themselves. A careful reading of the text, however, shows that in each and every presentation of the argument Kant includes a reference to the nature of the time series. Thus, in the paragraph at A194, B239, where he first concludes (in the A edition) to the existence of causal connections, a premise is that “The advance, on the other hand, from a given time to the determinate time that follows is a necessary advance.” In the paragraph at A196, B241, after making his second conclusion to causal connections, he explains the nature of his argument by saying, “Nevertheless the recognition of the rule [that everything that happens has a cause] as a condition of the synthetic unity of appearances in time, has been the ground of experience itself.” In the paragraph beginning at A198, B243, where he makes his third conclusion to causal connections, he says the determining cause is required for “connecting the event [the succession] in necessary relation with itself in the time-series.” The two 12. W. A. Suchting, “Kant’s Second Analogy of Experience,” in Kant Studies Today, ed. L. W. Beck (La Salle, Ill.: Open Court, 1969), 322–40. 13. Norman Kemp Smith, A Commentary to Kant’s Critique of Pure Reason (London: Macmillan, 1918), 363, 375. See also A. C. Ewing, Kant’s Treatment of Causality (London: K. Paul, Trench, Trubner, 1924), 73, and Allison, Kant’s Transcendental Idealism, 222.

Proofs of Substance and Causation

105

paragraphs that follow, which focus on the causal theory of time, do not then constitute a separate argument, but make clear in a general manner the issue (representing the necessary advance of time) that is involved in each of the presentations of the argument. The introductory statement to the Analogies likewise expresses that in each of the Analogies the proof of the category depends on the fact that “time, however, cannot itself be perceived [and so] the determination of the existence of objects in time can take place only through their relation [to one another] in time in general .l.l. through a representation of necessary connection of perceptions” (B219, p. 209).14

4 . c au s at i o n a n d t i m e a s a p u r e i n t u i t i o n A key premise of Kant’s argument is that time itself cannot be perceived (so that the necessary advance of time has to be represented within or via the connection of legitimate or proper reactions). Of course Kant, in the Aesthetic, has already argued that time is not an objective, self-subsistent entity (A32, B49, p. 76). Roughly, then, the reason he holds that time itself cannot be perceived is that time itself as an object does not exist. Indeed, time exists for Kant only as a pure intuition. In the Aesthetic Kant holds a constructivist theory of space and time that is somewhat parallel to a constructivist theory of number. Just as for the constructivist numbers exist only as termini of counting procedures, so too for Kant space and time exist only in flowing procedures or flowing constructions. The reason for this is that any space or time is a continuous expanse or extent. This, for Kant, implies that it is a whole that is prior to its parts or elements (A25, B39, p. 69) rather than composed out of them. Indeed, the “seamlessness” of a continuous expanse is just this fact that it is so seamless that it cannot be constituted out of elements. Now, an objective whole exists only by all its elements existing, and so is composed of them. By contrast, a flowing construction, such as sweeping out a line, is not composed out of any cuts or stops that construct parts of the flow. Only as flowing constructions, then, or motions of the subject (B155, p. 167), can stretches of space and time be continuous. An example of a temporal construction, or what we may call “temporizing,” would be a conductor who paces or tempers the orchestra’s playing of a note by a downbeat gesture, which is, indeed, a flowing performance. For something to be given in intuition, for Kant, is for it to be immediately presented. Space and time, being immediately presented by constructing or performing, rather 14. See also the formulation of the principle of the Analogies at A177, p. 208.

106

the principles

than by sensing, are thus given in pure intuition. Moments of time are limits or cuts of the flow. Thus the start of the downbeat and the terminus constitute the construction of successive moments of the extensivity that time is. This supports our contention that Kant is not holding a causal theory of the successive order of time (of the earlier-to-later relation of moments), since this order is given in pure intuition (in construction). Let us suppose then that in the Aesthetic Kant is holding a constructivist theory of space and time. How does this theory cohere with the partial causal theory we claim that he holds in the Second Analogy?15 To begin with, different aspects of time are involved in the two accounts. Whereas the expansiveness of time and its seriality are constructed, it is the necessary advance of the seriality that is represented causally.16 Second, the necessary advance cannot be represented in construction. Construction is not such that having constructed an extent forces or determines a further construction. Nor does the propriety or legitimacy of one construction force or make a second construction proper. The legitimacy of my now going ahead to temporize is not forced or determined by the propriety of preceding constructions. Indeed, each construction is proper or legitimate on its own. The order of proper constructions may be irreversible, but, as we have seen, irreversibility is not sufficient for determination. It is only proprieties of reacting or responding (via causation) that can force or produce subsequent proprieties. The causal theory of the Second Analogy, then, is not redundant. Third, a causal theory does not represent moments of substantival time, or even temporal relations as with relational time. The causal theory adds only a dynamical relation of production or necessary advance among occupants (proper reactions) to the aspects of time represented in construction. In other words, nothing specifically temporal that exists objectively (or outside of construction) is added. Space and time for Kant are not only pure intuitions, they are also forms of empirical intuition (that is, reacting or responding perceptually). It is, so to speak, by proper reacting being tied to (being in the course of) proper constructing that reactions “inherit” successive order separated by expanse. But now, in the Second Analogy, it is by proper 15. For the contention that his theory in the Analogies is inconsistent with his account in the Aesthetic, see, for example, T. K. Swing, Kant’s Transcendental Logic (New Haven: Yale University Press, 1969), 151–52, and Robert Paul Wolff, Kant’s Theory of Mental Activity (Cambridge: Harvard University Press, 1963), 263. 16. Toward the end of the Second Analogy (A207, B253, pp. 230ff.) Kant argues that alterations are continuous because time is a continuous magnitude; viz., the alterations don’t represent continuity but conform to the distinct continuity of time. This shows that Kant does not hold a causal theory of time magnitude.

Proofs of Substance and Causation

107

constructing being tied to (or encompassing) proper reacting (in accord with the determination by causation) that the successive order (of constructing) inherits necessary advance from the earlier to the later. Thus in the Second Analogy Kant is adding a dynamical dimension (causation) to the representation of reacting in the course of constructing, which is not adding any extra-constructive ontological time (whether relational or absolute).17 In this fashion, I believe, the partial causal theory Kant is holding in the Second Analogy is fully consistent with his account of time in the Aesthetic that time is merely a pure intuition that is the form of empirical intuition. The account in the Aesthetic also explains, I believe, why Kant can argue from the unperceivability of time to causation, despite the fact that he explicitly holds that causal necessity is likewise unperceivable. Time is unperceivable because it exists only constructively, not objectively. Causation, on the other hand, is a dynamical notion, not a constructive one. The crux is not unperceivability but extra-constructive existence, which causation has but time lacks.

5 . k a n t ’ s a n sw e r t o h u m e In our account so far, Kant’s argument is that objective succession requires first rules for reacting, since objectivity concerns the propriety or legitimacy of reacting. Second, it requires a connection of necessary advance between such proper reactions, in order to represent the necessary advance of the time series within which any objective succession takes place. Embedded in this account is Kant’s answer to Hume. First, proper reactions are connected by a tie or nexus of producing, forcing to happen, determining, or necessarily emerging into, etc. This bond derives from the nature of the time series that the relation among proper reactions has to represent, viz., that the earlier time determines or forces or necessarily advances to the later time. In particular, the bond or connection is completely independent of Humean propensities to transfer force and vivacity upon associating ideas with impressions. Kant, that is, “finds” the singular causal nexus in the nature of 17. The causal theory, in this sense, is not a version of the relational theory, because it does not add any specifically temporal objective relations to time construction. Rather, it adds the causal relation among proper reactions instead of any objective temporal relations. Equivalently, it adds a dynamical component to the constructivist account of time. As an analogy, adding a dynamical component to the notion of straightness (geodesic) in terms of force-free motion is not adding a further (relational or substantival) spatial component to that notion. Likewise, adding a dynamical component in terms of causation to the notion of time order or time series (specifically to its aspect of necessary advance), is not adding a further (relational or substantival) temporal component.

108

the principles

time, not in mental habits. Second, this causal (productive, determining) connection must pertain to every proper (objective) reaction whatsoever, since all such reactions have a place in the time series. In other words, both the causal nexus and the universality of the causal principle are conditions of the possibility of experience (or of the representation of objective succession). Kant is clearly addressing Hume in the paragraph at A196, B241, p. 223, where he begins by saying that the concept of cause does not depend on repeated succession of events; i.e., it doesn’t depend on constant conjunction in experience. Of course, on Hume’s analysis it does, since constant conjunction is involved in association and so in transference of force and vivacity. Kant says that such a concept of cause would be merely empirical. Notice that on his view the concept of cause (production, necessary determination) derives not from constant conjunction, but from the nature of the time series. Kant further says that Hume’s account makes the principle “that everything which happens has a cause” contingent (viz., it would be accidental depending on how much regularity there happens to be in experience). Kant’s proof of the principle, rather, is that “experience itself is brought about only by [its] means” (viz., the representation of the universal necessary advance of the time series requires universal causation, or that every proper reaction whatsoever emerges from, or is produced by, a preceding proper reaction). Indeed, this “rule determining the series of events” (that each event emerges or is produced from something preceding) Kant says is “a condition of the synthetic unity of appearances in time.” In this paragraph, then, Kant is holding that regularity (uniformity, constant conjunction) is irrelevant to establishing either the nature of the causal bond itself or the universal causal principle (that this bond is universally applicable). Nevertheless Kant does believe that the causal relation involves universality (that events similar to the cause are always followed by events similar to the effect). This strict universality follows, I believe, not from any inductive inference based on regularity, but from the necessity that holds between cause and effect.18 The necessity with which a preceding time emerges into the succeeding is an absolute or unrestricted necessity, as opposed to a necessity limited in some regard or fashion. But now suppose A causes B only in the present case, or only in some cases. Then the necessity by which A emerges into B would be restricted or conditional. It would not be absolutely necessary that B follows A, but only now (or in some cases) necessary. Thus the relation between A and B could not carry or represent the absolute necessity of the advance of 18. See B4, p. 44, where Kant says, “Necessity and strict universality are thus sure criteria of a priori knowledge, and are inseparable from one another.”

Proofs of Substance and Causation

109

the time series. Alternatively, the necessity in the advance of time has a constant character at all times. Each moment flows from or emerges from preceding time in the same way as any other moment. This homogeneity in the way time necessarily unfolds has to be represented by a homogeneity in the way events necessarily unfold into other events, viz., by this necessary unfolding having always the same character. But this universality of character of causation is just the same cause-same effect principle, or the principle that all causes and effects come under universal laws of what causes what. Thus, although regularity may be an empirical criterion for ascertaining what specifically causes what, it is not a basis for the strict universality that holds between cause and effect, which has its basis rather in the homogeneous nature of the necessary advance of time. So far Kant’s answer to Hume has not taken Kant outside the realm of proper reactions or proper perceptions. So far, that is, Kant, like Hume, is a phenomenalist; viz., the ontology includes only perceptions and related notions such as either imaginative propensities regarding perceptions (as with Hume) or else rules of proper perceptions (as with Kant). However, in the very first paragraph of the Second Analogy, and again in the paragraphs beginning at A204, B249, Kant makes clear that causation requires or involves substances, thus going beyond any version of phenomenalism. I shall consider first, in some detail, Kant’s proof of substance in the First Analogy and then return to his claim that causation imputes substances.

6 . k a n t ’ s p r o o f o f s u b s ta n c e In the First Analogy Kant says (B225, p. 213) that the time (form of intuition) in which all time relations are thought is permanent. Further, since time itself (that is, objective time) cannot be perceived, the “substratum” which represents time in general must be found in the objects of perception, and that this permanent is substance. Note that he says that time is the permanent form of intuition. If permanence is in time as a form of intuition, however, then why must it also be represented in the objects of perception? Suppose we accept that Kant is a constructivist regarding time. I can now carry out a temporizing construction or procedure such as marking time. But time exists prior to any such construction I can now carry out. My constructions, that is, do not begin a new time, but continue the flow of an already ongoing time. What this means is that if time as a form of intuition (as construction) is to be “permanent” (a continuation) then there must be some way of representing construction prior to any that I can go ahead to

110

the principles

perform. Of course, besides actual constructing, there are rules of how it is proper to construct (whether one actually does so or not). Such rules, however, will not enable a representation of earlier construction if they are simply rules for how it is proper or legitimate to go ahead and temporize. What I suggest now is that it can be presently proper or legitimate to be in the course of (at a middle or end stage of) procedures or constructions, whether or not one has actually performed the initial stages. Thus, suppose I have a rule for (a procedure of) baking a cake. I shall presume that this is an ordered, step-by-step procedure for adding ingredients, mixing ingredients, etc. Now suppose I come into a room and find that the first three ingredients are sitting in the bowl. Then I claim it is proper for me presently to be up to stage 4 in the cake-baking procedure, even though I haven’t added the first three ingredients. Note that it is only something in my present circumstance that can “make it” proper to be so far along in the procedure rather than at the beginning. If I represent that it is thus legitimate to presently be up to the fourth ingredient in the procedure, then I represent my present performance of putting in the fourth ingredient as a continuation of preceding stages of a larger procedure. In order, then, to represent temporizing as a continuation (rather than the construction of a new time) I must represent that it is legitimate for me to be up to a certain stage of a larger temporizing procedure, rather than at the beginning. Let the temporizing procedure be to mark time by a series of downbeat gestures, where each such flowing gesture is accompanied by reciting a numeral. Such a procedure would begin with reciting “1” at the end of a downbeat, then reciting “2” at the end of the next downbeat, etc. Then the representation of its being presently legitimate to be up to k in such a procedure would represent that a present construction I can perform (a downbeat together with a recitation of k + 1) is a continuation of a larger temporizing construction. In this manner, time as a form of intuition (as constructive) would be “permanent” (viz., it would be a continuation of constructive time). But now, as in the cake-baking case, something in my present circumstance must “make it proper” to be so far along in the temporizing procedure, rather than at the beginning. After all, a procedure is always to do first things first (put in the first ingredient first, recite “1” at a downbeat first, etc.) unless something presently sets me ahead in the procedure. Let us recapitulate. If time as a form of intuition is to be permanent (not something begun anew with a present construction), then it must be presently proper or legitimate to be in the course of a constructive temporizing procedure. But then something in the present circumstance must make it legitimate to be beyond or past earlier stages of the

Proofs of Substance and Causation

111

procedure (despite not having performed them); i.e., something present must be the basis or “substratum” of the permanence of time. Now, this something cannot be objective time itself. It is not, that is, that objective time is presently so far along in its unfolding that “to keep up with it” it is legitimate to be presently so far along in a temporizing procedure. Hence, it must be rather that something real is presently so far along in its existence that “to keep up with it” it is proper to be so far along in a temporizing procedure geared to its existence. Existence or reality for Kant pertains to that which affects us. And continued existence pertains to continued affection. Continued affection is what obtains in keeping track of what affects. What the temporizing procedure keeps up with, then, is tracking, so that we have finally the following representation of the substratum or basis of the permanence of time: With respect to what presently affects (that is, the real), it is proper to be so far along (up to k) in a temporizing-cum-tracking procedure (that is, a procedure that marks time while keeping track). But now this represents what presently affects as being something proper to have been tracking. It represents, that is, that what is presently real is also what is proper to be in the course of (at a non-beginning stage of) tracking. But this is exactly to represent the presently real as having previous existence, or as being “permanent” (its “permanence” extending back as far as the tracking procedure that is presently up to k). Permanence of the real, of course, is just substance for Kant, and so the substratum or basis of representing (past) time is substance.19 Note that in our account of the argument, substance is, as Kant says it is, the condition of time magnitude (A183, B226, p. 214; A177, B219, p. 209) or time duration. Our temporizing or marking-time construction, in numbering the flows (the downbeats), makes that flow a unit for measuring an extent of time or duration of time. The extensivity (continuity) of time is exactly what exists for Kant only in flowing construction. The point of the argument, then, is not that continuity (and with it, duration or magnitude) is represented outside of flowing constructions in substance. Rather the point is that representing past extensivity or duration is a matter of representing myself as being in the course of (past or beyond stages of) extensive flowing construction, and that this 19. Once again the fact that substance (permanence of the real) is no more perceivable than time itself is no objection to Kant’s argument. For this objection, see Paul Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987), 219. There is no objective continuous flow of time to keep up with, while tracking what is real (what affects) is holding or moving attention continuously. That Kant has to go to something objectively dynamical rather than objectively temporal is required by time’s being merely a form of intuition.

112

the principles

requires a basis in my present circumstance, and so it requires substance. Substance, that is, is required to extend the scope of construction “into the past.” Just as in the Second Analogy, the necessary advance of the time series goes beyond constructive time, and requires gearing the time order (succession) to dynamical causation, so too, here in the First Analogy, past duration (extensivity) of time goes beyond constructive time, unless it can be proper to be in the course of such construction, which requires gearing the construction of magnitude to tracking in regard to what presently affects. In both cases the representation of a mode of time (past duration in the case of the First Analogy, and necessary advance of the time series in the case of the Second Analogy) is made possible without adding objective (extra-constructive) temporality (but rather adding existence in the one case and causation in the other). In our reading of the paragraph at B225 so far, the real that presently affects is the substratum or basis of the permanence of time as a form of intuition (viz., of constructive time). Kant also holds, however, that only in this permanent form of intuition “can either coexistence or succession be represented.” In other words, the time series or time order is represented as within the extensivity that time is. Kant goes on to say that “all that belongs to existence can be thought only as a determination of substance.” What belongs to existence are appearances or proper reactions. Suppose that I want to represent that the propriety of a certain reaction arises in past time. I must represent myself as presently being beyond or past so reacting. This in turn requires representing myself as being beyond or past an extensivity of time at which the reaction is proper. But we have just seen that this requires representing myself, based on present reality, as being beyond a temporizing-cum-tracking procedure. It requires, that is, representing myself, with respect to what presently affects, as being up to k in marking time while tracking. It is this which represents the “permanence” of time as a form of intuition (ongoing temporizing). To represent the appearance (the proper reaction) in relation to the permanent form of intuition, then, is just to represent the reaction as what is proper at the beginning of the procedure. We have, then, the following representation of appearance in the “permanence” of time as a form of intuition: With respect to what presently affects, it is legitimate to be up to k in temporizing while keeping track from first reacting r. This represents the propriety of reacting r as something proper upon initially tracking what presently affects me (though I am now up to k in tracking it). But this is just to say that the appearance (reacting r) pertains to this which is before me, only I am past or beyond so reacting (tem-

Proofs of Substance and Causation

113

porally separated from so reacting by being in the course of a temporizing procedure rather than at the beginning, where the reaction is proper). In this way, the appearance (the proper reaction) is a “determination of substance” (of that which is before me), where, roughly, “determination” means that the reaction is to be had with or upon keeping track. Not only, then, is substance the basis of representing past time, but it is the basis of representing proper reactions in past time, which are represented, indeed, as pertaining to (that is, as determinations of) trackable existence. Thus Kant can say, “the permanent is the object itself .l.l.l; everything on the other hand which changes or can change [i.e., proper reactions] belongs only to the way in which substance or substances exist, and therefore to their determinations” (A184, B227, p. 214).20

7 . t h e r e l at i o n o f c au s at i o n t o s u b s ta n c e We return now to the relation of causation to substance in the Second Analogy (A202, B248–A205, B251, pp. 227–29). We begin with Kant’s illustrative example of a ball hollowing out a cushion. There is a succession of states in a substance (the cushion) from first flat to then hollowed out, and an influence of a second substance (the ball being placed on the cushion). Involved in the transition of states, then, is both the nature of the substance that undergoes it (its being a soft cushion) and the nature of a second substance that influences it (its being a hard ball). Note that the second state of hollowed out emerges from or necessarily advances from the first state; i.e., the key relationship of necessary advancement between states (between proper reactions) required for representing the necessary advancement of time still obtains. Now, however, it obtains because they are states of a substance with a nature under the influence of a second substance with a nature. The example, that is, does not overturn the key relationship of a necessary succession of states that is the crux of the argument of the Second Analogy.21 What I wish to suggest is that this model of causation follows 20. This answers Guyer’s second objection (Kant and the Claims of Knowledge, 220–21) that Kant equivocates in going from substance as permanent (or enduring) to substance as the bearer of properties. There is no “too hasty” transition here. The only way to represent past appearances (proper reactions) is to represent them in past time. The only way to represent past time is by its being geared to trackable existence (the permanent). Hence past appearances can only be represented at all as geared to trackable existence (i.e., as determinations of the permanent). There is a difference between focusing on what is real and keeping track of it versus specifically reacting to it. This, I claim, is the difference between substance and the “properties” (proper reactions) it bears. 21. Guyer (ibid., 260) makes the point clearly and definitively that the connection of causation to substance is compatible with the argument of the Second Analogy.

114

the principles

from the fact that the time series that causation is supposed to represent is itself also limited to individual substances. Kant holds, that is, what can be called a “substance-based” theory of time, according to which the time series exists basically or fundamentally in relation to individual substances.22 This substance-based theory is akin to the modern idea that time basically is “proper time” holding along individual world-lines, each world-line having its own proper time. The necessary advance in a time series, then, is an advance in regard to an individual substance. Further, this substance-based theory follows from the fact that the time series unfolds in relation to the extensivity that time is, and such time duration or time magnitude (as per the First Analogy) is itself representable as a procedure only in relation to individual substances. Thus Kant’s model of causation as involving a necessary succession of states of a single substance coheres with his understanding of the very nature of the time series that causation is meant to represent. One may ask, now, how this model of causation is compatible with objective successions between states of different substances. Thus first a ship is upstream, and then a bell on the shore rings. This succession of states is objective, but there is no causal connection between them, according to the model of causation in the Second Analogy.23 The ship upstream, to be an objective occurrence, must be represented in a necessarily advancing time series with regard to a single substance, and similarly for the bell ringing in regard to a second such series. It is only when they are each thus represented as being two objective occurrences (two occurrence arising in time) that any issue of their relationship (their succession) comes up. It is only in the Third Analogy that this subsequent issue of objective temporal relations between states of different substances is considered. Kant’s representation of the causal series can be set out as follows: With respect to what presently affects me, it is legitimate to be in the course of a series of successive reactions, each of which necessarily advances to the next. Since this series supposedly ends with present “irreversible” reactions (like ship upstream, ship downstream) this present objective succession 22. That Kant holds such a substance-based theory of time is clear from the paragraph at A189, B232, p. 217, of the First Analogy (beginning “Substances in the field of appearances .l.l.”). Each substance is a substratum of time determination. He does believe that the time series relative to various substances must also be relatable to one another, but this is a further matter. 23. Schopenhauer considers such cases to be coincident successions (as when I first leave a house and then a tile falls on my head). He thinks that they are therefore not causal but are nevertheless objective, thereby refuting Kant’s Second Analogy. For Schopenhauer’s objection as well as a reply, see Kemp Smith, Commentary, 378–79.

Proofs of Substance and Causation

115

is thus represented within a necessary advance of time order. Note that it is still a representation of proper reactions necessarily advancing. Despite the fact that substance is now invoked, there isn’t for Kant a second series of states of substance outside of proper reactions.24 The proper reactions themselves (as per the First Analogy) are the determinations of the substance. Thus, the Lovejoy-Strawson25 charge of a non sequitur is still bogus, even though, by now, Kant is not a phenomenalist.

8 . t h e r e l at i o n o f c au s at i o n a n d s u b s ta n c e t o t h e t r a n s c e n d e n ta l d e d u c t i o n Kant’s answer to Hume, then, is not only that the universal applicability of the causal tie or bond is required for representing proper reactions in time (for representing objective succession), but that substances, which are enduring realities, not proper perceptions, are required as well. Not only isn’t causation or necessary connection a “fiction” of the (empirical) imagination, but substance (or identity through time of reality) is not such a fiction either. Indeed, these are two concepts that must be applicable if possible appearances (i.e., proper reactions) are to be represented objectively in time. Kant says in a famous sentence that “The a priori conditions of a possible experience in general are at the same time conditions of the possibility of objects of experience” (A11, p. 138). This I suggest is neither a trivial claim nor a mere reminder of Kant’s Copernican revolution. Rather, it is a statement of Kant’s refutation of phenomenalism. “Possible experience” signifies all possible perceptions (that is, all proper reactions), which includes not just further perceptions we can locally and presently have but perceptions remote in past time and far away in space.26 What Kant is saying is that although the purport of representation or cognition is just proper reaction in the full scope of space and time, this cognition requires objects of experience (indeed, substances). In other words, phenomenalism is incoherent, since in order to represent the full scope of possible perception, substantial entities are necessary. Indeed, we have seen in 24. This is shown, I believe, by what Kant holds about the series of appearances in past time in the First Antinomy. Objective states of substances would have to form a finite or an infinite series, whereas the propriety of reacting can be limitless without being a finite or infinite totality. See, for example, A495, B523, p. 442. 25. See footnotes 8 and 9 above. 26. In the paragraph at A110, p. 138, Kant has said that there is one single experience that encompasses all (proper) perception, just as “there is only one space and time in which all modes of appearance .l.l. occur.” It is this single experience (viz., the entire spatio-temporal scope of appearances) that is signified by “a possible experience in general” in the sentence I have quoted.

116

the principles

our account of the First Analogy to represent a proper past reaction as in past extensive time requires representing it as a reaction at an initial stage of a procedure of tracking what is present (which procedure we are now in the course of rather than at the initial stage of). The proper reaction, that is, has to be represented as a reaction proper to what is presently real, only not now, but formerly. What Kant is saying then in this sentence is that concepts of objects (the relational categories) are necessary conditions of representing possible experience (the full propriety of reacting). By our account, the concepts both of substance and of causation are required to bring all proper reactions (appearances) to the unity of apperception. Apperception, for Kant (A119, p. 143), is the faculty of understanding (cognition via concepts, or thought), and the understanding is the faculty of rules (A126, p. 147; A118, pp. 142–43). To bring remote proper reactions to my present apperception, then, is to have rules that encompass those reactions. This, in turn, is by having rules that encompass the spatio-temporal manifold that in turn encompasses all proper reactions. This manifold is a matter of constructions by the productive imagination (B155, p. 167)27 involving essentially “motion as an act of the subject” (B155, p. 167). The rules then are rules for spatiotemporal constructions or procedures, within which reactions are proper. This much, I claim, constitutes Kant’s account of what cognition is.28 Any concepts that are required for bringing the full spatio-temporal manifold to present rules will necessarily apply to proper reactions (appearances), for only by these concepts are those proper reactions anything to my present cognition. Kant, indeed, characterizes the categories just this way, when he says that they are concepts “which contain the necessary unity of the pure synthesis of imagination in respect of all possible appearances” (A119, p. 143). The necessary unity, I suggest, is the unity of a rule, which is how it is proper to construct and react. The Transcendental Deduction, I believe, is a recipe or method for establishing the applicability of pure concepts to cognizable reality. The method is to show that such concepts effect (are required for) cognition. The heart of the Deduction is not to prove any categories but to give the account of cognition, in relation to which, then, the categories 27. Kant characterizes the imagination as “the faculty of representing in intuition an object that is not itself present” (B151, p. 165). If so, then pure imagination must be a faculty of representing in pure intuition what is not present or what is remote. This is exactly our contention that the spatio-temporal manifold is required not basically for organizing what is presently given but more fundamentally for representing what is not present at all (remote appearances). 28. For a more detailed discussion of his theory of cognition, see Essays 3 and 4 in this volume.

Proofs of Substance and Causation

117

might be required. Kant hearkens back to this same account in his summary of the Analogies. He says, “Our analogies therefore really portray the unity of nature in the connection of all appearances [that is, all possible or all purportedly proper reactions] under certain exponents [i.e., the categories] which express nothing save the relation of time [in so far as time comprehends all existence] to the unity of apperception [to my present cognitive ability to represent rules]—such unity being possible only in synthesis according to rules” (A216, B263, p. 237). By our account, the concept of substance is required to bring a pure synthesis of imagination (viz., time-extensivity in the past), and so proper reactions (past appearances) to rules. As we have said, substance makes possible rules for being in the course of temporizing, and so being in the course of temporizing procedures that begin with a proper reaction. This proof of substance, then, in the First Analogy exactly fits the recipe for proving a category outlined in the Transcendental Deduction. Likewise, causation makes possible rules for being in the course of a series of successive reactions (while tracking) which reactions necessarily advance to one another, and so it is via causation that the necessary advance of the manifold of time (the time series) is brought to present rule. If I am on the right track, then neither the proof of substance nor the proof of causation makes any sense at all, unless one starts with the theory of cognition set out in the Transcendental Deduction, according to which all cognition is a matter of presently representing myself as variously “situated” with respect to proper reactions. This representation can be, for example, that I am situated as “too late” for a proper reaction (as in being at the tail end of a procedure that begins with the reaction), or it can be that I am situated too far from the reaction (as in being at the beginning of a spatial procedure that ends with the reaction), etc. In each case, my representation of my situation with respect to proper reactions is via presently operative rules for spatial and temporal constructions. Our reconstruction of the two Analogies has turned entirely on this account of cognition. We note finally that with respect to substance Kant is not deducing that there is trackable reality (which is an empirical factor). Rather, he is legitimating our right to think of that which is trackable (if such there be) as one enduring entity, as opposed, say, to a succession of continuously successive replacements. Because this latter thought of one enduring entity is not a thought of any further feature reality may present or not, thinking it ( justifiably because of its necessity for cognition) is no different from its “really” pertaining or being so. It cannot fail to pertain for failing to pick out what reality is like. Similarly, with

118

the principles

respect to causation Kant is not deducing that there is orderly, regular, constant reality, which is an empirical feature. Rather, he is legitimating our right to think of regularity (if such there be) as regularity in (or of) a necessary succession. Thus if smoke regularly follows fire we are entitled to think of it as regularly emerging from fire. Because this latter thought does not pertain to a feature that events present, thinking it pertains ( justifiably so because of its requirement for cognizing the unfolding of time) is no different from its “really” pertaining. It cannot fail to pertain for (regular) reality failing to incorporate some further feature. In this way Stroud’s objections to the force of transcendental arguments29 fail, since for pure a priori concepts there is no such thing as their application to reality failing because reality lacks some feature that they supposedly depict. There is no difference, that is, between having to believe they apply (having to apply them for cognition to be possible) and their “really” applying. In sum, Kant’s proof of causation, in our account, turns fundamentally on a partial causal theory of time. This account makes the Second Analogy basically one single argument where objective succession is first distinguished from objective coexistence and then placed in a time series where the earlier necessarily advances to the later. It answers Hume both as to the nature of the causal tie or nexus (the necessary connection), and as to the universality of the causal principle. It is consistent with (indeed depends on) Kant’s theory of time as a pure intuition (something immediately presentable by construction) in the Aesthetic, and it coheres with (indeed depends upon) Kant’s theory of cognition in the Deduction (bringing all possible appearances to the unity of apperception, by that unity governing the transcendental synthesis of imagination). The account is also consistent with (indeed demands) a rejection of phenomenalism via the connection of causation to substance. Despite this rejection of phenomenalism, it is consistent with Kant’s transcendental idealism according to which space and time are mere forms of intuition (are exclusively constructions within which proper reactions arise), and according to which the categories (substance and causation), by being required for rules for spatio-temporal constructions, pertain only to possible appearances (proper reactions). Indeed the causal relation holds between proper reactions (the propriety of reacting one way necessarily advancing to the propriety of then 29. Barry Stroud, “Transcendental Arguments,” in Kant on Pure Reason, ed. R. C. S. Walker (Oxford: Oxford University Press, 1982). Although it is true that there are empirical elements (trackability, regularity) that are necessary for cognition, it is also true that Kant is not deducing these elements. For an interesting discussion of such empirical elements of cognition, see Kenneth R. Westphal, “Affinity, Idealism, and Naturalism: The Stability of Cinnabar and the Possibility of Experience,” in Kant-Studien 88, no. 2 (1997).

Proofs of Substance and Causation

119

acting another way), and states of substances are proper reactions (had in the course of keeping track of what affects).

9 . m o d e r n d e v e l o p m e n t s i n c au s at i o n The modern version of a causal theory of time is developed in great detail by Reichenbach.30 Unlike Kant’s, Reichenbach’s is a total causal theory where the causal relation fully constitutes time order itself, not any necessary advance of an already given order. He can thus use a probabilistic conception of causation as long as he can recover the order and direction of time from it. For Kant, by contrast, the exact role of the causal relation is to constitute the earlier time as necessarily advancing to or determining the later time, and hence causation must be deterministic. The “direction” of the causal relation (that what causes or produces precedes what is effected or produced) derives as well from this same exact role. Since the earlier time determines or yields the later, the relation that constitutes this must likewise be asymmetric. In sum, the relation between events whose role is to constitute the necessary advance of time must be both an asymmetric and deterministic one, the determining cause prior to the effect. Mackie is exactly wrong, then, when he says regarding Kant, “Surprisingly, in view of the importance which it would appear to have for his thesis that objective time order depends upon causation, Kant has little to say about causal priority [the direction of causation].”31 Although the priority of cause to effect follows from Kant’s view, it does so only to the extent that the linear advance of time is itself necessary. The necessary advance of such linear time order can be regarded simply as the existence of that order in a set of possible worlds. This family of possible worlds, for Kant at least, is characterizable as constituting the “real” possibilities. Kant’s view on the direction of causation doesn’t apply outside this family of worlds. Thus where time is cyclic or branching, a relation between events that constitutes or represent this order would likewise be cyclic or branching, violating respectively the direction and deterministic nature of causation.32 Kant’s view depends essentially on the idea that time “advances,” 30. Hans Reichenbach, The Direction of Time (Berkeley and Los Angeles: University of California Press, 1956). 31. J. L. Mackie, The Cement of the Universe (Oxford: Clarendon Press, 1974), 90. 32. This is just to say that for Kant “backwards causation” and nondeterministic causation are logically possible. General relativity, though perhaps allowing for cyclic time, doesn’t allow for branching time. Since locally everywhere the earlier necessarily advances to or determines the later, the relation between events constituting such advance would still be universal deterministic causation.

120

the principles

that the present “emerges” out of the past, etc. These features belong to what McTaggart33 called the A-series, and are features of time that many philosophers find troublesome. Mellor, for example, says, “I shall therefore ignore all accounts of causation which .l.l. involve time’s flow, e.g., by using the way it ‘fixes’ events as they become present to say how causes fix their effects .l.l. for .l.l. time does not flow.”34 Now, in Kant’s view it is not how time fixes events as they become present, but how preceding time fixes present time that is used to say how causes fix their effects. Clearly, though, Kant’s is an account of causation that involves time’s flow. I cannot go into a general discussion of McTaggart’s arguments against time passage, but I will note that the arguments are framed in terms of objective time and seem not to apply to Kant’s constructivist account of time. The reason is that in Kant’s account before and after (viz., McTaggart’s B-series) are not “constructible” apart from my presently being up to a certain stage in temporizing (the “cut” between the past and the present that belongs to McTaggart’s A-series). Since the B-series exists in construction only as dependent upon and fixed in terms of the A-series, McTaggart’s argument, which depends in effect upon an independent B-series, is blocked. It is nevertheless true that Kant’s entire theory of causation, not just the direction of causation as in Mackie’s account,35 depends on time’s “passage” or “flow.” What I wish to argue next is that Kant’s view can incorporate or contain the view defended by Fair and Salmon36 that causation in the actual world is just transference of energy or momentum in processes or interactions. We note that Kant’s causal theory of (the necessary advance of) time constitutes a functional analysis of the notion of causation. Causation is not taken as a primitive notion and then employed to constitute time order. Rather, causation itself is defined by its role or function of being that connection of events that represents or constitutes the necessary advance of time. The fact that the cause produces or determines the effect or that the effect emerges or derives from the cause is a consequence of that role or function. Since the bond derives from the role or function, it holds between events in so far as these events play this role or serve this function. This leaves open what event pairs in fact play this role in the actual world, and whether those same pairs 33. J. M. E. McTaggart, The Nature of Existence (Cambridge: Cambridge University Press, 1927), book 5, chap. 33. 34. D. H. Mellor, The Facts of Causation (London: Routledge, 1995). 35. Mackie, The Cement of the Universe, chap. 7. 36. David Fair, “Causation and the Flow of Energy,” Erkenntnis 14 (1979); Wesley Salmon, “Causality, Production, and Propagation,” in Causation, ed. Ernest Sosa and Michael Tooley (New York: Oxford University Press, 1993).

Proofs of Substance and Causation

121

play it in all (really) possible worlds.37 Thus it may be that in the actual world causation is always transference of energy or momentum. However, in Kant’s view such transference is not definitive of causation per se but simply characterizes what actually plays the role. It is the role itself that defines causation. In the Metaphysical First Principles of Natural Science Kant himself allows for a universal but empirical applicability of matter in motion as the causal connection in the actual world. The separation between this and the Critique’s account of causation is just the separation between the essential nature of causation per se and what causes what in actuality. The contemporary view38 of probable causation, or causation without determination, is of course incompatible with Kant’s account. Anscombe contends that causation goes with notions such as “derives from,” “comes from,” “arises out of,” and that these notions do not involve necessitation.39 In Kant’s theory, however, they do involve necessitation, since the source of these notions is the advance of time order which is also a necessitating advance. Anscombe gives as an example of such non-necessitating causation a bomb connected to a Geiger counter. She says, “There would be no doubt of the cause of the reading or of the explosion if the bomb did go off.”40 But exactly what does she think the cause of the reading is? The particle having left the nucleus? There is some reason to hold that until the reading there is no event going on that leads to the reading. In other words there is no preceding cause of the reading. This is just the standard understanding that in quantum mechanics there aren’t well-defined events that occur to produce the probable outcome. Mellor41 gives the example of enough fissionable material causing the explosion, although it only makes it extremely probable. It seems to me that enough fissionable material precisely doesn’t make the explosion happen; it only makes it extremely probable. Hence, at best, it causes it to be highly probable for there to be an explosion, but not the explosion (the event) itself. Even if there is no such thing as probabilistic causation, the lack of causation in quantum mechanics should, in Kant’s view, imply the failure of the necessary ad37. A necessary connection between cause and effect deriving from its role leaves open what causes what, just as an “authority connection” between sergeant and private leaves open who (which person) has authority over whom. 38. G. E. M. Anscombe, “Causality and Determination,” in Sosa and Tooley, Causation, 88–104. See also Mellor, The Facts of Causation. 39. Anscombe, “Causality and Determination,” 91–92. 40. Ibid., 101. 41. Mellor, The Facts of Causation, chap. 5. Again, as in the Anscombe example, no exact events go on between stages of the explosion process and so there is no causal chain from the sufficient material to stages and from stages to other stages.

122

the principles

vance of time order. The many-worlds interpretation of quantum mechanics comes closest to such a view if we think of the branching of the worlds as also being a branching of time.42 The contemporary regularity theory according to which singular causation is simply a matter of falling under nomic generalizations has its source in Hume. Mackie goes so far as to attribute such a view to Kant. According to Mackie, Kant “has nothing to say about any intimate tie between an individual cause and its effect.”43 Now, Kant does hold that cases of causation always come under universal laws. He holds this by deriving both the intimate tie between a cause and its effect and the universality of causes (same cause-same effect) from the role of causation in representing the necessary advance of time. As we have seen, that advance is not only necessary (determining, yielding) from moment to moment, but the necessity itself has the same character universally for all transitions in time. The upshot, for Kant, is a singular causal tie that is also universally generalizable. The singular tie is not defined in terms of a nomic law, for the law itself is a law of the universality of the singular tie. Thus the law is to the effect that always an event of a certain type yields or produces or necessitates an event of a second type. If anything, nomic lawfulness, as opposed to “accidental” universality, is to be defined in terms of the singular bond of causation rather than vice versa. Kant’s account of causation, as opposed to contemporary accounts, has all the advantages and disadvantages of being an account embedded in a more general metaphysical theory. The advantage, I believe, is obvious. Kant is able to derive (explain) the singular bond between cause and effect, the universality of particular causal connections (same cause-same effect), and the universal law of causation (every event has a cause) all from the role causation has in his account of the representation of reality. No contemporary theory, I believe, comes close to such a derivation of any of these three matters. The disadvantage is equally obvious. One must accept Kant’s specifically constructivist view of time and the features he attributes to time (necessary advance) or his whole account collapses.

10. modern developments regarding substance The most important contemporary view on the nature of substance derives from Quine’s account of reference. First of all, in Quine’s ac42. For Kant, however, such branching is not “really” possible since there are no constructions for going off into incompatible future times. 43. Mackie, The Cement of the Universe, 90.

Proofs of Substance and Causation

123

count reference to substances is a “posit” that is underdetermined by experience or by the data.44 As such, there is no “flat-out” truth to the existence of substances in the world, but only truth relative to some theory, decision, language, etc. Second, for Quine, apart from analytical hypotheses (which do not derive from anything about a person) there is no sense to the idea of a person’s referring to or using the concept of a substance.45 Hence there is no “flat-out” truth to a person’s having the concept of a substance, but only truth relative to a choice of analytical hypotheses. All Quine finds to the idea of substance is our use of the individuative apparatus of our language consisting of “plural endings, pronouns, numerals, the ‘is’ of identity and its adaptations ‘same’ and ‘other.’”46 It is this apparatus which is undetermined as to its truth and inscrutable as to its employment in cognition. This individuative apparatus essentially corresponds to Kant’s logical function of subject-predicate in being the “judgmental” or “intellectual” component of the notion of substance. Quine is surely right as against Kant that various aspects and components of judgment (not just subject-predicate) go with reference to substances. Quine looks for an empirical basis for the use of the individuative apparatus and he finds none. In Kant’s terms, the trouble is that the apparatus and the empirical are “heterogeneous,” requiring the intermediation of a schema. Without schemata the functions of judgment for Kant are empty or represent no object (A147, B187, p. 187). The schemata give both significance and applicability to the judgmental apparatus (A146, B186, p. 186). The schemata, last, are determinations of time. What Quine is lacking, then, that keeps him from finding significance and applicability for the individuative apparatus is its use in the determination of time. Let us see how this works for the schema of substance. The schema of substance for Kant is the permanence of the real functioning to represent the permanence of time; viz., to represent the fact that present time is a continuation of ongoing past time. As we have seen in a constructivist view of time, this representation involves being in the middle of a temporizing construction, which in turn requires that something present put us beyond the beginning of the construction or procedure. But to say that it is proper to be at the midstage of a procedure tracking what is now present is eo ipso to say that 44. See W. V. Quine, Word and Object (Cambridge: MIT Press, 1960), 22. 45. Ibid., 73. 46. W. V. Quine, “Ontological Relativity,” in Ontological Relativity and Other Essays (New York: Columbia University Press, 1969), 33.

124

the principles

what is now present is a substance (viz., something that has a past or that was), since a temporally extended procedure is proper with respect to it (the very reality that is present). So, in particular, I cannot be in the middle of tracking a present “slice” or “stage.” Since such a slice is a momentary existent it cannot be proper to be so far along in keeping track of it. Nor can I presently be in the course of tracking a series of such slices or stages the last of which is now present, because this would require there being past stages (stages existing in the past) which only subsequently are represented by being in the course of a temporizing procedure. But then there would be pastness pertaining to things antecedent to being in the course of temporizing procedures with regard to them, which would take us beyond constructivist time (the only time Kant allows). That there are substances as opposed to slices may be underdetermined by the empirical data, but it is not underdetermined by the necessity of the existence of past time. The pure functional role of substance in determining time to have a past (on a constructivist construal of time) ensures that there are substances, not slices. It ensures, that is, the “flat-out” truth of the existence of substances in the world. A person has a concept or a thought of what is present to him as being a substance if he thinks of himself as properly being up to a midstage in the procedure of tracking it. The concept of substance is “scrutable,” then, to the extent that there is evidence for thoughts of being at mid-stages of procedures. Suppose now that a person who has a procedure for baking a cake in ordered steps when he starts from scratch on his own comes into a room and sees the first three ingredients mixed in. Instead of putting in the fourth ingredient right away, suppose he first quickly goes through the motions of putting in the first three ingredients. Then he shows that he thinks of himself as being up to putting in the fourth ingredient in a procedure that begins with the first three. Similarly, suppose a person first quickly goes through the motions of timing to k and keeping track of what is present, before he tracks while marking time at slow pace from k + 1. The person shows that he thinks of himself as being up to k in tracking and keeping time (at slow pace) with what is present. Hence he shows that he thinks of what is present as a substance. In this way there is a “flat-out” truth (apart from analytical hypotheses) to a person’s having the concept of a substance.47 47. I deal with the “scrutability” of substance in detail in Representation of the World: A Naturalized Semantics (New York: Peter Lang, 1996).

7

Kant’s Refutation of Idealism in the B Edition

The Refutation of Idealism in the B edition attempts to show that if I know that I am now in a conscious state, then I must also be immediately aware of objects in space outside me. The argument proceeds in two steps, each of which concerns the reality of time. First, to know that I am now in a conscious state entails that the present time exists, and the present can only exist as emerging from the past. Second, the reality of the past entails the existence of objects outside me that are intuitable, or that I can be immediately aware of. Kant argues for the second step in the First Analogy.1 In part 1 I present this argument that the past presupposes something permanent in objects of perception. In part 2 I try to explain why this permanent must be an intuitable and a spatial thing. Much of the complexity of the Refutation is in these issues. In part 3 I defend the above two-step reconstruction of the argument of the Refutation. Finally, in part 4 I take up two issues that Kant raises in the notes to the Refutation: how it is that we can distinguish delusions and hallucinations from veridical perception, and how it can be made understandable that we are immediately conscious (of what is) outside ourselves.

1. the argument for permanence The proof of the First Analogy added in the B edition (B225, p. 213)2 begins with the claim that time as a form of intuition is permanent. Kant is presuming the result of the Aesthetic that time is a mere form of intuition. In the Aesthetic he contends that space and time are 1. For the unorthodox view that the Refutation does not incorporate the First Analogy, see Jonathan Bennett, Kant’s Analytic (Cambridge: Cambridge University Press, 1966), 202. For a convincing response to Bennett, see Henry Allison, Kant’s Transcendental Idealism (New Haven: Yale University Press, 1983), chap. 14. 2. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition. My discussion of the First Analogy is limited to this one paragraph.

125

126

the principles

given in pure intuition. This means that they are immediately represented by being present (i.e., they are intuited), and are so without sensory affection by objects (i.e., purely intuited). Kant holds, I contend, that space and time can be presented without affection by performing constructions, such as sweeping out a line, which constructions present expanses. Kant is a constructivist regarding the existence of space; i.e., space exists only in constructions, somewhat as for a constructivist regarding numbers, numbers exist only as termini of counting procedures.3 His argument for this is that space (any expanse of space) is continuous, and continuity can pertain only to flowing constructions such as sweeping out a line.4 A temporal construction would be something like a series of downbeats a conductor performs to temper the orchestra’s playing or to mark time for the orchestra. Each downbeat is a flowing construction, which is the construction of a temporal interval or expanse. That this pure intuition of time is also a form of empirical intuition means that what exists in time does so by arising at a stage of temporal construction. When Kant says, then, in the First Analogy that time as a form of intuition must be permanent, he is talking about constructive time or temporal construction. To say that constructive time must be permanent means that a temporizing construction that I now begin is not the beginning of time but a continuation of already ongoing time, and hence (since time is constructive) a continuation, in some manner, of construction. Now, in the Transcendental Deduction the construction of time is attributed to the productive synthesis of the imagination (A118, pp. 142–43), which involves “motion as an act of the subject” (B154, p. 167). This synthesis gets its unity from the unity of apperception (A118, p. 143). But apperception, for Kant, is the (faculty of) understanding (A119, p. 143) and the understanding is the faculty of rules (A126, p. 147). Hence the unity of apperception is the unity of a rule, and thus our representation of time is not in terms of actually constructing, but rather in terms of rules for proper temporal construction. But now I wish to suggest that not only can it be proper for me to begin a temporal procedure, it can also 3. Space and time for Kant, then, are not filters, “green glasses,” or cognitive maps, any more than numbers are for the mathematical constructivist. They are activities or constructions of the subject. 4. In such a flowing construction the whole is prior to the parts (See A25, B39, p. 69). The flow is not composed of a series of cuts but is a seamless construction. Any segmenting into parts occurs by subsequently imposing cuts on the line. This seamlessness (that the whole is not put together out of constituents) is Kant’s understanding of continuity. By contrast, an objective whole (an objective spatial region—whether relational or substantival) would be such that it exists or is all there only if all its parts exist, and so the whole would not be prior to the parts. Thus it would be a whole composed of constituents, and so not continuous.

Refutation of Idealism in the B Edition

127

be proper for me to now be in the course of a temporizing procedure (i.e., to now be past or beyond earlier stages of the procedure). It is only in this manner that constructive time can be “permanent” (can pertain to the past). Whatever is involved or entailed by the propriety of being in the course of a temporizing procedure is thus required for representing past time (indeed for there being past time) in a constructivist account of time. We need to see, then, what is involved in the propriety of being in the course of a procedure. Suppose I have a procedure or a rule for baking a cake by putting ingredients into a bowl according to a fixed order. It cannot simply now be proper to be up to the fourth stage of the procedure or up to putting in the fourth ingredient. After all, a procedure is just a way of doing things in a fixed order (first things first). But now suppose I have this procedure and I walk into a room where I find or detect a bowl with the first three ingredients in it. What I presently detect then makes it proper to be up to the fourth stage of the procedure. It does so because the cake-baking procedure is geared to something (getting a cake baked), and what I presently detect is such that to keep up with what the procedure is geared to requires me to be up to the fourth stage. Similarly, if it is proper to be in the course of a temporizing procedure (such as producing a series of downbeats accompanied by a recitation of ordinals “first,” “second,” etc.), this procedure must be geared to something (getting something time-marked), so that what I presently detect requires me to be up to a certain stage in order to keep up with what the procedure is geared to. Now, as Kant says, “time cannot by itself be perceived” (B225, p. 213). Temporizing, that is, cannot be geared to, and so cannot keep up with, objective or extra-constructive time.5 It is not, that is, that objective time is presently so far along that keeping up with it requires me to be so far along in a temporizing procedure that is geared to it, or that marks it. Hence, Kant says, the “substratum” (the basis) of representing constructive time “must be found in the objects of perception” (B225, p. 213). Something real in my present circumstances, that is, must be so far along in its existence and the temporizing procedure must be geared to that existence. Now the real, for Kant, is what corresponds to sensation and gives me the consciousness that I am affected (A166, p. 201; also B207, p. 201). The real, then, is what affects me. The continued existence of the real is continued affection through or while keeping track. If the temporizing 5. The reason, quite simply, is that objective or extra-constructive time does not exist. Time is continuous (expansive) and only constructive time can be continuous. See footnote 4.

128

the principles

procedure is geared to tracking or keeping track, then what presently affects me may be so far along in its existence (trackability) that keeping up with it by a temporizing-cum-tracking procedure requires me to be now so far along in, and hence in the course of, the procedure. We can represent this as follows: (1) Upon being presently affected, it is proper to be up to stage k in temporizingcum-tracking. But thinking this rule or propriety in regard to presently affecting reality is eo ipso thinking that what presently affects has duration, or is “permanent,” at least through the stages of temporizing. Thus it is to think the permanence of the real that, for Kant, is substance (B183, p. 184). Thus the substratum or basis of representing past time is substance. Kant next says that “all that belongs to existence can be thought only as determination of substance” (B225, p. 213). What belongs to existence, for Kant, are appearances. Appearances are possible perceptions that we may take to be proper reactions or responses. Any reaction proper in the past must be represented in past time, and hence via being connected to the procedure in (1) above that represents past time. For example, if the procedure is augmented to one of temporizingcum-tracking together with first reacting “red” (at the beginning of the first downbeat as one is tracked onto reality), then we can extend (1) to: (2) Upon being presently affected, it is proper to be up to stage k in temporizingcum-tracking from first reacting “red.” Thus the proper reaction (or possible appearance) is represented as proper in past time via its being represented as proper with respect to (and so as a determination of) the real that presently affects me (that is, substance). In this manner, possible appearances are brought to the unity of apperception (the unity of a present rule) by the category of substance applying to them (that is, by representing them in regard to the permanence of the real). This is as much of the argument of the First Analogy as is required for the Refutation of Idealism. The crux of the argument is simply that past time (past constructive time, which is the only time there is for Kant) can only be represented (indeed, can only exist) via something permanent that is presently perceived (that presently affects me).

Refutation of Idealism in the B Edition

129

2 . s pat i a l i t y a n d i n t u i ta b i l i t y of the permanent In our reconstruction of the First Analogy we accepted Kant’s inference that since the basis or substratum of being in the course of temporizing cannot be objective time itself, therefore it must be some intuitable reality (some present object of perception) and, further, a spatial reality. We discuss these components of his conclusion in turn, since they are each crucial to the Refutation. The inference to intuitability is essentially the idea that the basis cannot be some object that is purely conceptually represented or merely represented in thought. Kant doesn’t explicitly deal with this in the First Analogy, since the rejection of purely conceptual representation is a thread running throughout the Critique. Nevertheless, since eliminating this alternative of a purely conceptualized permanent as basis is crucial to the Refutation, I shall try to outline now why the basis has to be given in perception (indeed, immediately given in perception). Of course it may be that my gearing a temporizing procedure to something that it keeps up with might require that I be acquainted with that thing, in some sense other than mere conceptualization. However, couldn’t it be that a temporizing procedure in fact keeps up with something, whether I gear it to that thing or not? To make the hypothesis a little more complete, suppose we accept a representative theory of perception, according to which I am only directly aware (I only intuit) what goes on in my own states. The object outside me that causes such a state is not “directly” perceived, but only conceptualized as the entity that causes my state. Why shouldn’t that object be the basis of being in the course of temporizing? Equivalently, why shouldn’t the unperceived cause of my present mental state be what a temporizing procedure keeps up with? For Kant, that space and time are mere forms of sensible intuition means, in part, that only what affects us immediately can have spatial or temporal position or characteristics (A404, B522, p. 441; B69, p. 88; A26, B42, p. 71). What is “in” constructive space and time, that is, is what is presented in the course of constructions. In a spatial construction, such as drawing a line, my attention shifts or sweeps out linearly. If I am thereupon affected, that “posits” (A26, B42, p. 71) thus being affected as occurring at the limit of that constructed spatial sweep or expanse. If at the end I merely conceptualize or think of an entity that isn’t itself presented, that entity doesn’t attach to that point of space (the end stage of construction) by being something that happens to me (being affected) at that stage. Hence it must attach or pertain to the con-

130

the principles

struction in some other way. The only other way, apparently, is that it has some feature or characteristic belonging to it by which it goes with or matches stages of construction. Indeed, suppose I keep track in drawing out my attention linearly. On the representative theory of perception, what is encompassed within the linear construction (in the sense of what happens to me throughout its course) is internal (perhaps constant) sensation, not the external entity that produces it. If the construction is to pertain to the external reality, it must then be by some sort of matching or overlap. The entity, that is, has some sort of existence that matches the construction. But this is just to say that the entity has its own objective path through space, which my construction overlaps. The point is that space and time are activities. What happens to me at, or throughout, stages of these activities is thereby positioned with respect to the activity. On the other hand, what exists (what is represented conceptually, but doesn’t happen to me) can only be positioned with respect to activity, by having a relationship to it based partly on its own (spatial or temporal) matching existence. The entity, so to speak, stands outside or apart from the activity (the construction) and so is not encompassed by it, but merely relates to it externally. In sum, an object that is not intuited (an object that is merely conceptualized) would have to be thought of as existing in extra-constructive space and time for spatio-temporal construction to pertain to it. But for Kant there cannot be extra-constructive expansiveness (continuity), either spatial or temporal, that matches or externally relates to construction. Hence, spatio-temporal construction cannot pertain to merely conceptualized (unintuited) objects at all.6 As an analogy, suppose one is a constructivist regarding sequential order. By this I mean that one holds that a collection of objects on a desk, say, has no intrinsic “first” object, “second” object, etc. to it. Only when I “impose” an ordered construction (such as reciting in order “first,” “second,” etc.), do the objects obtain any sequential order. If the objects are given in perception, so that I point to them as I construct, they are thereby ordered. On the other hand, if I merely think entities or objects (without presenting them), my thoughts may have a certain 6. The considerations I have just rehearsed constitute the transition Kant makes from space and time being pure intuitions to their not representing “any determination that attaches to the objects in themselves .l.l. when abstraction has been made of the subjective conditions of intuition” (A26, B42, p. 71). This is just the transition from space and time being pure intuitions to thereby being mere forms of empirical intuition (rather than pertaining to things apart from intuiting them). The Metaphysical Exposition establishes that space and time are pure intuitions (constructions), and then Kant concludes that therefore they apply only to what can be intuited. The transition, then, is not basically from space and time being a priori to their being mere forms of empirical intuiting.

Refutation of Idealism in the B Edition

131

order, but the entities that I think of are yet to be encompassed by the construction that imposes order. Since I can earlier think of a later entity and later think of an earlier entity, it is not enough that my thoughts have a certain order, they must be thoughts of (the entities’ having) order for the thinking to pertain to the construction. The only way the collection on the desk then could pertain to the order of the construction would be, per absurdum, by thinking of those entities as having their own intrinsic matching order. Indeed, my contention is that for Kant the feature of continuity (seamlessness, expansiveness) has very much the character of sequential order in this analogy; viz., it obtains only in construction and therefore only to what is given within the construction, not to what would have to be thought of as matching it extra-constructively. In regard to our case of being in the course of temporizing, the temporizing procedure itself could only pertain to an entity that is purely conceptualized (that is not immediately given or intuited) only if that entity had its own matching duration (since the entity is not “encompassed” within the temporizing). Thus the temporizing procedure could only keep up with the entity’s corresponding (extra-constructive) duration, and so the propriety of being in the course of temporizing (in order to keep up with what the procedure pertains to) would be a matter of keeping up with the extra-constructive duration of the entity. But there is no extra-constructive duration (seamless expanse), and so the basis of being in the course of temporizing cannot be some nonintuited or purely conceptualized object. In particular, in regard to the representative theory of perception, the basis cannot be the nonintuited entity that is thought as cause of what I perceive in me. On the other hand, what affects (that is, an immediately intuitable reality, or an object of perception) is encompassed by the temporizing, for the tracking is what I do and the affection is what happens throughout the downbeats. The affection doesn’t have a second matching duration; its duration is just its governance by the temporizing procedure. Another example of an unintuited entity would be the subject of consciousness itself. For Kant, as for Descartes, the subject is not directly sensibly intuited or perceived. Indeed, Descartes insists in the second meditation that the “I” cannot be pictured at all (that it is non-sensory altogether). The subject, rather, is merely thought of or purely conceptually represented. Kant says in Note 2 to the Refutation that the consciousness of a thinker is a “merely intellectual representation” that, as such, cannot be a permanent that “might serve as correlate for the determination of time in inner sense.” Now the same point that arose against an inferred thing distinct from me as a basis of being in the

132

the principles

course of temporizing pertains to my being such a basis. Namely, as a merely intellectually represented entity, it could be such a basis only by having a matching objective (in the sense of extra-constructive) duration, but there is no such duration. It should be clear why Kant does not specifically deal with the supposed alternative of a nonintuited (purely intellectually represented) basis or substratum in the First Analogy. The impossibility of it follows directly from his central contention that space and time are mere forms of intuition; i.e., the constructions they are pertain only to what can be intuited or perceived (what can happen within the construction), not to things independent of being intuited. This contention is just the transcendental ideality of space and time.7 We turn now to the second aspect of Kant’s conclusion in the First Analogy, that the basis must be spatial. Why couldn’t the basis, for example, be some constant or lasting internal (nonspatial) representation such as an ongoing feeling of anguish or an ongoing hearing of a note or sound? Kant says in the General Note on the System of Principles that “in order to obtain something permanent in intuition .l.l. we require an intuition in space.l.l.l. For space alone is determined as permanent, while time, and therefore everything that is in inner sense, is in constant flux” (B291, pp. 254–55). It may be that time is in constant flux, always progressing to later time so that earlier times pass away, but why does this imply that everything in inner sense is in constant flux? Why couldn’t a feeling of anguish be enduring, so that it is presently proper to be in the course of a temporizing procedure geared to the feeling? One answer, I suggest, is that everything intuited in inner sense, whether qualitatively constant or variable, is a condition or state of the subject, not an entity. Now, conditions or states are aspects of the career or history of a thing, and the history of an entity (as opposed to the entity) has temporal parts or segments. The history, that is, is a temporal whole or series, whether or not it involves qualitative variation. Likewise, then, any condition or state that is part of the history is a temporal whole or series of parts, and so in this regard is in constant flux (viz., earlier stages of the condition give way to later stages). The same is true, by the way, in regard to outer sense. Thus an entity may remain red or heavy or whatever, but its condition of being thus red or heavy is 7. In the first-edition Refutation of Idealism, Kant constantly equates empirical idealism with transcendental realism. The equation is as follows. The idealist holds a representative theory of perception and so holds objects (the unintuited causes of perception) to be merely conceptualized, and so must hold that space is not a mere form of intuiting things but applies to things even apart from intuition (viz., the objects external to our perceptual state would have to be in objective space).

Refutation of Idealism in the B Edition

133

part of its history and so, like its history, is a temporal whole or series. On the other hand, the thing that is red or heavy (the thing that has the history) is not a temporal series with parts or stages, but something that exists completely and entirely at each time of its existence; viz., it has strict numerical identity over time. Its history may increase, but its existence or being is full and complete at any time. Suppose now that the lasting anguish is a temporal whole composed of stages or parts. Then the anguish presently felt is just a present stage of anguish and so is momentary. But now a momentary stage is not something that a stretching-out temporizing procedure can pertain to or be geared to, and hence the present (stage of) anguish cannot be the basis of being in the course of such a procedure. Of course the whole temporal spread of stages of anguish that includes the present one can be subject to temporizing. However, this spread includes components or stages that are not present but exist only in the past (unlike a present entity, which exists completely in the present or at any time that it exists). To think of the whole spread, then, as a basis would require thinking that, with regard to a series of anguish stages, some of which exist in the past, it is proper to now be in the course of a temporizing procedure. The past parts or segments would have to be presently thought of as having past temporal existence independent of temporizing procedures, since it is the whole series, including past elements, that is thought of as the basis. Once again, this is impossible for Kant, since there is no extra-constructive temporal expansiveness. In sum, everything given in inner intuition, even if qualitatively constant, is a state or condition of the subject and so an aspect of its history. Hence it is a temporal whole or series and so is in constant flux even if its quality is constant. Thus what is given in inner sense cannot be the basis of being in the course of temporizing.8 Indeed, this argument implies not only that the basis must be given in outer intuition but that it must be a thing (entity), not an outer condition or state of a thing, that is the basis, for a qualitatively unvarying condition or state of a spatial thing is likewise in constant flux (viz., is a temporal whole or series only the present stage of which is presently intuitable). On the other hand, 8. It is because everything given in inner intuition is a state or condition of ourselves (whether qualitatively permanent or not) that the basis or substratum cannot exist in inner intuition. Since space is the only other form of intuition, and what is given in space need not be a state or condition (viz., it may be a thing or substance distinct from ourselves), only what is given in space could be a basis for past time. The argument I have presented for the spatiality of the permanent that can represent time has affinities with the argument given by Jill Vance Buroker. See “On Kant’s Proof of the Existence of Material Objects,” in Proceedings of the Sixth International Kant Congress, ed. Gerhard Funke and T. M. Seebohm (Washington, D.C.: University Press of America, 1989).

134

the principles

since a thing exists entirely at each moment of its existence, I presently intuit the thing (substance), not some present temporal stage of it. Hence its representation as having a past is exclusively that it is what makes it proper to be in the course of temporizing (not that it is a temporal whole independent of temporizing, that temporizing matches or corresponds to).9 It follows from our discussions regarding intuitability and then spatiality that the basis of being in the course of temporizing must be a thing outside me (a thing in space) that I intuit or am immediately (noninferentially) aware of. It cannot be my own subject or an unperceived thing distinct from me, since neither of these are intuitable. Nor can it be a state or condition of me, or a state or condition of a thing distinct from me, since these are in “constant flux” even if qualitatively unchanging. That leaves only things outside in space that I immediately intuit. Some commentators10 have thought that this conclusion, which Kant holds in the B edition Refutation of Idealism, is inconsistent with transcendental idealism. Note first that this conclusion does not imply that we can intuit substance (permanence) apart from thought (indeed, categorial thought). It is the thought that what affects (what I presently intuit) makes it proper to be in the course of temporizing that represents what affects as permanent. Kant’s contention that we are immediately aware of what is outside us, then, is not the contention that we can intuit objects without thought, but rather a rejection of a representative theory of perception.11 Second, this contention is consistent with categories applying only to appearances. The category of substance, indeed, is what enables us to represent possible appearances (possible reactions, or reactions proper at earlier times). Further, it is consistent with space and time being mere forms of intuition applying only to what can be intuited or given. It simply holds that these intuitable items are things outside us.12 Since the two basic tenets of transcendental ide9. In effect, our considerations show that in a Kantian view of time we must think of substances as distinct from temporal wholes or temporal series if we are to represent the past at all, since if substances were thought of instead as temporal wholes (like conditions or histories) they too couldn’t be the basis of being in the course of temporizing procedures. 10. See Paul Guyer, “Kant’s Intentions in the Refutation of Idealism,” Philosophical Review 92 ( July 1983): 329–83. 11. For an excellent discussion of this point, see Buroker, “On Kant’s Proof.” 12. Of course, in the first-edition Refutation of Idealism Kant holds that we can only intuit (be sentient of) what is “in us” so that “the whole of our self-consciousness therefore yields nothing save our own determinations” (A378, p. 350). What he has changed his mind about is not whether space and time apply only to what can be intuited, but rather the status of what can be intuited. It is now no longer “merely our own determinations” that we can be immediately aware of.

Refutation of Idealism in the B Edition

135

alism are just that the categories and space and time apply only to appearances, the immediate awareness of things outside us in space is consistent with transcendental idealism.13

3 . t h e r e f u tat i o n o f i d e a l i s m The goal of the Refutation is to show that “even our inner experience, which for Descartes is indubitable, is possible only on the assumption of outer experience” (B275, p. 244). Kant is not only setting out to prove the existence of objects outside us, but that we are immediately or noninferentially aware of such objects. He is to prove this, further, starting from a premise that Descartes would accept. That premise is characterized in the statement of the thesis as the “mere, but empirically determined, consciousness of my own existence” which, Kant adds, is a consciousness of it “as determined in time.” I suggest that Kant’s premise is simply that I think that I am presently in a conscious state, as when, for example, I think that I now feel pain or that I am now having a visual experience of red. The premise is not that I am presently conscious but that I think that I am. I believe Descartes would accept this, since he is trying to come to an incontrovertible belief (indeed to a clear and distinct perception) that seems to require a propositional attitude toward my conscious state (believing it, thinking it, etc.). The premise does not require a sequence of conscious states in time. The reason some commentators14 may think it does is that they believe Kant’s argument for permanence has something to do specifically with making succession possible. This, as we have seen in part 1, is not the case. Besides, it is somewhat less clear that Descartes would accept the certainty of previous conscious states, even recent ones. The thought that I am now conscious (that, say, I now have a toothache) already is an “empirically determined” consciousness of my existence, without any specific connection to a succession of states. It is because I think that I am now conscious that I am conscious of 13. Kant’s change of mind (see preceding footnote) is not a shift from the A to the B edition of the Critique, since he already holds in the A edition First Analogy that we can be immediately aware of what is distinct from all our own determinations. If the B edition Refutation of Idealism were incompatible with transcendental idealism, the First Analogy would be too, which is absurd. Further, if the B edition Refutation is thought to commit Kant to things in themselves (is thought to require such things in order to have inner experience), the First Analogy would likewise be committed to substance (the permanence of the real) being a thing in itself, which once again is completely absurd. The truth is that Kant, very early on (before he even had the category of substance), held a phenomenalistic version of transcendental idealism, but gave it up already in the A edition. 14. Bennett, Kant’s Analytic, and Paul Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987).

136

the principles

my own existence “as determined in time.” What Kant is holding, I believe, is that the present moment is inherently and inseparably the limit of past time. Otherwise Descartes’s cogito would have to be formulated as “I am certain that I now* have a toothache,” where “now*” signifies neutrality as to whether time itself begins now. This would mean that I am leaving it open, as far as my certainty is concerned, that perhaps I am at a boundary of time. This is a more extreme idea than time’s being finite and unbounded. It is more extreme even than the idea that time can be finite and bounded, where the “boundary” is a singularity. It is the idea that I am at the boundary that, unlike the other two ideas, seems to be open to the absurdity of there then having to be what is on the other side of the boundary (what “comes before” time). Now, neutrality regarding this idea is every bit as problematic as the idea itself. I am being asked to say that I am certain of being presently conscious in a sense of “presently” that is unfathomable in leaving it open whether the present begins time. The proper attitude, it seems to me, toward thinking I am now* conscious is bafflement, not certainty. Descartes can convincingly siphon off or separate the internal from the external world because there are hallucinations or dreams, where things are apparently the same to my thought that I am conscious without relation to the external. He cannot similarly convincingly siphon off the present from the past. Descartes may convince us that perhaps we and all our memories were just now created, but not that time was just now created. In a reflection Kant says, “Our existence in time is determined in inner sense; and this presupposes the representation of time itself [der Zeit selbst]”;15 i.e., in order to think that I now exist in a conscious state, I must be able to represent “time itself ” (the time from which the present emerges). Note that this first point is not that to have the thought of presently having a toothache or presently having a visual image of red requires only the thought or idea of the past. Rather, since what I know is that I am now having a toothache (since I cannot know or be certain of now* having a toothache), it follows that the present exists and hence that the past exists. The knowledge of my own present “empirically determined” existence, or my present empirical consciousness, then, presupposes that there is past time. Alternatively, that there is past time is as certain as my thought that I am now conscious. As per the First Analogy, the existence of past time is just the propri15. Kants Gesammelten Schriften (Berlin: Walter de Gruyter, 1929), 18:611, #6311. He does go on to say, “time, however, contains the representation of change; change however presupposes something permanent.” I suggest that what he means by time containing the representation of change is that the present, or now, is what preceding time gives way to. It is this latter (the idea of past time giving way to the present) which requires the permanent.

Refutation of Idealism in the B Edition

137

ety of being in the course of a temporizing procedure, and that propriety, in turn, presupposes something present that is permanent (has duration into the past, in the sense of being that in regard to which it is proper to be in the course of temporizing). Hence Kant can say, “All determination of time presupposes something permanent.” Indeed, he says it presupposes something permanent in perception. His point, I believe, is to deny Descartes’s representative theory of perception, according to which I do not immediately perceive what is external but only infer it from my internal state. This contrast is what is expressed when Kant goes on to say, “Thus perception of this permanent is possible only through a thing outside me and not through the mere representation of a thing outside me.” Recall that the point of the Refutation is to prove that we have immediate (noninferential) experience of outer things. Now, an internal representative of a thing cannot be the basis of being in the course of a temporizing procedure, since it is a momentary state that exists only in the present. Since it has no continued existence, it is not at a stage of its existence that requires me (in order to keep up with it) to be in the course of temporizing. This leaves the possibility that maybe the inferred thing itself is the basis of being in the course of temporizing. However, as we saw in the preceding section, a thing that is not intuited (directly perceived), but merely conceptualized, cannot be the basis, without violating Kant’s doctrine that time is merely a form of intuition. Our argument in the preceding section likewise eliminated the self or the subject of experience as the basis. Thus the basis can neither be the subject nor an unintuited thing distinct from the subject. Further, again from our discussion in part 2, the basis cannot be any state or condition (even a qualitatively unchanging one) of me, and so cannot be found in inner sense. Indeed, as we saw, it cannot be a state or condition at all, not even one that is outerly intuited, but can only be a thing or an entity. All together, then, we have the conclusion that the permanent that determines my existence in time (viz., the basis of being in the course of temporizing) must be a thing outside me in space that I am immediately aware of.16 In sum, then, if I know that I am now conscious (seeing red, feeling 16. Kant’s refutation of Descartes, as I have reconstructed it, depends inextricably on his thesis that time is constructive. Although I presented this thesis in part 2 above as based on the continuity of time, I believe there are arguments for it that are independent of continuity. So far, all the argument has shown is that there must be at least one object in space. We can extend the argument to a whole physical world as follows: I know there is something in space outside me. I am not, however, at the edge of space (any more than I am presently at the beginning of time). Hence, since space is constructive, it must be proper to spatially construct (i.e., to move or to spatially resituate myself). But past time exists at all places, viz., at all stages of spatial construction. Hence, at each different stage of spatial construction it must be proper to be in the course of a temporizing procedure,

138

the principles

sad, etc.), then the present exists. The present exists, however, only as the present stage of ongoing time, which in turn exists only in the propriety of being in the course of a temporizing procedure. This, in its turn, requires there being a basis in present (intuitable and spatial) reality in relation to which it is proper to be in the course of temporizing, which implies that this reality is a substance (a permanent real) outside me that I immediately intuit. It follows that what I know if I know that I am presently conscious is that I am immediately conscious of things outside me. It will not do to say that the mere idea or conception of a permanent thing outside me is all that is required or presupposed for determining my existence in time. This would be like saying that the mere idea or conception of a mixing bowl with three ingredients in it can make it proper to be in the course of a cake-baking procedure. Just as it is the actual existence of the bowl that is presupposed (that can properly set me ahead in the procedure), so too it is the actual existence of something permanent that is presupposed (that can properly set me ahead in the temporizing-cum-tracking procedure). Kant is going to go on to say in the Notes that his conclusion is consistent with the fact that I can be mistaken as to perceiving a thing outside me. He says that “it does not follow that every intuitive representation of outer things involves the existence of these things.” However, if I am mistaken (deluded or having a hallucination), then surely it cannot be on that basis that it is proper to be in the course of a temporizing procedure. A hallucination of a bowl with three ingredients in it does not make it proper to be in the course of baking a cake. How then is the existence of hallucinations compatible with determining my existence in time? Kant says, “All that we have here sought to prove is that inner experience in general is possible only through outer experience in general” (emphasis added). Now, what I know (if I know that I am now conscious) is that if my present perception is hallucinatory, then there is some other attainable veridical perception of an outer thing that I can reach (say, by altering my position), so that my present hallucinatory consciousness (together with the sequence of conscious states accompanying my shift of position) can be determined in past time. In other words, if my present outer consciousness is not sufficient for time determination of my present inner experience, some other such outer consciousness must be sufficient for determining a sequence of inner experiences (that includes my present one) in time. The crux of Kant’s Refutation is the “indissolubility” of time; that the and hence there must be substances throughout space (throughout the course of spatial construction) making it proper to be in the middle of temporizing procedures. Thus there has to be a whole physical reality.

Refutation of Idealism in the B Edition

139

present inherently emerges from the past. This indissolubility (once brought to our attention) precludes being certain that I am presently conscious without being certain that there is a past. One might conclude from this that therefore what Kant has shown is that we really can’t be certain that we are presently conscious, since “so much” is involved in the supposed certainty. Thus, although Kant would have shown that “even our inner experience .l.l. is possible only on the assumption of outer experience,” he would not have shown “that we have [such] experience .l.l. of outer things.” Rather, all he shows is that either we have outer experience or else we cannot be certain of inner experience. The indissolubility does show, however, that we can’t siphon off a realm of purely present certain consciousness. Thus the skeptic can no longer say he is certain of just that realm, while the rest is less certain. What he is left with is the claim that perhaps none of it is real or true, which includes that perhaps I do not now think that I am in a conscious state. But in not having any contrast, he would no longer have any means of seducing us. We could say to him, “If I cannot know that I am now conscious, then you cannot properly believe that you are now wondering whether I can know this” (where he, of course, can be ourselves).

4. further issues Kant, in the Notes to the proof, brings up two significant issues. One issue pertains to what the anti-skeptical worth of the proof is, if we still have to distinguish hallucinations from veridical perceptions (Note 3). The second issue pertains to how it is comprehensible or explicable that we could be immediately conscious of what is outside us (footnote [a] to Note 1). I will discuss these issues in just this order because settling the issue of the externality of consciousness depends on settling the issue of distinguishing veridical perception from hallucination. Both Descartes and Kant hold that hallucinations (dreams, delusions) are distinguished from veridical perception by some sort of coherence. Both, however, owe an account of why that coherence goes with veridical perception. For Descartes, this means that he owes us an account of why coherent perceptions are the ones caused by (representative of) entities that we do not directly perceive. This would seem to be a daunting task, since it would seem to make equal sense to say that our perceptions that are coherent across time (which, in fact, is Descartes’s conception of coherence)17 are the ones due to the nature of our own selves, while the incoherent ones are the ones due to dis17. See the sixth meditation in Descartes’s Meditation on First Philosophy, trans. Laurence J. Lafleur (Indianapolis: Bobbs-Merrill, 1960).

140

the principles

turbing “extrusions” into our consciousness from external objects. After all, the operations of complex organized entities (such as machines) are such that the “coherent” workings are due to how they are built to function and so how they work on their own, while the “incoherent” workings are due to external disturbances. In the sixth Meditation Descartes just presumes, without argument, that coherence goes with what is externally caused. The opposite conclusion, however—that incoherence is what goes with what is externally caused as a disturbance—seems just as plausible. Even after Descartes distinguishes hallucinations from other internal states, his representative theory gives him no basis for connecting coherence to representativeness. Kant has an analogous problem. He too holds that coherence or “congruence with the criteria of all real existence” is what distinguishes veridical perception. The criteria he has in mind have something to do with causal coherence; i.e., our veridical perceptions are the ones that, taken as a group, have the regularity that will support (be criteria for) causal laws. But now the question is why should the “regular” perceptions that support causal laws be the ones that are perceptions of what is outside us (as opposed to the perceptions that merely seem to be of what is outside us)? Unlike Descartes, however, I believe Kant has some sort of answer to his version of the question of why veridicality goes with coherence, for in the Second Analogy he has already shown that there can be no time series, except if occurrences are governed by strict causal laws. Let us briefly look at this argument. Time is such that preceding times don’t just happen to give way to succeeding times. Rather, they necessarily advance to or determine succeeding times (A194, B239, p. 222; A199, B244, p. 225). Since time itself cannot be perceived (A200, B245, p. 246), this relationship of necessary advance must be represented in terms of occurrences or events which thus “must determine for one another their position in time, and make their time order a necessary order” (A200, B245, p. 246, emphasis added). Thus preceding occurrences necessarily advance to or determine succeeding occurrences (so as to form an entire series of necessarily advancing occurrences). But now the relation of necessary advance or determination between occurrences is just the causal relation. Further, the necessity of preceding times’ advancing to succeeding ones is an absolute or unrestricted (unconditioned) necessity, and so the advance from preceding occurrences to succeeding ones must be likewise absolutely necessary. Absolute necessity, however, implies strict universality (B4, p. 44), since if occurrences like e1 didn’t always determine or necessarily advance to occurrences like e2 (the succeeding), the advance from e1 to e2 wouldn’t be absolutely necessary, but neces-

Refutation of Idealism in the B Edition

141

sary only on the condition that it take place on certain occasions (i.e., it would be a restricted or conditional necessity). Thus occurrences are connected by strictly universal causal laws. Note that this argument doesn’t say that causation is a criterion of veridical perception, but rather that causation constitutes the nature of time (the necessary advance of the time series from the earlier to the later). We can now conclude that those of our perceptions that, by their outer content, do not have a regularity consistent with the existence of strictly universal causal laws cannot be taken at face value as far as their outer content is concerned, and hence must be the hallucinatory perceptions. Otherwise there would be no necessary advance in the time series, and so no existence (even of hallucinations) in the necessary advance of time.18 Kant says in the Postulate regarding actuality that the analogies of experience “define all real connection in experience in general” (A225, B272, p. 243, emphasis added). These analogies, Kant says, are principles for determining existence in time according to the three modes of time (A215, B262, p. 236). It is because the analogies constitute the nature of time that coherency of perceptual content in accord with the analogies is a criterion of veridical perception (of reality of content). Thus, just as the First Analogy is the key to the refutation of idealism itself, so the Second Analogy is the key to a distinction between veridical perceptions and hallucinations, without which the antiskeptical (if not the anti-idealist) import of the refutation would be minimal. We can turn now to the issue of how to understand the immediate consciousness of outer things. Of course in the Refutation Kant has proved that such consciousness must be possible if there is inner experience “be the possibility of this consciousness understood by us or not.”19 The problem of understanding such consciousness is well put by Kant in the first-edition Refutation of Idealism: “For we cannot be sentient outside ourselves, but only in us, and the whole of our self-consciousness therefore yields nothing save merely our own determinations.” 20 But outer objects, as shown by the B Refutation, are distinct 18. This doesn’t mean that hallucinations are not subject to universal causal laws. Rather, it means that their being so depends on characterizing them as internal simulations, and so not taking their outer content at face value. (Hence, they are conditioned by material reality quite differently from veridical perception). 19. Footnote (a) to B227, p. 246. In this footnote he doesn’t explain the possibility of such consciousness, he simply reiterates, in line with the Refutation itself, why it must be possible. In the preface to the B edition (footnote (a) to Bxli, p. 36) he says, “How this [that an external existent is part of experience] should be possible, we are as little capable of explaining further, as we are of accounting for our being able to think the abiding in time.” 20. A278, p. 350. In the first-edition Refutation Kant accepted that we can only be

142

the principles

from all our determinations, and so it seems that to be immediately conscious of them we would have to be “sentient outside ourselves.” From our preceding discussion on causal laws, we may presume that outer objects are material objects according to Newtonian physics. Indeed, since the empirical content of our perceptions is such that the causal laws they “lead to” are Newton’s laws, therefore our veridical perceptions are perceptions of Newtonian material objects. I think that it is only at this stage (based on the empirical content of our perceptions) that Kant is in a position to say that the outer objects that are the bases of being in the course of temporizing are, specifically, material objects. Of course, in Note 2 Kant exemplifies the result of the Refutation with the notions of matter and impenetrability, but there is nothing in the proof itself that leads to this conclusion. Presuming, then, that outer objects are material, the question becomes, how we can be immediately conscious of material (impenetrable) objects outside us? Recall that the basis of being in the course of temporizing is presently being affected. This may now be taken to signify a state of probing for gross mechanical interaction, such as attempting to push, pull, lift, etc., and detecting for mechanical resistance by some feedback mechanism. I suggest that as long as the brain (or mind) is suitably connected to our bodies, this detection of resistance to our probe will happen only with the force of an external material reality. The reason is that the body is affected in a large-scale way by the mechanical interaction (the lifting, pushing, etc.), and any internal simulation of this large-scale bodily state (by parts of one’s own body acting on other parts) would require such further drastic bodily changes that the brain would easily detect them (if it is at all well connected to the body). Further, as to external simulation, there is no such thing as fake or artificial matter for influencing the body, and any machinery that bypasses influencing the body by going directly to the brain would be detectable by the organism by feeling around his head, etc., (so long as his brain is not tampered with so as to desensitize its relation to his body). Thus this consciousness (detection with his feedback components normally in place) can only happen with the existence of the material object outside us, and so in this sense is a consciousness of (consciousness pertaining uniquely to) what is outside us. Further, it is an “immediate” or “direct” consciousness. It is not as if the external impression of the external force or the impact of the material object is a cause distinct from the being impressed or the being impacted of the sentient of our own determinations, but held that outer spatial objects were a species of our own determinations.

Refutation of Idealism in the B Edition

143

organism’s body. The mechanical transaction, that is, is not two distinct events, any more than the bombarding by one billiard ball of another is a distinct event from the being bombarded of the other billiard ball by the first one. If so, then it isn’t the case that my consciousness (detection) of being impacted is consciousness of something distinct from the impacting of the external object. The relationship between what I am conscious of and the external object, then, is not like that of a sign or image or even like that of an effect to a distinct cause. The relationship rather is that of a transaction characterized passively to that very transaction characterized actively. Hence I am immediately conscious of the impacting material object. Now of course we can and do hallucinate being impacted when our feedback mechanisms between brain and body are severed or impaired, as can happen in “altered states” such as when we are drugged or asleep. This doesn’t imply that our unimpaired consciousness is not immediately of external material reality. What it does imply (and here Kant must break with Descartes) is that I may not be able to tell or discern immediately which conscious (detective) state I am in (an impaired or an unimpaired one). The point is that we can be “sentient outside ourselves” so long as we allow that being in a conscious state does not entail being not mistaken that we are in that state.

PART IV THINGS IN THEMSELVES

8

Kant on Things in Themselves

Understanding Kant’s doctrine of things in themselves involves understanding three claims he makes. First, we do not cognize things in themselves. Second, they are not in space and time, and third, the categories do not apply to them. These claims, I contend, are utterly central to Kant’s entire theory of cognition in the Critique and cannot be discarded without discarding the Critique itself. In the first section of this essay I shall clarify and defend Kant’s claims. In the second section I shall discuss variations in Kant’s thinking regarding these claims that are evident in certain passages of the text.

1. cognition of phenomena In order to understand even what Kant means by the claim that we do not cognize things in themselves but only phenomena, we have to understand the basics of Kant’s account of cognition. The basic elements involve the nature of space and time, and then the nature of thought or representation. Space and time for Kant are not objective realities but rather mere activities or constructions or motions of the subject by which it is liable to be affected by things (A26, B42, p. 71).1 Thus, in order to be affected by something, I must move and thereby make the object present. This moving gives the object location as being presented that far along in the motion of the subject. The location, then, is in terms of the activity of the subject. As an analogy, consider the idea of there being an order to the several objects on my desk (so that one of them is first, another second, etc.). Apart from an activity or procedure of mine that orders the items, they themselves have no intrinsic order. Similarly, for Kant, apart from being manifest or present at stages of motion, objects have no intrinsic location. This is just Kant’s denial of both absolute and relational theories of space. Similar remarks apply to time, which for Kant is nothing but constructions or activities of timing (marking time). We shall 1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition.

147

148

t h i n g s i n t h e m s e lv e s

not consider time in this essay since the essence of Kant’s three claims regarding things in themselves can be revealed for the case of space. So far we merely have a subject moving about and getting affected (being bombarded), then moving further along and getting further affected, etc. This much characterizes what Kant calls sensibility (A22, B36, p. 67; A494, B522, p. 441), which consists of sensation (getting affected), and space or the form of sensation (the moving about that enables and precedes getting affected). Cognition or thought, for Kant, is the unity of sensible intuition. The unity of an activity (such as moving about and getting affected) is a rule that governs it (A103, p. 133). The rule gives the activity a beginning and an end and a unity of stages. If so, then thought is a unity of sensible intuition by being a rule for it, viz., a rule for moving about and getting affected. This, I claim, is Kant’s basic notion of cognition of the world. Cognition is nothing other than rules for moving about and getting affected in various ways or obtaining presentations. Cognition, we note, goes beyond actual presentations since it is in the form of rules for how it is proper to move about and be affected (obtain presentations), whether one actually does it or not. In this sense cognition pertains to actual and possible presentations, where possible presentation means proper or legitimate or correct presentation. In sum, for Kant a cognition is nothing more than a rule that we can set out as: (1) It is proper to take k steps to (thereupon) be affected and react in such and such a way. With this account of cognition we are now ready to clarify Kant’s contention that we cognize phenomena, not things in themselves. Phenomena are things whose entire existence is in their “showing up” or being manifest. Their whole existence, that is, is their presentability (their being presentable). This “phenomenality” or manifestability is all there is to their existence. In particular, they have no intrinsic existence beyond, or in addition to, their properly entering into relations of presentation or manifestation. A rule for moving about and being affected as in (1) above does not represent any existence or reality of things beyond presentability. Indeed, it is simply a rule for the propriety of presenting upon moving about. Now, a cognition of what has existence beyond, or in addition to, being a phenomenon, would represent not only the propriety, say, of taking ten steps and being affected, but also that there is an entity or object waiting there to be met with (A505, B533, p. 448), or waiting there to possibly affect us. This latter involves not only a rule for moving about (taking steps) and being affected, but also the idea that apart from the rules (and even grounding

On Things in Themselves

149

the rule) there is an entity existing in its own right that we move to and so get affected by. In contrast to (1) above, a supposed cognition of something as having an intrinsic existence in addition to mere manifestability would be (2) It is proper to take k steps to and to (thereupon) be affected by the object that is there waiting to affect us. When Kant says that we cognize mere phenomena, then, he is holding that our cognitions are rules for obtaining presentation as in (1) above, and do not extend to cognitions or representations of objects as also having an existence of their own (beyond the manifestability represented in the rule itself) as in (2). Phenomenal reality, understood as that whose existence is entirely as entering into presentation, certainly need not be mental or have the existence of mental contents. A presentation in space (in the course of moving about) is not an internal state of the subject but a relation of the subject to what affects or bombards it. If we consider presentation as an event, then it is an event that partially happens outside the subject.2 Kant does talk of our cognition as limited to (what exists in the course of) possible experience (A495, B523, p. 442). If we understand experience, however, to involve a relation, this is compatible with our cognition being of what exists outside us. The same holds for Kant’s talk of our cognition being limited to appearances or to possible representations. If an appearance is understood as something that happens partially outside us, and a representation is understood as involving what exists outside us, then the restrictions of our cognitions to appearances or representations need not be a commitment to phenomenalism (a restriction to actual and possible sensory states of the subject). The only point is that our cognition does not pertain to what exists in its own right and only besides has the further potential to become present or manifest to us as in the underlined in (2) above. The doctrine that our cognition is limited to phenomena Kant calls transcendental idealism, and he thinks he has established it already in the Aesthetic (A401, B519, p. 439). In the Aesthetic the conclusion is stated as the fact that what exists in space is merely appearance (A26, B42, p. 71). This conclusion is drawn right after Kant concludes that space is merely a form of intuition, which itself is supposed to follow from the Metaphysical Exposition. In the Exposition Kant argues that space is such that the whole precedes the parts. I believe he is referring to any extent of space whatsoever, and not specifically to the whole of 2. Indeed, for Kant, enduring substances are phenomena, though they clearly do not exist in our mental state.

150

t h i n g s i n t h e m s e lv e s

space. Further I believe that for Kant the whole must precede the parts because space (any extent of space) is continuous and what is continuous for Kant is what is so seamless that it cannot be put together out of or composed from parts. Now, anything that exists “objectively,” such as a system of relations or an empty receptacle, is such that a whole (any whole) of it exists only if all its parts do. The parts, thus, would be prior to the whole. By contrast, a flowing construction or motion is such that the construction of the whole is prior to the segmentation into parts. My sweeping out of a line, for example, is not composed of a series of “stops and starts” (cuts or segmentings) but is a seamless flow. I can potentially stop and so produce a cut in the flow or a part, but in fact in constructing the whole line (expanse) I do not actually construct parts. In sum, only if space is flowing motion can space be continuous. Since continuity, for Kant, is of the essence of space (in being its expansiveness or extensiveness), it follows that space is nothing but flowing motion. In particular, there isn’t any objective space (relational or absolute) that one moves through, since motion alone is continuous. This motion is what enables us to get affected or to obtain presentations (intuitions), and so we have the conclusion that space is nothing but the form of intuition (i.e., it cannot also be something objective). Why now should Kant conclude from this space being merely a form of intuition that what exists in space is mere appearance, i.e., something that exists entirely in appearing or in being manifest? Suppose that what properly appears in the course of moving about has an (intrinsic) existence beyond (the relation of) appearing. Then it exists in such a way that it does also appear at a specific stage of moving about. Its existence then is as something “waiting to be met with” (A493–94, B521–22, p. 441) or something waiting to appear at that stage. Since its existence is tied to our getting there to meet it (or for it to present itself), it must be in space, indeed in a place that our moving gets to (see [2] above). In sum, anything that appears can have existence beyond mere appearing only by its waiting for our motion to get to it, and so only being in an objective space that corresponds to our moving about. But, as per the first conclusion of the Metaphysical Exposition, there can be no such objective space, since space is nothing outside of moving about. What is in space, then, has no existence beyond its appearing in the course of motion. Kant has other arguments for the mere phenomenal existence of what is in space—in particular the argument of the Mathematical Antinomies,3 and so for the first claim that we do 3. Without the continuity argument of the Aesthetic Kant can still hold that purely discursive (descriptive, depictive) representation is empty; and without rules for proce-

On Things in Themselves

151

not cognize things in themselves. For now, however, we turn to the other two claims Kant makes regarding things in themselves. That things in themselves are not in space (and time) can be established as follows. If the supposed things that exist beyond mere manifestability don’t also manifest themselves to us at all in the course of moving about, then they don’t exist in space, because, whatever else, something in space (even supposed objective space) can be reached by moving to it. On the other hand, if these things that exist beyond mere manifestability do also manifest themselves, then they would be what appears in the course of moving about, which, as we have just seen (in regard to the second conclusion of the Metaphysical Exposition), is impossible, since what appears in the course of moving about has no existence beyond being manifest. Things in themselves (viz., what has intrinsic existence beyond manifestability) cannot be in space.4 The inapplicability of the categories to things in themselves holds for both pure and schematized categories. Pure categories, for us, are nothing more than logical functions of judgment. They require (rather than supply) some basis of semantic applicability to bring them to bear upon extra-conceptual reality, or reality that can be given in intuition (B148, p. 163). Now it is schemata that first bring judgment forms to bear on what can be given, but the schematized categories simply “contain the necessary unity of the pure synthesis of imagination in respect of all possible appearances” (A119, p. 143). The pure productive imagination involves the capacity for spatial construction (synthesis), the “necessary unity” of which is the unity of a rule. If so, then the semantic functioning of categories (pure concepts) is to enable (contain) the formulation or existence of rules for spatial (and temporal) construction that extend to or cover the full (spatio-temporal) range of possible appearance or presentability. The categories, then, do not pertain to any existence beyond presentability, since their function is exactly to enable the cognition of the full scope of (rules for) presenting reality. It is not surprising that Kant’s understanding of the real use (applicability) of pure concepts should be in terms of how they function in cognition according to his understanding of cognition. Since the latter does not pertain to any existence beyond presentability, neither do the pure concepts that function to effect it. dures for contacting or presenting, there would only be purely descriptive representation. Thus for nonempty representation we must incorporate rules for moving so as to make contact or rules for presentations. This still leaves (2) above as a nonempty representation of what has more than mere phenomenal existence. The Mathematical Antinomies give an argument against (2) based on issues of the general representation of what exists in space. I discuss this argument in Essay 9. 4. Similar results apply to time.

152

t h i n g s i n t h e m s e lv e s

If cognition is restricted to phenomena, then things in themselves have no cognitive role. Nevertheless, for Kant, the idea of existence or reality does not entail that reality is merely phenomenal. It is only the cognition (definite representation) of existence or reality that entails it. Hence existence in itself, though uncognizable, is not incoherent. The concept of such existence, which Kant labels noumenon in the negative sense, is said to be “a merely limiting concept, the function of which is to curb the pretensions of sensibility” (A255, B311, p. 272). For our purposes the important point is that the concept of the thing in itself is a mere limiting concept, since existence in itself is in no way, directly or indirectly, involved in cognition. In particular, it is not the case that cognition involves a transcendental object as ground of affection. Nor is it the case that we cognize things in themselves, but only as they appear. There are, however, passages where Kant seems to express each of these views. What I wish to argue now is that the position we have set out in section 1 is Kant’s considered view, but that other “earlier” (discarded) views still find expression in the Critique, and it is these expressions that give rise to many of the mistaken accounts of Kant’s doctrine of things in themselves.5

2. the development of kant’s views In the Transcendental Deduction from A104 to A110, pp. 134–38, Kant discusses the objectivity of our representations. To represent objects outside our sensible intuition is impossible for Kant, as it would require a purely conceptual representation of them.6 The cognitive function of such objects would be to constrain our sensible intuitions or our reactions. This function can be achieved without such objects, however, by rules for reacting, since a rule constrains actual reaction, and an actual reaction may conform or fail to conform (correspond or fail to correspond) to a rule for reaction. Thus objective representation is not via a representation of entities outside reactions corresponding to them but via rules for reacting.7 The unity of a rule is just the unity of apper5. For a much more detailed discussion and defense of this development of Kant’s views, see my Space, Time, and Thought in Kant (Dordrecht: D. Reidel, 1989). 6. Kant believes it would require, that is, exactly the kind of purely conceptual representation that he had accepted in the second part of the Inaugural Dissertation. It was the rejection of this kind of representation that initiated the Critique. 7. In this way Kant has imported objective representation into the spatio-temporal world. In the Inaugural Dissertation there are spatio-temporal procedures that enable sensible reactions, but no objective cognition in the sensible realm. The entire problematic of Kant’s “critical” theory of cognition is to bring objective representation (and ultimately conceptual representation in general) to bear on the sensible realm.

On Things in Themselves

153

ception;8 viz., the understanding thinks in rules. But now Kant adds that the “pure concept of this transcendental object, which in reality throughout all our knowledge is one and the same, is what alone can confer upon all our empirical concepts in general, relation to an object, that is, objective validity” (A109, p. 137). If rules subserve the function of objects (to be a constraint on our reactions, or to be something our reactions may or may not correspond to), then why do we need the transcendental object for objective representation? The answer, I suggest, concerns the basis or the authority of the rule. A rule can be conventional or arbitrary or made up. A conventional or made-up rule would hardly express objectivity. Thus if the rule to react “red,” say, is one that I invent for myself, this hardly constrains my reactions. It is true that if I go ahead to react “green” I fail to conform to the rule, but since the rule is quite arbitrary, it is not clear what objective significance this failure has. Objectivity then must be a matter of unconventional or nonarbitrary rules. Now, reality in one sense is exactly what contrasts with what is made up or invented. The basis of the rule for reacting, then, must be reality, or what is so, and this I suggest is the transcendental object. We can represent Kant’s account of an objective cognition in this A104–10 section, then, as follows: (1) In regard to what is so, it is legitimate or proper to react in such and such a way. This transcendental object (this reality outside all proper reactions) is indeed “one and the same” in all cognition. If there is a second rule of how it is proper to react, it too (to be nonarbitrary) is in regard to what is so, but there aren’t different what-is-so’s basing different rules. In this account of objective cognition, the operative contrast is between appearances (i.e., proper reactions or possible perceptions) that are mere representations, and the transcendental object (which is outside of all proper reaction). Kant does say that “appearances are not things in themselves; they are only representations.” He doesn’t say, however, that they are not things in themselves because they are things only as they appear. His point seems simply to be that appearances themselves are simply not things at all, but merely representations. Appearances, then, are not being contrasted to things in themselves but are being denied the status of thinghood. Note that in this account of cognition, uncognizable reality (the transcendental object) is involved in objective cognition. If this were Kant’s final word on the matter, then 8. Equivalently, it can be characterized as a necessary unity, since once again the unity of a rule for reacting is how it is required or necessary to react.

154

t h i n g s i n t h e m s e lv e s

indeed Kant’s account of cognition would be inseparable from uncognizable reality. I believe it is this account in (1) above that is the textual basis of the idea of transcendental affection, viz., the idea that objectivity requires that appearances (sensible perceptions) are grounded in, or due to, a reality beyond our cognition. This idea in turn gives rise to one of the supposed incoherencies in Kant’s notion of the thing in itself, namely, such transcendental affection seems to involve a nonempirical employment of the category of causation, while Kant insists there is no transcendental employment of the categories at all. Note, however, that Kant is not basing actual perceptions (which are events) on the transcendental object as the cause or ground of these occurrences. Rather, it is the rule for proper reacting that is based on what is so (as opposed to being an arbitrary or contrived rule). The transcendental object is the basis of the rule, not in the sense that it produces it (even a-temporally), but rather in the sense that it is the topic of the rule, or what the rule pertains to. The notion of what rules pertain to or are in regard to is not, even transcendentally, an employment of the category of causation. Thus, although this account of objective cognition is the textual source of attributing transcendental affection with all its problems to Kant, he is really holding only that the transcendental object is the “correlate” of sensibility (that which goes with rules for perceptions, or that which rules for perceptions pertain to), which is not a doctrine of affection at all. The account in (1) above is also, I believe, the source of the “twoworlds” interpretation of Kant’s distinction between appearances and the thing in itself. Appearances, indeed, are sensible representations. Since these, together with their propriety, are all there is that is cognizable, Kant is a phenomenalist regarding cognizable reality in this account of cognition. There simply are no phenomenal entities or objects of experience. There are only proper reactions. On the other hand, the basis of the rules for reacting is ontologically distinct from any reactions, since it is certainly not actual reactions that are the basis of how it is proper to react, nor actual reactions “considered in some other way” (whatever that would mean) that are the basis. I am going to claim presently that Kant drops the account of cognition represented by (1) above. If so, then the two-worlds interpretation of Kant, and any (even mistaken) attribution of transcendental affection to Kant also go by the boards as mere remnants of an early attempt at a theory of cognition. My contention is that commentators often attribute to Kant not only views that he abandoned, such as the two-worlds view, but even views that combine what he abandoned with what he later came to hold. We

On Things in Themselves

155

shall argue below that Kant eventually came to hold that the thing in itself (noumenon in the negative sense) is not a component of a theory of cognition at all, but merely a limiting concept that reminds us that it is only cognizable reality, not reality per se, that is subject to a theory of cognition. This of course contradicts the view he holds on the early account in (1) above, where the thing in itself (the transcendental object) is clearly a component of cognition.9 Nicholas Rescher is one such commentator who attributes the abandoned early view together with the later mere limitative view, for he holds that the commitment to things-in-themselves both blocks phenomenalism and is a limitative concept. Indeed, Rescher holds that it is a limiting concept because it blocks phenomenalism. The limitation is due to “the very nature of human reason to .l.l. take them [to be] not just appearances, but appearances of things as they are in themselves.”10 The early view of objective representation is repeated in the Second Analogy and in the Fourth Paralogism. What I claim now is that it was the issue of causation that drove Kant from his early view. The argument for causation is that there must be necessary determination between appearances (proper reactions or proper perceptions) in order to represent the necessary advance of time. An objective succession exists in a series of (preceding) times in which each time doesn’t just happen to advance to the next but necessarily advances to or determines the next. To represent this requires that the objective succession be part of an ongoing series of further (preceding) appearances or proper reactions, each of which necessarily advances to or determines the next. Hence, if it is proper to react r1–r2, there must be a preceding reaction r0, such that its being proper to react r0 forces or determines it to be proper to react r1–r2. But Kant came to see that causation (series of necessarily advancing appearances) presupposes substances or a plurality of entities. For example, the objective succession of a ship being upstream and then a bell ringing on the shore is not a succession in a se9. One cannot avoid this contradiction by saying the noumenon in the negative sense is a notion distinct from the transcendental object, and so Kant can hold the former to be only limitative, while holding the latter to be involved in cognition. Kant, in effect, identifies these notions when he characterizes the noumenon in the negative sense as “a something in general outside our sensibility” (see B307, p. 268), which is exactly a characterization of the transcendental object. At this late stage, however, this transcendental object has only a limiting function. 10. Nicholas Rescher, Kant’s Theory of Knowledge and Reality (Washington, D.C.: University Press of America, 1983), 9, 13–14. Rescher isn’t quite correct about things in themselves “blocking” phenomenalism at the early stage. Rather, the transcendental object enables phenomenalism to contain objective representation. For another phenomenalist two-worlds view, see James van Cleve, Problems from Kant (New York: Oxford University Press, 1999).

156

t h i n g s i n t h e m s e lv e s

ries of preceding appearances that necessarily advance to it. The reason is that the proper reactions pertain to different entities, and each reaction is the tail end of a different series of determining appearances (one series for each entity). Now, in Kant’s early view, there is objective representation but no plurality of entities or objects at all, and so the early view has to be abandoned. We consider now the revision that constitutes what can be called his middle view. The second “preparatory” part of the A edition Deduction at A111– 14, pp. 138–40, for the first time talks of objects of experience (viz., of a plurality of entities). Further, it explicitly ties causation (a series of events each determining the next) to objects. Recall, now, that in rejecting his view in the Inaugural Dissertation, Kant was rejecting a purely conceptual representation of entities or objects. The early view fit the rejection by simply sidestepping the representation of entities altogether, as being no part of objective cognition. If Kant, forced by causation, is to reintroduce representation of a plurality of objects, then intuition or sensibility must be involved in that representation. The solution Kant comes up with, I suggest, is the idea of things as they appear, which for now we can simply say are things as they arise in the course of spatio-temporal construction on the part of the subject. In the present section of the Deduction, Kant uses the phrase “objects of nature themselves” to contrast with objects of experience, which must then be not objects themselves but objects as they appear. We can represent Kant’s account of cognition at this second, or middle, stage, as follows: (2) .l.l. with respect to the thing as it appears there, it is legitimate or proper to react in such and such a way, where the three dots stand in for some spatial construction or shift of attention, such as taking five steps to the right. Different spatial constructions, then, are what determine a representation of a plurality of things appearing, so that Kant has reintroduced a plurality of entities within the spatio-temporal realm. This solves his problem with causation, because we can differentiate the proper reaction “ship upstream” as proper with respect to one thing as it appears, and the proper reaction “bell ringing” as proper with respect to another thing as it appears, and then we can represent each propriety in its own series of preceding appearances (proper reactions) where earlier ones determine later ones.11 11. So, if I represent the ship upstream (and then downstream) as follows: Take three steps—with respect to the thing as it appears there it is proper to react “upstream,” “downstream,” then I represent this succession in an ongoing necessary advance of time as follows:

On Things in Themselves

157

Because his theory of cognition now involves a plurality of things, Kant would be particularly concerned to deny that he has lapsed back into the Dissertation view according to which we can represent those things purely intelligibly (i.e., as noumena in the positive sense).12 This concern, missing from his early view of cognition, is what leads him at this stage to add a section on phenomena and noumena. The section he adds is the part of the chapter deleted in the B Edition at A249–53, pp. 265–71. This indeed is the key section in which Kant repeatedly uses the terminology of things as they appear as opposed to things in themselves. He says, “if the senses represent to us something merely as it appears, this something must also in itself be a thing” (A249, pp. 266–67).13 Later on he says that sensibility “does not have to do with things in themselves, but only with the mode in which .l.l. they appear” (A251, p. 269). A thing as it appears in the course of a spatial procedure must be something in itself because my reaching it by a spatial construction only enables it to appear; it doesn’t make it be something. It must already be something waiting to appear (awaiting only my spatial shift of attention in a motion I perform in order to appear). It is the thing’s appearing that arises in the course of a spatial construction, not the thing per se. Hence the thing as it appears must also be something apart from sensibility. Kant says, “the word appearance must be recognized as already indicating a relation to something, the immediate representation of which is, indeed, sensible, but which, even apart from the constitution of our sensibility .l.l. must be something in itself ” (A252–53, pp. 269–70). Note that he is making two claims here. Firstly, an appearance (a proper reaction) must be related to a thing represented sensibly (a thing as it appears), which, second, must also be something in itself. Of course at this stage his account of cognition still involves appearances or proper reactions, only now these proper reactions pertain to things as they appear, which also must be something in themselves. If this is correct, then Kant’s theory of cognition at this stage still involves uncognizable reality, so that the thing in itself is still not a mere limiting notion. It does not, however, involve the intelligible represenTake three steps—with respect to the thing as it appears there it is proper to be in the course of a series of reactions, each determining the next, ending with presently reacting “upstream,” “downstream.” 12. For the denial of a positive sense of noumena, see A252, p. 270. 13. He goes on to say it must therefore be “an object of .l.l. the understanding” (i.e., noumenon in the positive sense), but this is only his setup of the problem that apparently it must be such an object. The section, indeed, is concerned to reject that idea.

158

t h i n g s i n t h e m s e lv e s

tation of a plurality of objects (noumena in the positive sense). The reason is that although each of a plurality of things as they appear must be something in itself, it doesn’t follow that they are each different things in themselves. The representation of the difference, or the plurality, is totally within the appearing (different for appearing at different stages of the spatial shift of attention). Hence Kant can still use the terminology of the “transcendental object” (undivided reality that is one and the same throughout all our cognition) to express the in-itself reality of things that appear. Only now, as opposed to the early view, the transcendental object (what is so) is not the “direct” basis of proper reactions. Rather, what is so is, in relation to our spatio-temporal intuition, a plurality of things as they appear, which in turn are the bases of proper reactions.14 Finally, at this stage Kant is no longer a phenomenalist. Besides proper reactions, there are things as they appear, which are not proper reactions at all but what proper reactions pertain to. Indeed, the representation of causation (of the necessary advance of perceptions) now is that in regard to a thing as it appears (at a certain stage of the subject’s spatial shift) it is proper to be in the course of a series of reactions (each necessarily advancing to the next), ending, say, with the reactions ship upstream, ship downstream. The thing as it appears, then, has a history of reactions that are proper in regard to it, and so is a substance.15 Henry Allison is one commentator who conflates this middle view of Kant’s with the late view (according to which the concept of the thingin-itself is a mere limiting notion).16 The view he wants to hold is a double-aspect (rather than a two-worlds) view, according to which things may be considered in relation to intuition but also as they are in themselves. This, for Allison, is an “epistemic” distinction of two ways of considering things, not an ontological distinction of two ways of existing. Its being thus epistemic would seem to imply that Allison’s view is basically the late one, according to which the consideration of those things 14. It is at this stage that Kant is most vulnerable to the charge of being incoherent in speaking of a plurality of things in themselves, though plurality is a category, and categories are not supposed to apply to uncognizable reality. We see, however, that Kant himself goes out of his way to deny that we must speak of a plurality of things in themselves. 15. Back in the section of the A edition Deduction (A111–14, pp. 138-40), Kant says, “The a priori conditions of a possible experience in general are at the same time conditions of the possibility of objects of experience.” His point, I believe, is that in order to represent causation, and so the necessary advance of the time series, objects or substances are required. That these objects are required to represent possible experience (proper reactions in the necessary advance of time), constitutes a refutation of phenomenalism. 16. Henry Allison, Idealism and Freedom (Cambridge: Cambridge University Press, 1996), part 1, chap. 1.

On Things in Themselves

159

apart from intuition is merely a limiting notion. But now I claim that it is only at the middle stage that Kant uses the terminology of a double aspect, viz., the terminology of things as they appear in relation to intuition versus those things in themselves. At the middle stage, however, this precisely expresses an ontological distinction, not an epistemic one. Things in themselves are something outside of the progression of intuition, while they appear only within it. That Allison is, indeed, deriving his two-aspect talk from the middle view is made clear by the fact that he exactly uses the middle-stage section of the phenomena and noumena chapter to deny (as Kant denies) that there is a plurality of things considered in themselves isomorphic to things considered as they appear. Although the early view of cognition leads to a two-worlds view (appearances versus the transcendental object), and the middle view of cognition leads to an ontological “double-aspect” view, Kant simply never holds Allison’s epistemic “double-aspect” view of things considered in relation to intuition versus those same things considered in themselves.17 What I wish to claim now is that the issue of the divisibility of substances forced Kant to abandon his second account of cognition. Once he had reintroduced things into the sensible realm in his middle view, he would have naturally been led to Leibniz’s divisibility argument against the existence of spatial things. That argument, roughly, is that a spatial entity is an aggregate of parts, where each part is an aggregate of parts, so that in totality all there is are aggregates or pluralities of aggregates ad infinitum. Hence there is no true unity in the spatial realm. Kant, I contend, believed he could avoid this argument in his middle view by avoiding the idea that there is a totality of parts. Thus the divisibility of a thing as it appears would simply consist in its being subject to an ongoing or limitless, but not total, division into other things as they appear (parts). The ongoing divisibility of space, like its continuity, is a matter of construction. We can take the relevant construction here to be bisection. Thus, after delineating the whole object, we can introduce parts by bisecting that delineation, and parts of parts by bisecting the bisection, etc., where the parts only appear (are only exhibited) within or relative to the spatial construction. The rule, then, would be that with regard to a thing as it appears, it is legitimate or proper to keep bisecting into things as they appear at any of those stages of the bisection. In this way Kant thought that at each stage of bisection there was unity (since no further parts appear at that stage), so there is never (in sum) 17. It seems to me that Allison’s conflation regarding the thing in itself is part of a tendency on his part to conflate the Copernican revolution with transcendental idealism.

160

t h i n g s i n t h e m s e lv e s

a totality of actual mere pluralities. This middle-stage solution of Leibniz’s problem would have been in a Second Antinomy in which both thesis and antithesis were true. In the sensible realm there are no simples (every part, though so far a unity, is potentially further divisible), while intellectually represented substance would, to have unity, require simples. This is obviously not the Second Antinomy as we officially have it in the Critique.18 The reason, I suggest, is that Kant came to see that his solution was incoherent. There can be ongoing or limitless or unfinishable rules (the “potential” infinite), but there cannot be limitless pluralities of individuals. A plurality (set or aggregate) is either finite or actually infinite. It makes no sense to say there are limitlessly many (or ongoingly many) things as they appear. But this is exactly what the rule of bisection above says. Kant’s answer to Leibniz’s argument against spatial substance, then, collapses into the absurdity of potentially infinite pluralities. Kant needs some sort of diversity of substances for causation (for diverse causal series), but this diversity cannot be a plurality of entities, not even as they appear. His solution, I suggest, is what he calls the permanence of the real (A143, B183, p. 184). Sensation, for Kant, gives us the consciousness that we are affected (B207, p. 201), and the real (“realitas phaenomenon”) is what corresponds to sensation (A168, B209, p. 203), and hence that which affects us or influences us. As I move around spatially (shifting the flow of my spatial attention) I am variously affected. Upon first being affected, it may be proper to react “red,” while upon being affected further along in the spatial shift, it may be proper to react “green.” Different reactions, then, are proper, depending on where I am affected in the construction of space. We have thus diversity of affection (diversity in being affected) consequent simply upon the shift in spatial construction, and so a diversity without a plurality of entities. We can represent this as follows: (3) .l.l. upon being affected, it is proper or legitimate to react in such and such a way, where the three dots stand in for some spatial construction, such as taking five steps to the right. The permanence of the real (or, for our purposes, the lastingness of the real) would consist in the fact that reactions in past time are proper upon being presently affected. So, for example, if we represent that upon being affected it is proper to be in the course of 18. The official Second Antinomy, with its both-false resolution, is completed after Kant abandons the middle view. For the idea that this Antinomy involves a switch between incompatible views, see Norman Kemp Smith, A Commentary to Kant’s Critique of Pure Reason (London: Macmillan, 1918), 491.

On Things in Themselves

161

(at the tail end of) a series of reactions, each advancing necessarily to the next, we are representing what presently affects as the subject of a series of reactions ending with a present one. This indeed, is to think of what presently affects as having existed. Upon being affected further along in the spatial construction, it may be proper to be in the course of a different series of necessarily advancing reactions, so that we have diverse causal series with respect to a diversity of the permanence (or lastingness) of the real, all without the representation of a plurality of entities that led to the failed response to Leibniz’s argument against spatial substances. Indeed, now Kant can represent the rule of potential infinite divisibility as follows: Upon being affected, it is legitimate to keep bisecting and as you do to keep being affected. That the rule is to keep being affected signifies that there is no failure of affection upon any bisection (viz., there are always further parts potentially), but the rule does not represent a plurality of things at all, and so not a potentially infinite plurality of things. Roughly, it says that being affected is ongoing, not that there are ongoingly many affections. The new basis of rules in the permanence of the real, or phenomenal substance, solves both the problem of causation from the early view, and the problem of divisibility from the middle view. Like the middle view, this late view is not phenomenalism, since being affected licenses a series of reactions with regard to it, and so indicates a reality that endures and that proper reactions pertain to. In this late view, the basis of proper reactions is the phenomenally real that affects within spatial construction. An example of the conflation of the early view with the late view is the whole doctrine of “double affection”19 according to which both the thing-in-itself and, perhaps in another relation, phenomenal substance affect us. The real that affects (realitas phaenomenon) on this late view is certainly not the transcendental object (the unintuited reality) of the early stage, since this real is exhibited and individuated by spatial intuition. Reading the early view, then, as a transcendental affection, and combining it with this late view, gives a double affection. Most importantly for our purposes, the late view of cognition, for the first time, completely severs the theory of cognition from (the idea of) uncognizable reality. The nonarbitrary basis of rules for proper reactions (appearances) is neither uncognizable reality itself (the transcendental object of the early view), nor things as they appear, which must also have an uncognizable aspect of being something in themselves out19. See Erich Adickes, Kant’s Lehre von der doppelten affection des Ich als Schluessel zu seiner Erkenntnistheorie (Tübingen: J. C. B. Mohr, 1929). I am grateful to Paul Guyer for this reference.

162

t h i n g s i n t h e m s e lv e s

side the progression of intuition. Rather, the basis is phenomenal substance that exists totally within the progression of intuition. That leaves only a limiting function for any notion of uncognizable reality. The section of the phenomena and noumena chapter added in the B edition (B305–9, pp. 266–70) to replace the A edition section that expresses the discarded middle view states this. Kant says, “The understanding, when it entitles an object in a certain relation [to intuition] mere phenomenon, at the same time forms apart from that relation a representation of an object in itself ” (emphasis added). The understanding’s “title” of “mere phenomenon,” then, is a limiting notion formed by contrasting phenomena with noumenon in the negative sense, which is “a something in general outside our sensibility.”20 The section of the phenomena and noumena chapter from A254, B309, pp. 271–A260, B315, p. 275 that was retained in the B edition also expresses the mere limiting function.21 Thus the concept of a noumenon is said to be “a merely limiting concept, the function of which is to curb the pretensions of sensibility.” Indeed, this late distinction between phenomenal reality and noumenon in the negative sense is just the distinction involved in Kant’s announcement of a Copernican revolution. Kant, recall, is going to investigate a priori the structure of reality, by seeing how the object must conform to our cognition. The limitation of this method is that it gives us a metaphysics of cognizable reality, not reality per se. Of course, in the passage just quoted Kant has his specific theory of cognition in mind, viz., that it involves sensibility (forms of intuition) governed by rules. Other than this specificity, however, the late distinction that curbs the pretensions of sensibility is just the Copernican distinction that curbs the pretensions of a theory of cognition as a theory of reality per se. I have tried to argue that all accounts, other than the sheer limiting account, are either remnants of discarded views, conflations of views, or misreadings of views. Thus the two-worlds view, with things in themselves affecting us, is derived from Kant’s early view, and misreads the 20. Note what has happened to the transcendental object (i.e., this something in general outside our sensibility). Originally, it itself was the basis of rules. Then, at the middle stage, it, in relation to intuition as a plurality of things as they appear, was the basis. Now it is merely a limiting concept. 21. That this section of the chapter is late is clear, I believe, from what Kant says at A258, B313, p. 274: “When, therefore, we say that the senses represent objects as they appear, and the understanding objects as they are, the latter statement is to be taken not in the transcendental, but in the merely empirical meaning of the terms.” Kant, I contend, is going out of his way to say here that the terminology of the middle stage (things as they appear versus as they are) has only an empirical use, so that he is rejecting his middlestage distinction.

On Things in Themselves

163

relationship as one of affection (problematic transcendental causation), whereas it is really a relationship of “topic” (viz., the legitimacy of our reactions is in regard to what is so). The two-aspects view (things as they appear versus those things as they are in themselves), in its ontological guise, is derived from Kant’s middle view, but it does not require a problematic representation of a plurality of things in themselves. The two-aspects view, in its epistemological guise (two ways of considering or reflecting on things—as they appear versus as they are in themselves), collapses into the mere limiting view, unless it confusedly reads the middle view that way. The double-affection view simply conflates the early and the late views.

9

Kant’s Proof of Transcendental Idealism in the First Antinomy

1 . t h e d o c t r i n e o f t r a n s c e n d e n ta l i d e a l i s m Kant thinks the Antinomies establish transcendental idealism as against an opposing view that he calls transcendental realism. The idealist holds that the world “is only to be met with in the regressive synthesis itself,” while the realist views the field of appearance “as a thing given in and by itself prior to all regress” (A505, B533, p. 448).1 It is not that the realist holds that there is no regressive synthesis. Rather, he holds there is also the sensible world standing against the synthesis waiting to be met with by the synthesis. The idealist, by contrast, holds that the world exists “only in the empirical regress of the series of appearances, and is not to be met with as something in itself ” (A505, B533, p. 448, emphasis added). We have first to make clear, then, what the regressive synthesis is. The regress, for Kant, is a regress of appearances arising in space and time. Of course space and time are not themselves objective; they are the forms of the regress itself, not realities standing against it. I suggest now that a spatial or temporal regress means, for Kant, a spatial or temporal procedure. Thus, drawing a line of twenty units is a procedure, as is taking five steps to the right. Perceptions may arise in the course of either of these two procedures. To say that space exists only in (only as the form of) a regress, then, is to say that space exists only in such procedures. In particular, it is to deny that space is that through which, or in which, these procedures are carried out, for that would imply that space also exists apart from the regress as an objective expanse within which the regress takes place. Thus the “motions of the subject” (B154, p. 167) in drawing a line or taking five steps fully constitute space, and so do not take place through (a corresponding objective) space waiting to be met with or reached by these motions. In the Aesthetic Kant’s argument for this position turns, I believe, on the contention that conti1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition.

164

Transcendental Idealism in the First Antinomy

165

nuity (and so expansiveness) exists only in flowing constructions (that is, motions of the subject). Continuity is that kind of seamlessness of a whole whereby the whole cannot be composed of parts or elements at all, but rather is prior to the parts (A25, p. 69). Now, in a flowing construction, such as drawing a line, the flow is not composed of (done by) a series of stops and starts, and so is not composed of points or subintervals. Rather, points and subintervals are subsequently imposed (marked off) on the line (the whole). On the other hand an objective reality is such that it exists only if all its parts or elements exist, and so, in this sense, is composed of them. Thus, if space were an objective reality, it would be composed of components, and so, for Kant, would not be continuous (or expansive) at all. It follows, then, that the flowing construction is not a motion through a corresponding objective expanse. Indeed, such an expanse would be composed of components, and then the flow itself would have to “get through” all these components, destroying thereby the continuity of the flow (and leading to Zeno’s paradoxes). It is only by flowing motion constituting space, rather than taking place in it, that expansiveness is possible. Similarly, in moving five steps to the right, each step is a flowing motion of the foot, so that the procedure is just a series of flows. For the transcendental idealist space does not exist only in procedures that are actually carried out. Rather, it exists in “the rule of the advance” (A496, B524, p. 442), viz., in the propriety of procedures. Thus, it may be proper to take 254 steps whether or not I actually do so, and space extends or exists as far as the propriety of spatial construction. In the Transcendental Deduction Kant says that space is cognized or thought by the synthesis of the productive imagination (spatial construction) being brought under the unity of apperception (A117– 19, pp. 142–43). Apperception, for Kant, is the faculty of understanding (the faculty of cognition or thought), and this faculty, Kant says, can be characterized as “the faculty of rules,” which “fruitfully” and “more closely to its essential nature” characterizes the faculty of understanding (A126, p. 147). If so, then the unity of apperception is the unity of a rule, and hence spatial construction is cognized or thought by being brought under a rule (of how it is proper to construct). Note that in thinking it is proper to spatially construct a certain way, we are thinking of space existing only in the (proper) construction. The rule is to draw a line of five units, not to draw it through a space that objectively exists. For the transcendental idealist, then, space exists only in the rule of the regress, and is nothing outside that rule. So far, then, transcendental idealism regarding space comes down to a constructivist view of space. A constructivist regarding numbers holds

166

t h i n g s i n t h e m s e lv e s

that numbers exist only in procedures (say, as termini of counting procedures), and allows this to encompass not only procedures actually carried out but procedures it is proper to perform as well. Further, he holds that these procedures don’t traverse (pass through, match, or correspond to) an objective series, say, of Platonic numbers. Similarly, Kant holds that space exists only in flowing procedures that are motions or acts of the subject (B154, p. 167), including procedures that it is proper to perform, whether they are performed or not, and that these procedures don’t traverse objective space. Time, like space, is continuous and so must as well exist only in flowing constructions. Kant attributes time as well as space to motions or acts of the subject (B154, p. 167). A temporal (or a temporizing) construction would be something like a series of downbeat gestures by a conductor to pace (or time) the play of the orchestra. Each such gesture is a flow, and the series is a marking-time procedure, so long as it is understood that there is no objective (extra-constructive) time that is marked. Rather, time exists only in such procedures. Note that this construction (as opposed to drawing a line or taking five steps) “gets nowhere” spatially as the construction proceeds. The downbeat interval is indeed spatial, but it is repeated through the series. Although the spatial extent repeats, the temporal extent increases. This is what Kant means, I believe, by saying that although time is “figuratively” represented spatially, in the time construction we “attend to the succession” (B154, p. 167), viz., to the increase in production, even if it doesn’t increase spatially. Time, like space, is brought to the unity of apperception by being brought under a rule, and so the transcendental ideality of time is just the constructivist doctrine that time exists only in proper temporizing constructions. For Kant to be consistent, this ideality must include or encompass past time. Now, there cannot be rules for going back in time, but there can be rules for presently being in the course of temporizing procedures; viz., it can be proper for me now to be up to stage k of a temporizing procedure even if I haven’t been actually constructing. We shall see below how Kant thinks it can be proper to thus be in the course of a procedure such as a series of downbeats. For now, I suggest that such rules for presently being in the course of temporizing represent past time constructively, since they represent my present situation as one of being past or beyond stages of time marking, without representing an objective (extra-constructive) past time through which these stages take place. In this manner, the transcendental ideality of time (even past time), like that of space, is a matter of time existing only “in the regress,” i.e., only in rules for procedures (including rules for being in the course of procedures).

Transcendental Idealism in the First Antinomy

167

What exists in space and time, now, is what arises within the regress or within the course of proper spatio-temporal constructive procedures. What thus arises we can call “appearances” in order to signify that what is given is what exists (appears) only within the regress. As Kant puts it in the Aesthetic, space “can be ascribed to things only in so far as they appear to us, that is, only to objects of sensibility” (A27, B43, p. 72). Space, that is, does not pertain to things apart from their being intuited in the progression of spatial construction. This is claimed, by Kant, to be a conclusion from the fact that space does not exist objectively. To see how this conclusion follows, suppose one is a constructivist regarding serial order; i.e., one holds that such order exists only in relation to ordered procedures such as reciting numerals. Items on my desk, then, get a serial order by their arising in the course of such a reciting procedure. Thus I point to an item and recite “one,” point to another item and recite “two,” etc. It is items, so far as they are pointed to, that arise in the course of the procedure, and so have a serial order by (merely thus) having their place within the procedure. On the other hand, we cannot represent the items themselves, apart from pointing, as having a serial order, since apart from pointing the items themselves don’t arise in the course of the procedure, and so if they have a serial order, it would have to be by their having their own intrinsic order (which, then, the procedure or regress matches or corresponds to). But this latter contradicts the premise that serial order does not exist extra-constructively. Similarly now, it is the intuition or perception of (or the attention to) items that arises (occurs) in the course of spatial construction, and so has position merely by having a place within the spatial procedure. Items apart from intuition, since they don’t arise in the course of the construction, could only have a place relative to the construction if they had their own intrinsic place in extra-constructive space that matches or corresponds to the spatial construction. This again contradicts the premise that (continuous) space does not exist extra-constructively. Exactly similar remarks pertain to existence in time; viz., time pertains only to what arises within the course of the temporal regress (temporal constructive procedures). Although space and time apply only to what exists (arises) within the regress, this doesn’t imply that they apply only to possible perceptual states of ours. For Kant, perception involves sensation (a state of the subject), but also the consciousness that we are affected (B207, p. 201). It involves, that is, both sensation “and the real which corresponds to it in the object” (A166, p. 201). Upon drawing out a line or taking five steps to the right I get affected. Being thus affected arises at the tail end of the procedure. This doesn’t imply that being thus affected involves

168

t h i n g s i n t h e m s e lv e s

only a momentary state (sensation), since I might conceptualize what is then given (the real that then affects) as being permanent or having a duration (and so, as being a substance). To conceptualize it in this way is just to think that what affects (the “real” that corresponds to sensation) has a history or has existence in past time. The thought of past time for a constructivist, recall, is just the thought of its being proper to be in the course of (rather than at the beginning of) a temporizing procedure. Thus, to think of what presently affects as having past existence is just to think that it is proper to be in the course of a temporizing procedure that is geared to it (geared, that is, to what affects). A temporizing procedure that governs keeping track of what affects is a procedure geared to what affects, and hence, finally, to think of what affects as a substance is to think the following: (1) Upon being presently affected, it is proper to be up to stage k in temporizingcum-tracking. Note that (1) represents what presently affects as having existence in past time without going outside the regress (without going beyond what arises, or remains, within the course of spatio-temporal procedures). Thus being presently affected itself arises at a stage of, say, drawing a line or taking five steps, and past existence in time arises throughout the compass of a procedure that temporizes keeping track. Indeed, not only can we represent substance without going outside the regress, but, Kant holds, we must represent substance if we are to represent time at all. Substance is a category, and categories for Kant bring the synthesis of the productive imagination (the spatio-temporal construction) to the unity of apperception or the unity of a rule (A119, p. 143). A category, that is, is a concept required for thinking the regress of space and time. As we mentioned above, rules pertaining to past time are rules for presently being in the course of temporizing procedures. Now, it cannot just be proper to be up to k in a series of downbeats, since a procedure has a fixed order and begins from the beginning. However, if the procedure is geared to something, and what it is geared to is presently so far along, then it can be proper to be that far along in the procedure. So, for example, a cake-baking procedure of mixing ingredients in order is geared to getting a cake baked. If I come in upon a bowl with the first three ingredients mixed, so that what the procedure is geared to is so far along, then it is proper for me to be that far along (up to putting in the fourth ingredient) in the procedure. Similarly now, if (and only if) a temporizing procedure is geared to something that is presently so far along in its existence (its track-ability) can it be proper to be in the course of the temporizing procedure.

Transcendental Idealism in the First Antinomy

169

Thus (1) above not only represents what presently affects as substance but only by thus representing what presently affects as substance can past time (the propriety of being so far along in a temporizing procedure) be represented. In this manner, substance is a category; i.e., a concept required for thinking the regress of past time or, equivalently, for having a rule for being in the course of temporizing. That Kant clearly means to include categorial structure as a component of the regress in the Antinomies is made clear when he says that what arises in the course of the spatio-temporal regress in so far as it is “determinable according to laws of the unity of experience” are objects (A494, B522, p. 441). By “laws of the unity of experience” Kant means the categories (or the Principles), and it is these by which we think objects in the regress. Thus the regress, for Kant, includes objects of experience, not just perceptual states. It is only so far as perceptual states are subject to the categories that they are cognizable. For example, to represent a possible perception “in the past” we must represent it as arising in past time. This, in turn, requires representing it as at an earlier stage of a temporizing procedure that we are presently in the course of. In terms of (1) above the representation would be: (2) Upon being presently affected, it is proper to be up to stage k in temporizingcum-tracking-cum-first perceiving so-and-so. Since, in (2), perceiving is represented as what is proper with first or initially keeping track of what presently affects, (2) represents the perceiving as how what presently affects was, and so represents the perceiving as of a state of substance. In this manner, perceptual states (proper reactions) are cognizable only as states of substances. In sum, then, both perceptual states and the substances they pertain to are represented as arising within the regress (as what is intuited in the compass of spatio-temporal construction). This distinguishes transcendental idealism from any sort of phenomenalism, since representing appearances (i.e., possible perceptions) requires a category of substance according to which what is given in experience (as one is presently affected) is not a perceptual state, but an enduring object. To sum up, Kant’s doctrine of transcendental idealism has two components. First, space and time exist only “serially” in the regress; viz., they exist only in rules for spatializing and temporizing constructive procedures. They are nothing apart from the activity of the subject. Second, what exists in space and time exists only as arising within the spatio-temporal regress, viz., exists only as arising in the course of rules for spatio-temporal construction. The argument for the first component is

170

t h i n g s i n t h e m s e lv e s

that space and time are continuous (expansive), and continuity exists only in (rules for) flowing constructions. The argument for the second component is that what stands against or outside spatio-temporal construction (what doesn’t exist only as arising within it) can only be determined with respect to such construction, by having its own intrinsic, corresponding (spatio-temporal) determination, which latter (that is, extra-constructive space and time) is impossible. Kant establishes both of these results in the Aesthetic. That what thus arises within the regress are not only appearances (possible perceptions), but also objects of sensibility (substances) that appearances pertain to, Kant establishes in the Principles. The categories (including the category of substance) don’t take us outside the regress but rather are constitutive of it (in first enabling rules for spatio-temporal construction to have full scope). Since the two components of transcendental idealism are established in the Aesthetic, it makes sense for Kant to say of the Antinomies that they offer an indirect proof of what he has already proved in the Aesthetic (A506, B534, pp. 448–49). Since, further, the proof in the Aesthetic turns exclusively on the continuity of spatio-temporal expansiveness, the indirect proof in the Antinomies, if it is to be further corroboration, must be independent of this issue of continuity. Before seeing what this proof is, we first need to get clearer about what the transcendental realist holds.

2 . t h e d o c t r i n e o f t r a n s c e n d e n ta l r e a l i s m In the Critical Solution of the Antinomies Kant contrasts his view with transcendental realism as follows: “The series of conditions is only to be met with in the regressive synthesis itself [that is, transcendental idealism] not in the [field of] appearance viewed as a thing given in and by itself prior to all regress [that is, transcendental realism]” (A505, B533, p. 448). The realist, that is, views the appearance, or what is given in the regressive synthesis, as also a thing prior to all regress. The realist does not deny the regress but denies rather that the world exists only in the regress. Thus, for example, the realist doesn’t deny that it is proper to move about and thereupon be affected; but he holds also that what stands outside this procedure is an object at a position of extra-constructive space, which object is there waiting to affect us. The realist treats affection, which arises only in the regress, as affection by an object that stands against the regress or exists outside the regress, and he treats the spatial construction (moving about) as taking place through a matching objective (extra-constructive) space. Whereas the idealist holds only that

Transcendental Idealism in the First Antinomy

171

(3) It is proper to take five steps, to be affected, and to perceive so-and-so, the realist holds that (4) It is proper to take five steps, be affected by the object that is there (at the corresponding position of objective space), and perceive so-and-so. For the realist there is a cause of the advance of the regress; viz., the reason it is proper to take five steps and be affected is that there is an object at a place (which place is reached by taking five steps) waiting to affect us. In other words, the existence of an entity in objective space is the ground or basis of the regress. For Kant, by contrast, the only thing that exists outside the regress, which is ground or basis of the regress, is the transcendental object (A494, B522, p. 441). The transcendental object is something like God or the ultimate basis of what there is that is responsible for there being the full range of phenomenal reality. It is not spatio-temporal at all and is not a plurality of entities. This is in contrast to the realist, for whom the bases of our various procedures of moving about and being affected are a plurality of different spatial objects at different places (that these various procedures get to). Kant characterizes transcendental realism, which is the doctrine that the world “remains even if I suspend the series of its appearances” (A504, B532, p. 447) as a transcendental illusion (A504, B532, pp. 447–48). It is therefore no arbitrary doctrine. Indeed, by our interpretation it is the quite seductive doctrine that it can only be proper to move about and be affected because there is (extra-constructive) space to move through, and objects existing in this space to be affected by. This doctrine, at once realist and completely seductive, is just the doctrine, first, that space is not merely a form of intuition but is also something objective, and, second, that spatial objects are not mere appearances (arising merely in the course of spatial construction), but also things existing in themselves (in objective space) outside the regress. We note that the first component of this doctrine of transcendental realism is exactly Trendelenburg’s “overlooked alternative.” 2 If my understanding of transcendental realism is correct, then, rather than Kant overlooking this alternative, the argument of the Mathematical Antinomies is expressly directed against it. A similar characterization of transcendental realism applies to the temporal regress. For the idealist, past time exists only in rules for being in the course of temporizing procedures, and what exists in past time (i.e., states of substance) is only what arises within these proce2. For a discussion of the Trendelenburg objection, see Hans Vaihinger, Kommentar zu Kants Kritik der Reinen Vernunft (Berlin: Union Deutsche Verlagsgesellschaft, 1892), 2:268.

172

t h i n g s i n t h e m s e lv e s

dures (A495, B523, p. 442). This is what is expressed by (2) above, which we repeat here for convenience: (5) Upon being affected, it is proper to be up to stage k in temporizing-cumtracking-cum-first perceiving so-and-so. In (5) time exists only as a construction or procedure that it is now proper to be in the course of, and the past state of what presently affects is just a perception proper at (arising at) the first stage of tracking what presently affects. Once again, the realist doesn’t deny that it can be proper to be in the course of temporizing procedures that begin with proper perceptions. Rather he holds that it is proper because there are past states of substances existing outside the regress (at past objective times) to perceive (which are proper to perceive). Whereas the idealist holds (5), the realist holds that the full representation is (6) Upon being affected by that object, it is proper to be up to stage k in temporizing-cum-tracking from first perceiving the state of that object at that initial time. The realist then holds that time is not merely a form of intuition but also something existing objectively (so that temporizing procedures match or correspond to the unfolding of extra-constructive time), and he holds there are past states of things existing at those objective times, rather than just proper perceptions arising within proper procedures. For the transcendental realist, procedures are not just grafted onto a world that he represents independently of those procedures. The reason for this is that, for Kant, purely conceptual representation of objects is empty. To represent an object in space and time independently of representing it as met with or standing against our procedures is to represent it as the object having such and such spatio-temporal position and such and such properties; viz., it is to represent it by purely descriptive (conceptual) individuating conditions, even if these conditions include spatio-temporal predicates. It is exactly such representation by purely conceptual individuation that Kant considers empty.3 Because of this, the transcendental realist, who also accepts the emptiness of purely conceptual representation,4 represents an (extra-regressive) 3. Representation by conceptual conditions is always of “infima species,” not of individuals. Thus even a complete description (including descriptions of spatio-temporal position) does not discriminate between a purely possible or a fictional individual as opposed to a real one, since the former can satisfy the description as well. It is only by the domain (the potential satisfiers of the description) first being fixed to the real world that descriptions “refer” to individual real objects, as opposed to a completely specific “type” of object. But descriptions themselves don’t fix their own domain. 4. Kant rejected purely conceptual representation after the Inaugural Dissertation,

Transcendental Idealism in the First Antinomy

173

object in space and time in terms of the regress or procedures which meet it. Thus, in the italicized portion of (6) above, he represents the object standing against the procedure as “that object,” where “that” is a relative pronoun that gets its significance or purport from the regress (the nonitalicized portion of [6]). In this manner, it is not purely conceptually represented independently of the regress but is represented exactly as standing against the regress. Similarly the temporal position of the past state is represented as “that initial time,” where again “that” is a relative pronoun signifying nothing apart from the standing against the regress that fixes its purport. Although the objects at their spatio-temporal positions are represented as being (having existence) independent of the regress, their representation is via their being met with or reached by the regress. In sum, the transcendental realist agrees with Kant that all cognition or representation of objects, even the realist’s objects standing against the regress, requires the forms of intuition and the procedures for empirically intuiting. The difference is that the realist uses these procedures to represent what corresponds to them, while for the idealist, what arises in the procedures is all there is. The nature of the dispute can perhaps be clarified by considering what the issue would be as regards mathematical constructivism. The “pure” mathematical constructivist (analogous to the transcendental idealist) holds that numbers exist only as termini of possible (i.e., proper) countings. The “pure” mathematical objectivist or Platonist holds they exist apart from such countings as some sort of extra-procedural linear order. But the Platonist has the issue of the cognition (“grasp”) of these objective entities. Even if they somehow exist, it is not at all clear how we can think of or cognize them. If the Platonist now says that we cognize them only in terms of possible counts (as what are met with or matched by our proper counting procedures), then he maintains his realism (viz., that they exist standing against our procedures), while enabling their cognition. It is this latter sort of Platonist who is analogous to the transcendental realist.

3. the argument of the first antinomy We turn first to what the “antinomy” is that the transcendental realist is argued to fall into. Since the transcendental realist holds there are entities corresponding to (entities to be met with by) our procedures, he must hold that in sum there is either a finite totality of such entities and this rejection colors the entire Critique, but is not the specific target of the Antinomies.

174

t h i n g s i n t h e m s e lv e s

or an infinite totality of such entities. In particular, he cannot hold that there are “indefinitely many” such entities, where this signifies some sort of nontotality. The reason is that there simply is no such thing as an indefinite sum or collection of entities. There may be specifications of totalities or collections such that it is indeterminate whether certain objects belong to the specified collection, but there cannot be collections that determinately have indefinitely many members. In any case, there is no indeterminacy in the realist’s specification “object to be met with by a regress,” so that the collection specified by this condition is determinately finite or infinite. By contrast, the argument of the First Antinomy shows there can be neither an infinite totality of entities standing against the regress (the first thesis), nor a finite totality of such entities (the first antithesis). The antinomy, then, is that the full range of entities waiting to be met with by our procedures must be either finite or infinite, but it cannot be either (A506, B534, p. 449). Thus the nature of the antinomy is that it is simply a contradiction, and the realist winding up in the antinomy shows that transcendental realism is literally incoherent or inconsistent. We consider now the arguments for the First Antinomy, beginning with the arguments regarding the extent of entities standing against the spatial regress. The first thesis pertaining to space argues that the world (as understood by the transcendental realist) in space cannot be infinite. What would the transcendental realist’s representation be of an infinite world in space? To begin with, he must represent this world in terms of the spatial regress. For convenience, we limit ourselves to the extent of the world in a certain direction in space. What we need is a representation of the objects there are not just upon taking five steps (as in [4] above), but the objects there are in relation to the full scope of taking steps in a certain direction. This full scope, since space is not finite, is just that it is proper to keep on taking steps. Indeed, for the constructivist, universal generality is just the propriety of a potentially infinite (i.e., endless, or unfinishable) construction or procedure. Thus, the full scope of numbers is represented not by the propriety of any particular count or recitation but by the propriety of an ongoing or incompletable recitation. As Kant says in the Observation on the First Thesis, “The true transcendental [i.e., the constructivist] concept of infinitude is this, that the successive synthesis of units required for the enumeration of a quantum can never be completed” (A432, B460, p. 401). If so, then the representation by the realist of the full scope of entities is the representation of what entities are met with by the full scope of (the endless) spatial regress, viz.:

Transcendental Idealism in the First Antinomy

175

(7) It is proper to keep taking steps and to keep being affected by the entities that are to be met with at any of those places.5 However, (7) is impossible since it represents “ongoingly many” or “unfinishably many” entities, which is absurd. Recall that the realist doesn’t antecedently (independently of the regress) represent a totality of entities that he then subsequently represents as reachable by the regress.6 Rather, the entities are represented exactly as what (collection) is reachable by the regress they stand against. Since, then, the regress is ongoing or limitless, the entities thus represented only relatively to it form an ongoing or limitless plurality. But there cannot be a limitless plurality of entities. The potential infinite pertains only to procedures or rules, not to objectively existing individuals. Since the realist thus has to represent an actual infinite totality, and since he has to represent it in terms of the regress, he has no recourse but to represent the regress itself as actually total. Thus, to avoid the absurdity of (7) he is forced to posit that (8) It is proper to take all steps, while being affected by those entities that are to be met with at any of those places (or steps). But now the realist has been forced to represent an infinite completable procedure (synthesis). As Kant says in the first thesis, “In order, therefore, to think, as a whole, the world which fills all spaces, the successive synthesis of the parts of an infinite world must be viewed as completed” (A428, B436, p. 398).7 But the idea of completing an infinite procedure, for Kant, is impossible. Kant’s argument in this first thesis regarding space, then, is quite simple. The only notion of infinity applicable to rules or procedures is potential infinity (incompletability), whereas the only notion of infinity applicable to extra-constructive entities is an actual total infinity. But the transcendental realist has to represent extra-constructive entities in terms of the regress, which leads either to an impossible potential infinity of entities (as in [7]), or an impossible actual infinity of a procedure (as in [8]). Thus the trans5. Formula (7) actually represents entities to be met with at each stage, which is more specific than representing infinitely many entities. However, this is irrelevant to the nature of Kant’s argument. 6. In this case, he could first represent an actually infinite totality of entities and then represent limitlessness or ongoingness in the regress that meets these infinitely many entities. The impossibility of purely conceptual (regress-independent) representation is thus crucial to Kant’s argument against the transcendental realist. 7. Kant actually goes on to say that to view the synthesis as completed “an infinite time must be viewed as having elapsed in the enumeration” (A428, B436, p. 398). His point, I believe, is that to think of the synthesis as completed is to think of it as being done with, so that there is a first time at which it is done with. The question then becomes, what was I doing right before that first time?

176

t h i n g s i n t h e m s e lv e s

cendental realist cannot represent the infinitude of the world in space. Turning now to the first antithesis regarding space, one might think the problem of the thesis will no longer arise, since only finitely many entities have to be represented, and hence a finite regress would be sufficient. It is not true, however, that a finite regress is sufficient; viz., the following does not represent the finitude of the world in space: (9) It is proper to keep taking steps up to n and to keep being affected by those entities at any of those places be met with. Obviously, (9) only represents a finite collection of entities within a finite portion of space, which says nothing about the finitude of entities in all space. The realist has to represent that there are no entities beyond n steps, as, for example, by positing that (10) It is proper to keep taking steps beyond n and to keep failing to be affected by entities at any of those places. The realist is still representing objective places as empty, since his contention is that space exists objectively as what our regress meets or goes through. But now there can no more be an indefinite totality of places than an indefinite totality of entities. Potential infinity pertains only to constructions or procedures, not to collections of extra-constructive individuals, however “abstract” or “concrete” these individuals are. The transcendental realist, then, must represent empty space (beyond n steps) independently of the regress, augmenting (9) above to (11) It is proper to take steps up to n being affected by those objects met with at those places and S is otherwise empty, where “S” signifies space as a whole.8 Kant’s comment in the first antithesis regarding this realist representation of the finitude of the world in space is that “Things will therefore not only be related in space, but also related to space .l.l. [but] the relation of the world to empty space would be a relation of it to no object” (A429, B457, pp. 397–98). His footnote to the antithesis points out that a consequence of this purported relation of the world (what goes on in and what stands against the regress up to n in [11] to empty space (to S in [11]), is an indeterminacy of the position of the regress (the spatial motion) with respect to empty space. In (11), that is, the realist has not specified where, with respect to S, the step taking occurs. Surely it occurs at those places it reaches, but where are those places with respect to S? If the realist an8. Note that the realist can’t represent entities in space beyond the regress without lapsing into purely conceptual representation. But in (11) he is representing space beyond the regress, not a plurality of entities.

Transcendental Idealism in the First Antinomy

177

swers, those places right before places met with by further stages of step taking, his representation will collapse back into a version of (10).9 Thus the transcendental realist cannot represent the finitude of the world in space. Kant’s arguments in the First Antinomy regarding time are essentially parallel to his arguments regarding space. Once again, the issue is past time and the extent of states (of substance) in the full scope of past time. Since the realist represents states in past time in terms of the temporal regress, we need a representation of states not back to just k units ago (as in [6] above), but of states in relation to the full scope of being past (or beyond) stages of temporizing. A generalization of (6) which expresses this full scope is (12) Upon being affected by that object, it is proper to be as far as you please (up to whatever stage you please) in temporizing-cum-tracking-cum-perceiving those states of that object at initial times, where in (12) the potential infinite is represented not by incompletability of finishing, but by arbitrariness of how far along one is. But now there cannot be “as far as you please” many states. As in the space case, the potential infinite applies only to procedures, not to collections of extra-constructive individuals (states). Thus the realist has to replace the potentially infinite regress with an actually infinite regress, as in (13) Upon being affected by that object, it is proper to be beyond all stages of temporizing-cum-tracking-cum-perceiving those states of that object at those times. In (13) the realist, via the regress, is representing an infinite totality of states standing against the regress, but at the cost of representing finishing an infinite regress. But Kant says that “the infinity of a series [the regress] consists in the fact that it can never be completed through successive synthesis” (A426, B454, p. 297). Kant is denying, that is, that an infinite regress can be (represented as) now done with, as in (13). Indeed, if it is now done with (now proper to be beyond all stages) then what was it proper to be doing a moment ago? Producing a kth downbeat? That’s impossible since I would then be now up to k + 1. Similarly, there is nothing it was proper to be doing two moments ago, three moments ago, etc. In terms of accomplishing the regress, then, every 9. Surely objective space is what gets “covered” by taking further steps; otherwise S in (11) might signify space that doesn’t get covered, in which case (11) doesn’t even signify that further steps beyond n don’t meet objects, and so wouldn’t signify the finitude of the world met with by the spatial regress.

178

t h i n g s i n t h e m s e lv e s

moment of past time back from now would be the same, or as Kant puts it, “up to every given moment an eternity has elapsed” (A426, B454, p. 397).10 Thus the transcendental realist cannot represent an infinite extent of states (of substance) standing against the regress of perceptions in past time. As to the first antithesis regarding extent of states in past time, the realist has to represent finitely many such states within further (preceding) empty time. If he tries to represent the empty times via the regress, he gets (14) Upon being affected, it is proper to be further than k (but otherwise) as far as you please in temporizing-cum-failing-to-perceive states of the object at those times, which, once again, is impossible because there cannot be “as far as you please” many times. The realist, then, can only represent the emptiness of time (beyond k units ago) without relation to the regress, as in (15) Upon being affected, it is proper to be up to k in temporizing-cum-trackingcum perceiving those states of the object at those times and T is otherwise empty (where T signifies the whole of objective time). Kant’s comment regarding this attempt is that “no coming to be of a thing is possible in an empty time, because no part of such a time possesses, as compared with any other, a distinguishing condition of existence” (A427, B455, p. 397). His point, I believe, as in the corresponding space case (see [11] above), is just that (15) fails to specify where, with respect to T, the regress in temporizing takes place. Surely the regress takes place at those times of T that the regress corresponds to, but where are those times with respect to the global time T? If the realist says that they are after the times at which the regress is empty (at which the regress fails in perception of a state), (15) will collapse back into (14). Thus the realist cannot represent the finitude of the series of states in past time.11

4 . t r a n s c e n d e n ta l i d e a l i s m a s t h e s o l u t i o n Despite the complexity of the details, Kant’s arguments in the First Antinomy are quite simple at their core. The regress of space and time 10. I believe it is this absurdity of now being done with an infinite procedure that led Kant (contrary to his practice in the Aesthetic) to put the argument regarding time before the argument regarding space. In the thesis he later rejects completing the spatial regress by tying the idea to the time case. 11. Surely, whatever else, objective time is what gets “covered” by temporizing, otherwise T in (15) might signify time that doesn’t get covered, in which case (15) doesn’t even signify that preceding stages of the regress are empty, and so doesn’t express the

Transcendental Idealism in the First Antinomy

179

cannot be actually infinite, while the realist’s representation of entities in terms of the regress (as corresponding to it) requires an actually infinite regress, either to represent infinitely many objects and states (the thesis), or to represent infinitely many empty spaces and times (the antithesis). The realist’s initially plausible idea to use the regress to represent the world as what is all there waiting to be met with by the regress thus collapses into antinomy, since this met-with world must be a finite or an infinite totality, but it cannot be either. Kant’s arguments turn on his opponent’s having to relate entities to a synthesis or regress, but this is not circular 12 since the realist, on pain of otherwise lapsing into empty merely conceptual representation, accepts that objective entities, though outside the regress waiting to be met with by it, are to be represented precisely as what is thus reachable by the regress. Further, Kant’s arguments cannot be faulted for a supposed lack of understanding of infinity.13 He accepts the cogency of the idea of an actual (total) infinite. His charge against the realist is not that such an infinite is incoherent but rather that the completion of an infinite regress that the realist needs to represent a totality of entities is incoherent. Even if Kant’s arguments in the First Antinomy are cogent, they only prove that transcendental realism is antinomial, not that transcendental idealism is true, since it may be that transcendental idealism is antinomial also. However, the idealist, since he denies that there is anything standing against the regress, is not committed to the idea that there is either a finite or an infinite totality of entities standing against the regress. As Kant says, “Since the world does not exist in itself, independently of the regressive series of my representations, it exists in itself neither as an infinite whole nor as a finite whole. It exists only in the empirical regress of the series of appearances, and is not to be met with as something in itself ” (A505, B533, p. 448). Thus transcendental idealism avoids the antinomy, since it requires neither a finite nor an infinite totality. The idealist can represent the full extent of the world (that is, what exists only in the regress) simply via the potential infinity that is suitable for procedures. So, for example, the following represents the infinite (the nontotal or the nonactual infinite) extent of the world in space: (16) It is proper to keep taking steps and to keep being affected. finitude of the world. This just reinforces the point that the extra-regressive representation of T in (15) is not available to the realist. 12. The circularity objection is made by both Peter Strawson in The Bounds of Sense (London: Methuen, 1966) and Paul Guyer in Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987). 13. Strawson also makes this objection in The Bounds of Sense.

180

t h i n g s i n t h e m s e lv e s

In (16) the world is represented as ongoing (never-ending) affection in the course of (arising in) ongoing step taking. The finite extent of the world can be represented as: (17) It is proper to keep taking steps to n and to keep being affected, and to keep taking steps beyond n failing to be affected. In (17) the world is represented as a limitation on being affected (which, note, is not a representation of a finite totality of entities standing against the regress). Thus, not only does transcendental idealism avoid the antinomy, but it is adequate for representing the character of the full extent of the world.14 Completely similar remarks pertain to the case of the extent of the world in past time. Transcendental realism, as I have characterized it, is no straw man, but a highly natural response to transcendental idealism, which response can be turned into two supposed objections to Kant’s view. The first objection is that spatial construction is a species of motion, and motion is extrinsic or relative. Motion, that is, is motion through space, or is a change of position in space, and hence if there is proper spatial construction, there must also be extra-constructive space. Hence, as the transcendental realist holds, space cannot exist only in the regress, but must also be something that stands against the regress (something the regress moves through). In answering this objection, we may first dismiss the idea that because we cannot visually detect motion except against some stable background (say, of objects or parts of an object relatively at rest with respect to one another), and so can only detect extrinsic motion, therefore motion is extrinsic. This is to argue from features of visual phenomenology of motion to the nature of motion itself. One might as well argue that since we cannot visually detect motion without color contrast, that therefore motion inherently involves color. Besides, a blind person can represent motion by detecting the 14. Kant’s actual claim is that for the idealist the regress that represents the world in space proceeds “in indefinitum” (A521, B549, p. 457). What this means is that it is indefinite where (as I keep taking steps) I will be affected, and Kant represents this as the propriety of searching for affection at each stage of the regress (see A514, B542, p. 453). This is compatible with either limited or unlimited affection. The reason Kant puts it this way, I believe, is that both (16) and (17) are too specific in expressing limitless or limited affection, as they both express affection arising at each of a series of stages. Affection could be limitless, for example, by arising at every other stage, at random stages, etc. The expression of this would require embedding an existential quantifier within a universal as in the following: It is proper to keep taking steps so that upon being affected it is proper to keep taking further steps until you are further affected, which expresses that whenever there is affection there is further affection. Presuming that Kant had no way to express this indefiniteness as to which steps, the best he could do would be to license the propriety of an ongoing (step-by-step) search for affection.

Transcendental Idealism in the First Antinomy

181

force of a moving object on him without having to detect any stable background. A better argument is that motion is extrinsic because by its very nature motion is a change of relations (viz., a change in either spatial relations to other things or a change in relation to absolute space itself). However, there is an intrinsic dynamical conception of motion, according to which the motion of an object is characterized completely in terms of forces acting (locally) on it—and so without relation to other objects or to extrinsic space. This conception of motion is closer to the blind person’s detection. Within relativistic physics, for example, this is the conception of geodesic motion, which is the motive state of a mass that is not acted on by mechanical (electromagnetic, etc.) forces. This is an intrinsic conception because whether an object is thus moving geodesically is completely independent of its relations to other distal objects (reference frames) or to supposed substantival space. Indeed, the relativist “builds” space (space-time) out of local geodesic motion, so that space (space-time) is the total structure of possible geodesic motion. In this way geodesic motion is “fundamental” (not defined in terms of any antecedent extrinsic relations) and space is “defined” in terms of it. There is a parallel here to Kant’s transcendental idealism, which likewise has as its fundamental basis a system of canonical motions (constructions), and understands all spatial structure and attributions derivatively in terms of the full scope of such (proper) motions. Thus, for Kant, to take k steps is an intrinsic motion or construction that doesn’t require gearing how one moves in relation to the fixed stars or to absolute space. Further, to say that something is so far away is just to say that affection arises so far along in the motion or construction, so that spatial relations are defined in terms of motion, rather than vice versa. The idea, then, that transcendental idealism is incomplete or incoherent without the addition of extra-constructive (extra-motive) relational or absolute space rests on a bogus claim that motion has to be extrinsic.15 Similar remarks pertain to Kant’s account of time. In the Kantian view time is basically local and intrinsic. Indeed, a temporal construction, such as a series of downbeats, makes the subject a “local” clock in the relativistic sense, so that for Kant all time fundamentally is “proper” time. Just as for the relativist there is no extrinsic time against which, or in which, proper time is measured, so too for Kant time basically unfolds with tracking a substance (see [2] above), and so “along” a world line.16 15. Indeed, I claim that Kant’s “kinetic” or “motion” (construction) account of space is closer to (i.e., more suitable for representing) relativistic space than either classical relational or substantival accounts. See Essay 5 above. 16. At the end of the First Analogy Kant tries to argue for “building up” distant time coordination from the times (histories) that pertain to individual substances (see A189,

182

t h i n g s i n t h e m s e lv e s

The second altogether natural objection of the transcendental realist is that the world cannot be just what arises in the spatio-temporal regress, since what thus arises is affection or being affected, and there can be no affection without objects (there and then) to affect us. Thus Kant’s conception of affection in the course of spatio-temporal construction is inherently incomplete and incoherent, unless completed by objects standing against the regress to be affected by, which is just transcendental realism. Note that this objection is not that perceptions must be perceptions of objects, because the idealist already has a distinction between being affected and having a perception. As in (2) above, the real that corresponds to perception (viz., being affected) is permanent or endures, whereas perceptions are momentary. Equivalently, already for the idealist the real in the regress is subject to tracking (keeping track), while perceptions are responses in the course of tracking.17 All this is just to say that for the idealist there are substances (enduring trackable reality). Transcendental idealism then is not a phenomenalism, since besides possible (i.e., proper) perceptions there are also phenomenal substances. The objection of the transcendental realist is not against a supposed phenomenalism in transcendental idealism. Rather, the objection is that “being affected” (which is already distinct from having a perception in that it is subject to tracking) is a two-place relation requiring an object as well as the subject. This last way of formulating the objection is also the key to understanding Kant’s response, for the formulation is a formal one; viz., there has to be a second relatum or a second subject of the two place relation “.l.l. is affected by .l.l.” Kant allows the applicability of the subject-predicate form of judgment. However, these forms are merely syntactic (forms of the relations of judgments to one another in reasoning) unless schematized. The schema of the subject-predicate form of judgment is substance (i.e., the permanence of the real), which schema appears in (2) above. It follows, then that Kant holds that the subjectpredicate judgment (and so the notion of a second subject term for the relation) has applicability in regard to affection that arises in the B252, p. 217). He thought, of course, that the topology of time was absolutist (viz., distant simultaneity always holds everywhere throughout space, and does so relative to any substance). There is nothing in his theory, however, that forces this to be the case. 17. We can make this distinction out as follows. “Being affected” can be taken to signify grossly mechanically interacting, as when one lifts, pushes, pulls, etc. In these cases one’s whole body is involved in mechanically interacting. Thus I am affected if I attempt a gross mechanical interaction and there is resistance. The propriety of being in the course of tracking upon presently being affected is just then the endurance of mechanical existence. Perception, on the other hand, is a matter of further detecting (via sense organs) consequent upon mechanically interacting.

Transcendental Idealism in the First Antinomy

183

regress. I suggest that Kant can incorporate this form of judgment by allowing the mention of it within the regress (rather than its use to anticipate what stands against the regress). Thus, whereas the realist, as in (4) above, says that (18) It is proper to take k steps and be affected by the object there that is soand-so, the idealist says, rather, (19) It is proper to take k steps, to be affected, and to say “that object here affects me and is so-and-so.” Whereas the realist in (18) makes a reference to an object waiting to affect me, the idealist in (19) merely licenses a reference as one that is proper within the course of the regress. This allows Kant to have affection as a two-place relation, while still avoiding antinomy. Whereas (18) upon generalization leads to the problems of the First Antinomy, (19) upon generalization simply becomes (20) It is proper to keep taking steps, to keep being affected, and to keep saying “that object here affects me and is so-and-so.” In (20) there is the ongoing or incompletable regress of saying something, not an ongoing or incompletable totality of entities. Kant’s idealism, then, allows reference to affecting objects, but only demonstrative reference; viz., it licenses demonstratively referring in the course of the regress. What it precludes is the idea that we can now anticipate the regress by now referring to objects waiting to be met with or to affect us. The distinction from realism, then, is not over whether there are objects outside us that affect us, but whether these objects are within the regress or are part of a world standing against the regress. Not only, then, does transcendental idealism avoid the antinomy of transcendental realism, but it is an autonomous coherent view, not requiring transcendental realism (anticipating the regress) to be coherent. It requires neither extra-regressive space and time for the coherence of spatio-temporal procedures nor an extra-regressive world of objects for the coherence of being affected within these procedures. If I am correct, then the importance of the argument of the First Antinomy is not just as a proof of transcendental idealism but also as important textual evidence of what this idealism is. It is not phenomenalism since there are substances that endure, not just possible perceptions. It is not verificationism, since the representations of the past are not representations of verificatory procedures. It is basically a constructivism, holding that the world exists in procedures (or rules for procedures) of being affected and responding in the course of spatio-temporal con-

184

t h i n g s i n t h e m s e lv e s

structions. Each such procedure (or rule for a procedure) situates us (as we are here and now) with respect to interacting with reality, either as separated by a spatial construction we are at the beginning of, or separated by a temporal construction we are in the middle of. Such interaction with reality is not itself a matter of construction but a matter of what properly arises passively in construction. Kant’s repeated talk of incompletability (the potential infinite) is perhaps the most definitive evidence we have in his text that his idealism is a constructivism, since such a notion of the infinite only makes sense in a constructivist context.

10

Macroscopic Facts, Quantum Mechanics, and Metaphysical Realism

There is a plausible way of understanding quantum mechanics according to which reality has quantum-mechanical structure whether or not quantum phenomena are conceptualizable. Further, reality’s having that structure determines or produces macroscopic facts. If so, then such macroscopic facts arise from or are determined by a reality that is independent of conceptualization. Thus, although macroscopic facts are conceptualizable, they do not owe their reality to being conceptualizable. In this understanding of quantum mechanics, then, macroscopic facts are real apart from any intrinsic relation to conceptualization, and so are metaphysically real. This has negative implications for Kant’s Copernican revolution, according to which studying the conditions of conceptualizability can function as a method for doing metaphysics.

1 . m ac r o s c o p i c fac t s In this section we consider the nature of macroscopic facts themselves. I argue that they exist as part of the macroscopic scientific causal order, and so apart from any role they have in semantic theorizing (as truth conditions for statements). This still leaves it open that the scientific order of things itself exists only in relation to conceptualization. In subsequent sections I shall argue that this macroscopic order is due to the structure of reality apart from any relation to conceptualization. The contention that facts, if they are real at all, are real only in relation to semantic theorizing is held, for example, by Strawson, Davidson, and Putnam.1 Strawson and Davidson then reject facts at least partly on the basis that they have no such semantic function, while Putnam accepts facts, though their reality for him is still internal to conceptualiza1. Peter Strawson, “Truth,” in Truth, ed. George Pitcher (Englewood Cliffs, N.J.: Prentice-Hall, 1964); Donald Davidson, “True to the Facts,” in Inquiries into Truth and Interpretation (Oxford: Clarendon Press, 1984), 37–54; Hilary Putnam, Realism and Reason, vol. 3 of Philosophical Papers (Cambridge: Cambridge University Press, 1985).

185

186

t h i n g s i n t h e m s e lv e s

tion.2 A full response to Putnam is the burden of subsequent sections. In this section we argue against Strawson and Davidson that facts are not “shadowy” entities or hypostatised “dummy” entities that are nothing beyond truth of statements. We shall carry out this argument by considering Davidson’s rejection of the existence of facts. Davidson contends that if the only understanding of facts is in the sentential context “_____ corresponds to the fact that _____,” then by substitution of co-referring singular terms and logically equivalent sentences, it follows that there is only one fact. For our purposes we can accept Davidson’s contention. Note, however, that the exact same point applies to the sentential context “_____ in the event that _____.” Hence, if this context were our only understanding of events, it would follow that there is only one event. Davidson avoids this conclusion because he thinks we have an understanding of events apart from this context, as, rather, individual relata of singular causation. What I wish to suggest is that we have an understanding of facts as well outside the sentential context, as individual relata of singular causation. Hence we likewise avoid the conclusion that there is only one fact. Davidson3 explicitly denies that a fact such as there being sufficient oxygen in a room is part of the cause of a match lighting. It is the striking of the match (that event) that is the cause of its lighting. Davidson thinks the striking is the same event as the striking in a room with oxygen, so “sufficient oxygen in the room” functions only as part of a fuller description of the striking event, which is the cause, not as a further component of the cause itself. Consider, however, a case of each of two people pushing on a stalled car to get it moving. The first person’s pushing seems to be the same event as the first person’s pushing near a second person who is pushing. Hence the second person’s pushing functions only as a fuller description of the event of the first person’s pushing and, by parity with Davidson’s argument, we ought to conclude that the second person’s pushing is no part of the cause of the car moving. This, of course, is absurd because the second person’s pushing is clearly involved in the action that gets the car moving. But, similarly, the room’s containing oxygen is involved in the action that leads to the match’s lighting. It is involved not merely in law-like explanation but in 2. For Putnam, correspondence to reality has the semantical role of explaining valid reasoning. This gives reality a status in relation to first-order quantificational structure, which is pretty much the same conception of semantic structure found in Davidson and Strawson. In this understanding semantic structure requires a notion of individual entities, and it is just such a notion, we shall see, that is problematic in relation to core quantum mechanics. 3. Donald Davidson, “Causal Relations,” in Essays on Actions and Events (Oxford: Clarendon Press, 1980).

Macroscopic Facts and Realism

187

the very individual process (transformation of energy) that takes place on that occasion to light the match. Davidson is correct against Mill that a singular cause is not a sufficient condition. After all, without forces maintaining a perched bucket of water so that it doesn’t fall on the match just as it is struck, the match wouldn’t light. Hence, the striking alone is not sufficient. What I contend, however, is that a singular cause is constituted by all the factors that contribute to the action whose upshot is the effect. Since the room’s being oxygenated is a factor contributing to the chemical processes that are the action, it is part of the cause. Other conditions, such as constraining forces on the bucket, pertain to what may inhibit, interfere with, or override the action that leads to the effect. They are not involved in producing or sustaining the action but in “letting it happen” and so are not part of the cause, though part of the sufficient condition. However vague this distinction between producing action versus letting it go on, it seems to be central to the idea of singular causation in classical macroscopic physics, where a cause yields an effect by being the productive source of an action that, played out without interference, issues in the effect. Thus a moving object’s being charged is a source of a transmission of energy to the effect of the motion of a second body. Intervening between cause and effect is an action (transference of energy or momentum) from the cause to the effect. This constitutes, to some extent, the tie or bond between cause and effect that Hume demanded for singular causation. On this understanding of causation, objects being in certain conditions or states (i.e., facts) are a component of singular causation. The striking of the match (an event) together with the room’s being oxygenated (a fact) produces a process of energy variation to the upshot of the match lighting. A causal claim, then, involves singular terms for facts and states a relationship into which facts enter. Without any confusion of causes with their descriptions, or any confusion of causes with causal laws or explanations,4 facts (objects being in certain conditions) enter into singular causal transactions.5 4. Causal laws are general statements of kinds of action production, no different in specificity from the instances that state particular causal transactions. 5. Of course events as well as facts enter into singular causal relationships, but events as conditions or changes in objects or bodies, not “absolute” events in Davidson’s sense. The classical scientific causal order of the world always “locates” events or conditions in bodies because motion and location pertain to bodies (masses), and spatial location is a component of all other (electromagnetic, thermal, etc.) causal transactions. Without bodies, that is, there would be no singular or lawful tracing of causal sources of action across time and space. To the extent, then, that causal action depends on spatial location,

188

t h i n g s i n t h e m s e lv e s

Facts, then, are real and exist apart from any explicit relation to semantic theorizing (to truth conditions of statements) as elements of the general scientific causal order. These elements of the causal order also have “intelligible” structure, or the usual structure of facts invoked in correspondence theories of truth. Thus, on a correspondence theory, the statement that the room has oxygen is true in virtue of the room’s being oxygenated (the very fact involved in the match lighting). Likewise the statement “the apple is red” is true in virtue of the apple’s being red, where the apple’s being red is the inherence of a property or a feature in an individual thing, or an individual thing’s being in a certain condition or state. But this is the very fact that is involved in certain changes in light-frequency detectors. The facts, then, that are related to the truth of thoughts or statements are also, apart from that relation, elements of the general causal order of things. This is the result we need in order to establish the metaphysical reality of facts in the following sections. The semantic role of facts could be integrated with their role as elements of the general causal order if that semantic role itself were just another case of what goes on in the causal order. I sketch briefly how this unification might be effected.6 Suppose we have a hypothetical heat-seeking missile that has a mechanism for reaching and then destroying a hot object. In particular, its mechanism is not for targeting heat in the air but heat in an object. Thus we may suppose that it is drawn toward heat but that, upon reaching it, it unlocks extensors that probe for cohesive mechanical interaction. Only if the extensors, upon pushing out, meet mechanical resistance does the missile go on to fire. The missile, then, has in it a mechanism for destroying a hot object. Barring a breakdown, and barring interference (such as the missile being shot down, blocked, etc.), this mechanism will fully unfold (to explosion) given the existence of a hot object. Suppose then that there is a hot object in its range. That object’s being hot, together with the mechanism in place in the missile, is the source of action for that same mechanism to unfold fully in output. The object’s being hot is a causal condition of the mechanism in the missile being realized. In sum, there is an initial condition or state “inside” the missile that may or may not be realized and that is realized only with the fact of an object’s being hot. Similarly, a thought is supposed to be a condition “inside” an orand neither “absolute” events nor “absolute” persisting states position themselves apart from bodies, it is events such as an object’s increasing its charge, and not Davidsonian events, such as the charge’s increasing, that figure into the scientific causal order. 6. For a fuller development of the following sketch, see Essay 4 above or my Representation of the World: A Naturalized Semantics (New York: Peter Lang, 1996).

Macroscopic Facts and Realism

189

ganism that may or may not be true, and is true only with the existence of a fact. The parallel structure of these two cases suggests identifying thoughts in some way with mechanisms for (environment-interacting) behavior, and then equating their truth with their realizability according to the existence of facts outside the organism. Suppose a cat has been stalking a mouse in the bushes for several nights, and on this night it detects a rustling in the bushes. It may think to run to the bush to eat or kill the mouse. This thought, roughly, is its being set or prepared to do just that. Further, this state of the cat may or may not be realizable in that behavior’s actually unfolding. Barring interference, the condition of that thought’s being realizable is the object in the bush’s being a mouse. On the identification of thoughts with mechanisms for behavior, then, facts are truth conditions of thoughts by being causal conditions of the realizability of mechanisms. But then facts as truth conditions are thereby a special case of facts in general as elements of singular causal transactions, where the “special” case is that the causal transaction is to the unfolding of mechanisms in place in organisms. Our causal account, rather than being inconsistent with the “intensionality” of facts as they relate to thoughts, would to some extent explain it. Suppose I have a thought that there is a red apple in the refrigerator, and suppose, roughly, this thought is a mechanism in place in me for going there, reaching out to grasp, and upon detecting red picking up what is grasped. If when I reach out to grasp (mechanically interact), there is no resistance, the mechanism doesn’t continue to unfold. Thus there being an object there (a cohesive mass that is a source of mechanical resistance) is a condition of the mechanism’s fully unfolding. Suppose now that next to the apple is an electromagnetic detector that indicates that some set of frequencies of light waves are reflected from the apple, and that these would be sufficient on the occasion for the apple’s being detected to be red. Suppose further that instead of looking at the apple, I, who know no physics, look at the detector that is indicating the range of frequencies. That detection will not act to make the mechanism in me continue to unfold (to make me pick up the apple) since, recall, it is only on discerning the red of the apple that I pick it up. Now the apple’s reflecting that set of frequencies on that occasion is the scientific formulation of the fact of the apple’s being red, but the fact, so formulated, is not the condition of the realizability of my thought. Equivalently, it is not how the condition of the apple influences other things (not how it enters into casual transactions producing electromagnetic-detector changes), but how that condition influences my

190

t h i n g s i n t h e m s e lv e s

sense organs that is involved in the thought’s realizability. That very fact (condition of the apple) enters into all sorts of causal transactions with other phenomena, but only some of those causal transactions (say, ones pertaining to certain actions on my sense organs) are involved in the realizability of my thought. Roughly, then, the apple’s being red is a “partial” characterization of the fact (the object’s being in that electromagnetically potent condition), in the sense that it characterizes the fact in terms of only some of its potential for entering into causal transactions. The unification of the semantic role of facts (as truth conditions) with their existence as elements of the general objective scientific causal order is simply that it is the very same facts in both cases, only the semantic role involves a partial (subject-perspectival) characterization of these facts.7

2 . m e ta p h y s i c a l r e a l i s m a n d q ua n t u m m e c h a n i c s Even if one accepts that facts belong to the scientific causal order, this doesn’t imply they are metaphysically real. If science itself involves representation of reality, then, apparently, the reality science represents cannot go beyond the limits of representation or conceptualization. Scientific reality, and hence facts as part of that reality, are thus internal to representation. The supposedly necessary claim that science deals with no reality beyond what is scientifically conceptualizable seems to contradict the claim that quantum mechanics deals with a reality that it is unable to conceptualize. I wish to argue that the latter claim can be given a plausible defense, thus calling into question the contention that science cannot go beyond conceptualizable reality, and with it Kant’s Copernican revolution by which a theory of cognition can set limits to all reality (including scientific reality) that we can ever be concerned with. Conceptualizability of what goes on, as I understand it, does not require the ability to picture what goes on; nor does it require that what goes on obeys specific laws or paradigms of classical physics. What it does require is simply a notion of what goes on as something happening to an individual entity or to a plurality of entities. It requires, that is, some notion of individuation or what counts as an individual entity, which allows what goes on to be conceived as what happens in one or 7. This would also explain the conceptualization of the macroscopic scientific order in terms of mechanisms for the “complete” detection of the full causal potential of facts. The scientist who goes to the refrigerator will pick up the red apple simply upon detecting the indication of its condition by the electromagnetic detector.

Macroscopic Facts and Realism

191

more such individual entities. Now, we are specifically interested in the conceptualization of what goes on quantum mechanically apart from interaction with macroscopic phenomena. Even if we have a conception of an interacting electron, or a conception of the detection of an electron, this does not imply that we have a conception of what goes on, while undetected, as happening to an individual electron. For example, suppose we prepare to send an electron through a two-slit apparatus, to then detect an electron on a screen past the slits. We have a conception of preparing an electron at the beginning of the experiment, and of detecting an electron at the end. This, however, gives us no conception of what happens in the interim as happening to an electron. We know that the mathematical structure of what goes on in the interim is a superposition of different states, but this by itself gives us no conception that in the interim it is an electron that is in the superposed state. Indeed, it is simply unclear what it is for a particle to be in a superposed state. Since being in a superposed state seems to imply that what goes on is not localizable, is it, then, one particle or several that is thus “scattered”? Is it one particle that is partially at different places or several particles fully at different places? There are problems, as well, with thinking that in the interim what goes on is an individual three-dimensional wave propagating. The mathematical structure does not determine what individual or individuals have or bear that structure, even to the extent of whether there are one or many. Thus, in saying that what goes on in the interim (between the initial preparation and the final detection) has a superposed structure, we are not yet per se conceptualizing what individuals in what relationships have that structure. Hence, if conceptualizing reality requires having an idea of what individuals or entities there are, it is not clear that we can conceptualize quantum mechanical reality in the interim (outside of its detectable interactions with macroscopic reality). By a “realist interpretation” of quantum mechanics, I mean an understanding of how quantum phenomena in themselves (apart from measurements) are to be conceptualized, which in turn involves an understanding of what quantum entities there are in themselves. A realist interpretation would then give us a semantics according to which statements or thoughts specifically pertain to identifiable individuals. The interpretation may or may not involve hidden variables; it may or may not involve particles, etc. The only requirement is that it involve individuals with identity conditions even when they are not interacting with macroscopic reality. Now, I do not mean to suggest by the remarks in the preceding paragraph that there cannot be, or even that there may not already be, realist interpretations of quantum mechanics. My point

192

t h i n g s i n t h e m s e lv e s

is only that it is not definitively clear that there is a realist interpretation and this, it seems, would be enough to imply that quantum reality, the fundamental scientific reality, may be beyond our conceptualization. This, in turn, would contradict internal realism, or the thesis that there cannot possibly be reality apart from the limits of conceptualizability. Internal realism would not be threatened by the possible failure of a realist interpretation if this failure meant there was no quantum mechanical reality at all. The very phrase “realist interpretation” seems to suggest this. However, the issue of there being a realist interpretation is an issue over conceptualizing quantum reality, not per se the issue of whether there is quantum reality. To hold that the failure of a realist interpretation implies that there is no quantum reality is to hold that the failure of conceptualizability implies that there is no reality, which is just the claim of internal realism. What I wish to suggest is that the transition sanctioned by internal realism is mistaken in relation to quantum mechanics in that it both neglects a central feature of our idea of reality and closes off real issues in understanding the relation of quantum phenomena to macroscopic phenomena. A central feature of our idea of reality, and one that is neutral with respect to the dispute between internal realism and metaphysical realism, is that what is detectable is real and what determines (yields, produces) what is detectable is real (whether the latter is detectable or not). Thus, in an elementary case, since detected rustlings in the bushes along with detected footsteps in the snow are real, so is what determines there to be such detectable reality; namely, some animal, and this is so whether the animal is detected or not. This feature of our idea of reality has two components, then. If a phenomenon is detected, then it is real, and what makes it to be the case that there is that phenomenon is likewise real. The first component allows that all reality falls within the limits of conceptualization. In particular it allows that detection is always already conceptualized detection; not a detection of an “absolute given” that conceptualization subsequently latches onto. Since I am not concerned to argue for metaphysical realism by holding that there is detectable reality apart from, or prior to, all conceptualization, I shall simply assume that there is no absolute given. The second component of our idea of reality allows that what determines detectable reality is conceptualizable, but it doesn’t straight off entail it. If the very notion of determination entailed that only what is conceptualizable can determine the detectably real, then this component of our idea of reality would fit with internal realism. What I wish to suggest, however, is that the situation in quantum mechanics illustrates the independence of the issue of determination from the issue of conceptualization.

Macroscopic Facts and Realism

193

There is a plausible way of understanding quantum mechanics, that is, according to which the issue of quantum reality’s determining the detectably real or not is split off from the issue of the conceptualizability of quantum phenomena themselves. This split gives sense to the idea that there may indeed be quantum reality (determining the detectably real) even though such reality is not conceptualizable, thus refuting internal realism. There is a core structure to quantum mechanics independent of the conceptualizability of quantum phenomena. This core includes states in Hilbert space, the superposition of such states, the Schroedinger evolution of such states, and operators on states with eigenvectors and eigenvalues. This core structure does not seem to determine detectable reality, since the application of it to quantum states interacting with macroscopic reality seems to lead only to superpositions of macroscopic states, such as Schroedinger’s part-dead part-alive cat. Suppose, however, it pans out that when the core structure is suitably applied, what we actually get are decoherence effects 8 according to which macroscopic interferences or superpositions are (almost immediately and to an overwhelming approximation) destroyed. That the macroscopic object then is in some definite state or other is determined by the core structure of quantum mechanics according to decoherence effects. Further, these effects are not mathematical artifices but essentially sorts of processes due to macroscopic perturbations.9 The collapse of superposition in the detectable macroscopic instrument, then, is determined by the quantum mechanical structure of the interaction of the atomic with the macroscopic. There are three aspects to the decoherence account that are important for our purposes. First, detectable macroscopic definiteness is determined quantum mechanically, not classically. Second, the determination is not a sheer mathematical transformation but a process taking time, albeit an undetectably small amount of time. This gives us, thus, a robust sense of the quantum mechanical determination of the macroscopic as an occurrent process. Third, the determination is by the application of the core structure of quantum mechanics to macroscopic objects with large numbers of degrees of freedom, without any semantic interpretation of that structure that would make it conceptualizable as to what quantum phenomena there are that bear that structure. In accord, now, with the principle that what determines or produces the de8. For a discussion of decoherence, see Roland Omnes, The Interpretation of Quantum Mechanics (Princeton: Princeton University Press, 1994), chap. 7. 9. Ibid.

194

t h i n g s i n t h e m s e lv e s

tectably real is itself real, the first and second aspects imply that core quantum mechanical structure is real. Reality, that is, is structured quantum mechanically. The third component implies that this is so independent of whether this quantum structure of reality is semantically conceptualizable, i.e., independent of any interpretation of what the individuals are that have this structure. Hence, all together, the decoherence account establishes metaphysical realism in regard to quantum mechanics i.e., the possibility that reality is quantum mechanical even if such reality is not conceptualizable.10 It may be that decoherence effects leave an infinitesimal probability of macroscopic superposition that never completely disappears. The situation then would be that the intrinsic structure of quantum mechanics probabilistically (with high probability) determines that there be definite macroscopic reality. Now, it is plausible to hold that what “determines” the detectably real with high probability is itself real. Equivalently it is plausible to hold that what makes something that is detected highly probable (by processes that probabilistically yield the detected reality) is itself something real. If so, then even if decoherence effects never completely erase superposition, they still establish the core structure as the structure of reality on specific occasions on which macroscopic definiteness (as in measurement) is yielded and detected. Determination by decoherence effects manages at once to be something less than a realist interpretation, which makes quantum phenomena conceptualizable, but more than an instrumentalism, which makes core quantum structure merely a mathematical instrument for prediction. As opposed to a realist interpretation, determination by decoherence does not involve any specific interpretation of the core structure as a structure of indeterminate particles with potencies, or as a structure of waves, or as a structure that is an effect of “hidden” determinate particles in a quantum potential.11 The difference of determination by de10. The determination of the macroscopic by quantum structure is not redundant because it has as a consequence the usual probabilistic experimental results that verify quantum mechanics. The postulation of real quantum structure in the world leads not just to macroscopic definiteness but to all the probabilistic predictions of quantum mechanics with respect to macroscopically detectable reality (viz., the results of measurements). 11. Note that a realist interpretation, whether including hidden variables or not, may likewise determine macroscopic definiteness, thus bypassing the “collapse” of the wave function. The key point, however, is that the issue of bypassing wave function collapse is theoretically distinct from the issue of conceptualizing quantum phenomena, since determination by decoherence effects yields macroscopic definiteness from the core structure without conceptualizing that core structure. It is this that allows the distinction between reality (that which determines the detectably real) being quantum mechanical versus its being conceptualizable.

Macroscopic Facts and Realism

195

coherence from instrumentalism can be seen by contrasting decoherence with the Bohr interpretation, which holds that there is no intrinsic quantum reality. The core structure involves states evolving over time according to Schroedinger’s equation when there is no interaction with the macroscopic. We can call this the intrinsic aspect of the core structure. This aspect is crucial for the prediction of the probabilities of macroscopic detection and so it does at least have an instrumentalist function. For Bohr, however, this intrinsic structure does not determine macroscopic reality, since its application is actually inconsistent with any definite macroscopic detection.12 This intrinsic structure, therefore, has merely an instrumentalist function, since it is not a structure of what determines detectable reality (quantum-to-macroscopic interactions),13 but only a structure that predicts detectable reality. Whereas realist interpretations are something more than sheer determination of the detectably real by the core structure, instrumentalist interpretations are something less than determination of the detectable real by the core structure. Determination of the macroscopic by decoherence effects is a program, not a fully worked-out account. Nevertheless, it presents a picture of how science, as a quest for discovering what determines the detectably real, might lead us to a reality that is beyond conceptualization. It may be that in addition to the decoherence program panning out, an interpretation can be given (either in terms of hidden variables or not) to make the determining quantum reality amenable to our conceptualization. This would not refute metaphysical realism, which doesn’t hold that reality is or must be beyond conceptualization but only that it might be. The separation of the issue of determination of the detectably real from the issue of an interpretation making the real conceptualizable is sufficient for establishing metaphysical realism. We can now characterize that realism as the thesis that the structure of reality that determines the detectably real may or may not go beyond our ability to cognize what entities or individuals have that structure. 12. This inconsistency ensures that the core structure is a mere calculating device as far as prediction goes. 13. Bohr’s reasoning as to why there cannot be anything more than this instrumentalist function to intrinsic quantum structure is, I believe, approximately as follows. For Bohr only classical reality is conceptualizable, and there is reality at all only where there is conceptualizability (viz., Bohr doesn’t allow for metaphysical realism). It follows, then, that there is no intrinsic quantum reality. Quantum phenomena in interaction with classical macroscopic objects are conceptualizable (via complementarity) along classical lines in those interactions as being particles or waves. Because they are thus conceptualizable, Bohr is a realist regarding quantum phenomena in interaction with the macroscopic. Some sort of Kantian Copernican idealism (i.e., internal realism) drives Bohr’s views. A

196

t h i n g s i n t h e m s e lv e s

3 . m e ta p h y s i c a l r e a l i s m a n d s c i e n t i f i c p r ac t i c e Metaphysical realism is an empirical or scientific possibility, in the sense that the best scientific theory may impute a structure determining (yielding by action) the detectably real, without incorporating a way of conceptualizing what individuals bear that structure. It is simply this property of a theory that makes it a theory of unconceptualized reality. This is true even if a theory’s having this property is not a part of the empirical adequacy of theories, and not a part of the acceptability of theories according to scientific practice. For the pragmatist, realism is only scientific to the extent that it is a component of, or an explanation of, scientific practice or scientific success. Those pragmatists who hold that it is not such a component deny therefore that science can establish realism (whether beyond conceptualization or not). Suppose, then, that empirical adequacy is the basis of the acceptability of theories, and that this adequacy requires only the predictability of detectable phenomena. As we have seen, such predictability by itself is not the same as the determination of the detectable phenomena that are predicted. Thus, in Bohr’s interpretation intrinsic quantum structure predicts the detectable without determining it, and so, for Bohr, it is empirically adequate but not realist. For a pragmatist who equates empirical adequacy with acceptability, there is simply no ground or reason for accepting a realist theory (or a realist understanding of a theory) over a nonrealist theory. However, pragmatic issues of acceptability seem to me to be quite beside the point. Even if a theory T that predicts is perfectly acceptable, if T has the property or character of also determining, and not merely predicting, the detectable, then it is a theory of reality, whether or not that exact fact matters for scientific practice. Similarly, whether reality “explains” scientific success or not, if a scientific theory has the character of determining the detectably real, then it is a theory of reality. The question is over the character of the theory, not whether theories having that character can explain scientific success. Once core quantum structure is instrumentally necessary for predictive adequacy, whether it has the further character of determining the phenomena that it predicts is an issue regarding that theory, and not a pragmatic issue regarding acceptability of theories, success of scientific practice, etc. Metaphysical realism, then, is not dependent upon (an account of realism as involved in) scientific practice. sufficient reason for instrumentalism, apart from this Kantian idealism, would be the supposed inconsistency of wave packet collapse with core structure.

Macroscopic Facts and Realism

197

Further, the supposed support for antirealism from the underdetermination of theories by the data, to the extent that this underdetermination pertains to the conceptualization of entities, is irrelevant to the metaphysical realist who requires only a structure to reality, not a conceptualization of entities. In this regard, the under-determination of theories is more a problem for the conceptual realist than for the metaphysical realist. Akin to a nonrealist account of scientific practice is a nonrealist account of scientific talk. The antirealist pragmatist can understand that talk along the lines of an assertabilist semantics. Consider, for a moment, an instrumentalist version of pragmatism. Supposedly, there will be allowable ways of using the mathematical apparatus of the theory as a calculating device for predicting the detectably real in various circumstances. These ways of calculating and tying calculations to detectable circumstances constitute the assertability conditions of making scientific statements. In quantum mechanics, for example, techniques warranted by the theory for calculating Hamiltonians of systems, and techniques for preparing system measurements, will be part of a body of standards that govern scientific assertions. For the instrumentalist such assertability conditions constitute all there is to making and understanding meaningful scientific assertions. In particular there is no realist semantics that recursively coordinates elements of statements to individual entities, or coordinates the truth of statements to properties of, or relations between, such entities. Thus in Bohr’s understanding of quantum mechanics any talk concerning intrinsic quantum structure, such as Schroedinger’s evolution of the wave function on its own, is to be understood as making calculative steps on the way to making further statements about detectable reality (which, for Bohr, includes quantum-macroscopic interaction). In particular there is no realist assignment of individual entities in reality to the components of such intrinsic statements. Roughly, the significance of an assertion is just its role in the practice of calculating and making predictions. For our purposes, the significance of an assertability semantics is that it frees the meaningfulness of scientific discourse from any realist semantics. For Kant, for example, the scientist must conform to the general conditions or constraints on realist conceptualization, because otherwise there would be no meaningfulness or cognitive significance to scientific discourse (or thought). For this reason Kant can say that science cannot go beyond phenomena (i.e., conceptualizable reality) without going beyond all sense whatsoever. This is the source of Kant’s rejection of metaphysical realism, at least for theoretical reason. The possibility of an assertability semantics shows that the meaningfulness

198

t h i n g s i n t h e m s e lv e s

of science is not limited by any general conditions for conceptualizing reality, since an assertabilist account precisely divorces making sense from conceptualizing entities. The use of assertability conditions by the instrumentalist goes hand in hand with his antirealism. However, to think that they necessarily go hand in hand is exactly to think that one can only be a realist where there is conceptualization; viz., it is exactly to be an internal realist. Indeed, it is Bohr’s Kantian internal realism that leads him to instrumentalism regarding intrinsic quantum structure. Bohr believed that intrinsic structure went beyond the general conditions of conceptualization (which conditions, for Bohr as for Kant, were perhaps tied too closely to the outlook of classical physics), and so such structure signified nothing real. More generally, to think that the issue of reality goes hand in hand with the issue of what sort of semantics pertains to statements is just to deny metaphysical realism. Suppose that we include, as part of the assertability conditions, norms or standards not just for calculation and prediction of the detectably real, but for determination of the detectably real as well. This would include, for example, derivations from core structure of particular structural configurations that remove superposition of states in some process-like way (i.e., decoherence effects). In this way the meaningfulness of statements of decoherence is severed from a conceptualization of reality. Unlike instrumentalist assertabilism, however, this assertabilist semantics is compatible with, and indeed implies, realism (although in the guise of unconceptualized reality). Indeed, to some extent the collapse of the wave packet in the Bohr-like understanding of quantum mechanics was a problem for assertabilism, because the Bohr-like understanding had no standards for where the “cut” is between the quantum mechanical and the classical, even if this is understood merely as having standards for when we can or cannot make assertions of definiteness. The discovery of large-scale “macroscopic” quantum effects exacerbates this problem. If, on the other hand, the program of characterizing decoherence effects systematizes as well when such effects do not take place macroscopically, then decoherence has a significant role even on an assertabilist understanding of scientific statements. So long as determination of the detectably real has thus an assertabilist role, metaphysical realism is part of the assertability conditions of scientific talk. In other words, scientific statements can “indicate” reality without those statements’ having a realist semantics. To the extent that they lack such a realist semantics, they thereby indicate unconceptualizable reality. In this way scientific sense is liberated from the Kantlike constraint of having to be a conceptualization of the real.

Macroscopic Facts and Realism

199

In sum, metaphysical realism is not refuted either by (1) an antirealist conception of empirical adequacy as sufficient for theory acceptance, or by (2) an antirealist conception of the meaningfulness of scientific talk. As to (1), whatever the canons of theory acceptance, the theory either has or lacks the character of containing a determination of the detectably real, and does so independent of whether it conceptualizes what the determining reality is. As to (2), determination of the detectably real can figure into the sense of scientific discourse as an assertability condition, without there having to be any realist semantics for that discourse.

4 . m ac r o s c o p i c fac t s a g a i n Suppose now that core quantum structure determines or produces not only macroscopic definiteness but essentially the entire classical nature of the macroscopic, viz., a world of objects in conditions that together act to yield effects in accord with the laws of classical physics. Suppose, that is, that core quantum structure determines the world of facts entering into causal processes described in our discussion in section 1. The program of determination by decoherence effects includes this entire determination.14 Decoherence effects, that is, are meant to be able to determine the macroscopic so that, to such a degree of approximation and within such time limits, the detectable duplicates classical structure. This derivation, as usual, is independent of any and all specific interpretations of the core structure. Because of this we can say that the program is to determine macroscopic facts from quantum reality, independent of the conceptualization of that reality. Since quantum reality itself is not determined by the nature of conceptualization, it follows that the reality of the facts that it determines is likewise not determined by the nature of our conceptualization. Let us suppose that we do conceptualize facts, or that facts have intelligible (conceptualizable) structure. Even so, it is not the nature of our conceptualization that determines them thus to have that structure. Rather, it is the nature of quantum reality that determines that structure, and that determination is independent of the nature of conceptualization. We have, then, a version of an Aristotelian metaphysics, as opposed to a Kantian metaphysics of the macroscopic, according to which conceptualization is of that which has an intelligible structure independ14. See Omnes, The Interpretation of Quantum Mechanics, chap. 6. The core quantum structure doesn’t determine which deterministic history will occur, but it does determine that some deterministic history or other will occur.

200

t h i n g s i n t h e m s e lv e s

ent of that conceptualization. Determination by core quantum structure not only implies metaphysical realism regarding quantum reality, but metaphysical realism as well regarding intelligible macroscopic reality. We argued in section 1 that the existence of facts is independent of semantic issues, since it is settled already by the causal structure of scientific reality. We were left with the possibility, however, that scientific reality itself might not be independent of semantic issues. We have seen in sections 2 and 3 that scientific reality in its fundamental quantum nature is not limited by issues of semantic conceptualization, and now in this section that likewise the scientific reality of macroscopic facts is independent of all semantic issues. The consequence of this is that it is not up to semantic inquiry to decide whether facts are real or not. By the time we do our semantic theorizing over the nature of cognition or of language, the issue of the existence of facts in the world is already settled. Of course it still may turn out that correspondence to the facts is not a component of understanding cognition or of semantic theory. What our conclusion as to the metaphysical reality of facts does imply, however, is that a full robust correspondence-to-the-facts theory of truth cannot be ruled out simply by objections to the existence of facts. Consider, for example, the supposedly basic dilemma of a correspondence theory that either the facts lack conceptual structure, in which case our thoughts can’t correspond to them, or else they have conceptual structure, in which case our thoughts only correspond to what is “internal” to the nature of thought. The picture I have presented resolves this dilemma by giving facts intelligible structure without thereby making them internal to thought. For another example, consider the supposed objection that we cannot get outside our thoughts to a reality they are supposed to correspond to. Again, in the picture I have presented, quantum theorizing does get outside our (individuative referential) thoughts to a structure of reality that determines the macroscopic facts that thoughts correspond to. Our picture, then, has two important consequences for the nature of semantic theorizing. First, it takes metaphysics (issues of what is real) outside the umbrella of semantics, in opposition to Kant’s Copernican revolution in all its guises. Second, it provides a world of facts, so that the issue of whether truth is correspondence to the facts is no longer to be decided by any rejection of the existence of facts.

s u m m a ry Macroscopic facts exist as part of the macroscopic causal order of things discerned by science, not by semantic theorizing. Further, sci-

Macroscopic Facts and Realism

201

ence itself is not limited by conditions of conceptualization. According to the principle that what determines or produces the detectably real is itself real, so long as scientific theorizing includes a structure that determines the detectable real, and does so independently of the conceptualizability of what individuals bear that structure, science can go beyond conceptualized reality. Quantum mechanics may present an illustration of this in the determination of the detectably real from core quantum structure by decoherence effects, which determination, in and of itself, involves no conceptualization of what bears the core structure. Such determination holds apart from pragmatic issues of theory acceptability or scientific practice generally. Further, such determination without conceptualization is compatible with the meaningfulness of scientific discourse via an assertabilist semantics that incorporates or gives a role to that determination. The macroscopic causal order of things, and with it macroscopic facts (as thus determined by reality independent of conceptualization), exists and has its intelligible structure independent of all conceptualization.

PART V FREEDOM AND MORALITY

11

Reason, Freedom, and Determinism in the Third Antinomy

In Kant’s view, genuine practical reasoning cannot be causally determined. Practical reasoning is open-ended in the sense that there is never a fixed stock of reasons that are definitive or conclusive as to how to live, and so as to what to do. For such open-ended reasoning to be efficacious, its concrete realization cannot be fixed or determined. Rather, only free, undetermined choice can close off practical reason without violating its open-ended nature. The dependence of practical reasoning on free choice leads Kant to the view that practical reasoning cannot be identical to any natural (and hence, for him, determined) occurrence. This leaves open the possibility of genuine reasoning outside of time or the natural order, that is, where such reasoning is efficacious by our free choices being part of God’s determination of that order. Although this leaves freedom outside of time, taking responsibility for one’s actions remains in time. I argue that Kant’s resolution of the apparent conflict between undetermined free reasoning and natural determinism is successful, but that the indeterminism of quantum mechanics can model the genuine open-endedness of practical reasoning in a theoretically more satisfying way.

1 . p r e l i m i n a ry d i s c u s s i o n o f undetermined choice The issue of freedom in the Third Antinomy is the issue of causally undetermined choice, not the issue of what Kant calls autonomy. As such it pertains to practical deliberation generally, not to moral legislation specifically. Kant says, “Whether what is willed be an object of mere sensibility (the pleasant) or of pure reason (the good) reason will not give way to any ground which is empirically given” (A548, B576, p. 473).1 Thus even apart from specifically moral considerations, when 1. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (Boston: Bedford, 1965). All references are to this edition.

205

206

freedom and morality

the object of the will is the pleasant, there is a freedom (spontaneity) of practical reason, and it is this aspect of freedom that is at issue in the Third Antinomy.2 Irwin calls this aspect of freedom “metaphysical freedom” and distinguishes it from autonomy.3 Allison and Beck identify it as Kant’s conception of Willkur, which they distinguish from the autonomy of Wille.4 Kant believes that the decision to act based on practical reasoning involves free or undetermined choice. This isn’t the whole of freedom, however, since limitation of one’s options when one chooses can limit freedom. Thus, even if I have a “free” choice based on deliberation as to whether to give someone who is holding a gun to my head my money or my life, the limitation on my options restricts my freedom. The options, we may say, are “heteronomous” despite whatever uncaused choice I still may have. I believe that, for Kant, options that conflict with morality are heteronomous or limiting to a rational being. My own various desires or goals, if they conflict with morality, are like a gun to my head, since I am not, by Kant’s lights, the rational author of these options. Allowing options only in accord with morality is what gives me any rational authority over my options and therefore my autonomy (viz., not only free choice among options, but authority over the options). Because there is free choice even with heteronomous options, it is possible in Kant’s view to use one’s “metaphysical” freedom (Willkur) to choose what is morally wrong. In this essay I shall be concerned exclusively with the issue of free (causally undetermined) practical deliberation, whether heteronomous or autonomous. The first large question is why Kant holds that freedom requires causally undetermined choice. Given that Kant identifies freedom in the Third Antinomy with the causality of reason,5 or 2. This agrees with Allison, who holds that freedom in the Third Antinomy pertains to practical reason in general, not just morality. See Henry Allison, Kant’s Theory of Freedom (Cambridge: Cambridge University Press, 1990), 33. 3. See Terrence Irwin, “Morality and Personality: Kant and Green,” in Self and Nature in Kant’s Philosophy, ed. Allen W. Wood (Ithaca: Cornell University Press, 1984), 33. Frederick Rauscher distinguishes it as “the power of decision” vs. (autonomous) moral legislation. See Rauscher, “Kant’s Conflation of Pure Practical Reason and Will,” in Proceedings of the Eighth International Kant Congress, ed. Hoke Robinson (Milwaukee: Marquette University Press, 1995), 2:581. 4. Allison, Kant’s Theory of Freedom, 131; Lewis White Beck, Commentary on Kant’s Critique of Practical Reason (Chicago: University of Chicago Press, 1960), 177. 5. An act or decision is free, that is, if it is governable by reason, which may apply even to cases where I act against what I think is reasonable due to some contrary desire. So, for example, I may eat a piece of chocolate cake though all my reasons are against it. The act nevertheless is governable by reason (and so free) if further, or other, or even repeated reasoning would have determined me not to eat the cake. On the other hand, if no further reasoning is relevant, so that the desire is immune to any reasoning (as in the case of

Reason, Freedom, and Determinism

207

with practical deliberation being the source of our actions, the question is why Kant holds that practical deliberation itself requires causally undetermined choice. After all, compatibilists, in the matter of the nature of freedom, often rest their compatibilism precisely on the thesis that free acts are acts caused by our own practical deliberation, where this deliberation itself may well be causally determined. Before turning to this question, we consider whether we may bypass it altogether by deriving causally undetermined choice not from the nature of practical reason itself but from the imputation of responsibility. Kant says, “That our reason has causality, or that we at least represent it to ourselves as having causality, is evident from the imperatives which in all matters of conduct we impose as rules upon our active powers” (A547, B575, p. 472). We seem to be able to tell ourselves that we ought to do such and such and others tell us that we ought to have done differently. If these imperatives make sense only if we have free (causally undetermined) choice, then that would seem to be a sufficient basis for rejecting compatibilism. Note, however, that in this passage Kant is talking about the effectiveness of reason (that it has causality in producing action), not its source (not whether it, in turn, is determined or not). It seems true that accountability and responsibility are directly tied to causality by reason (or by practical deliberation). Thus to hold someone accountable is to hold them to accounting for their action, viz., to give an accounting of their reasons for acting. Similarly, to hold someone responsible is to hold them to responding to reasons against what they did, and thus to justifying the reasons for which they acted. If so, then if reason wasn’t the determinant of their actions, it would be otiose to hold them accountable or responsible. But this is independent of whether the reasonable deliberation that produced the action was itself determined or not. Since Kant ties accountability (imperatives) to the effectiveness of reason, not its source, it is plausible to conclude that he is not deriving free choice from accountability, but rather is holding that practical reasoning itself involves free undetermined choice, and that it is only the effectiveness of reasoning that is supported by accountability. In a later passage Kant says, “Our blame is based on a law of reason whereby we regard reason as a cause that irrespective of all the above-mentioned empirical conditions could have determined and ought to have determined the agent to act otherwise. .l.l. Reason, irrespective of all the empirical conditions of the act is completely free and the lie is due entirely to its default” (A555, B583, a compulsive hand washer), then the act is not only contrary to reason but not governable by reason at all and so not free.

208

freedom and morality

p. 477). This again seems to be a derivation of undetermined choice from blame. However, Kant says he is using blame to “illustrate” the “empirical employment” of the concept of free (undetermined) reason, which seems to suggest that he is not deriving the concept of free reason from it. Further, in characterizing reason in the preceding paragraph (A553–54, B581–82, p. 476), he distinguishes the undetermined nature of reason (that it is not “dynamically determined in the chain of causes”) from its effectiveness in producing action (in “originating a series of events”). Since Kant distinguishes the undetermined nature of reason from its effectiveness, and connects blame to its empirical employment (its effectiveness), it is plausible to hold that he believes that even apart from issues of blame, there is something about reason itself that requires it to be undetermined. Reason, to be reason at all, must be undetermined. If that is so, then of course Kant can go on to give as an example of the employment of the concept of reason the imputation of blame or responsibility couched in “incompatibilist” terms (how the agent acting from reason could have done otherwise). We are left, then, with our basic question of why practical reasoning, to be reasoning at all, must be free of determining natural causes. A line suggested by many passages in Kant equates these supposed natural causes with sensuous impulses or sensuous motives (A534, B562, p. 465). If these were the only possible natural determinants of our practical deliberations, then it would be plausible to hold that practical reason must be undetermined, since desires or impulses determining deliberations would amount to their distorting or skewing those deliberations. It is the role of reason to consider and assess desires, not to be determined by them. Determination by desire would turn reasoning into rationalization. I would not be practically deliberating (coming to a reasoned decision), but “inventing” reasons for what is already decided. However, there can be physiological or neurological determinants of practical reasoning that, unlike desire or sensuous impulses, don’t skew or distort reasoning, and don’t turn it into a rationalization for already motivationally determined action. Since it isn’t the role of practical deliberation to consider and assess neurological factors, determination of reason by these factors doesn’t thereby violate the nature of reasoning.6 6. Allison’s “incorporation thesis,” according to which desires must be made into appropriate reasons, shows that the adoption of a maxim “cannot itself be regarded as causal consequence of the desire.” However, this by itself cannot show that therefore the adoption “must be conceived as an act of spontaneity on the part of the agent,” since it leaves the possibility of neurophysiological causal determination. See Allison, Kant’s Theory of Freedom, 40.

Reason, Freedom, and Determinism

209

Fortunately, Kant’s argument for the undetermined nature of practical reasoning doesn’t turn on any equation of natural causes with desires. He says that when we consider actions in relation to practical reason, “we find a rule and order altogether different from the order of nature. It may be that all that has happened in the course of nature, and in accordance with its empirical grounds must inevitably have happened, ought not to have happened” (A550, B578, p. 474). One way of reading this remark is that it is the claim that the laws governing practical reason are distinct from any natural laws, whether laws of desire or neurophysiological laws. Indeed, an action that comes under natural laws (viz., has happened) may fail to come under, or may violate laws of, practical reason (viz., ought not to have happened). On this reading, the laws of practical reasoning cannot be reduced to natural laws. This is a weak reading, in the sense that it is perfectly compatible with it that practical reasoning is always determined by natural causes. Indeed, the disparity of law is compatible with each case of practical reasoning being identical with some neurophysiological process, and hence with being causally determined.7 This latter is true even if there are no universal laws of practical reason, as in the Davidsonian position of anomalous monism applied to practical reasoning. Despite the causal determination of any case or token of practical reasoning in such a view, this view has been attributed to Kant by Ralf Meerbote.8 In Meerbote’s view the freedom of reason is just its anomalous nature (that it is not governed by universal laws), which is compatible with any case of reasoning being identical to natural (and so causally determined) occurrences. The trouble is that Kant believes that the freedom of reason resides in our intelligible character, and this character grounds (and so is not identical to) any natural temporal processes that determine our will (the latter being cases of our empirical character). Thus at best Kant holds that it is empirical processes determining the will that are token identical to natural (and hence causally determined) processes, but it is not in these processes that the freedom of reason resides. Kant holds, then, that genuine practical reasoning cannot be token identical to any natural occurrences because it is not causally determined. Practi7. The laws of chess are disparate in this way from the laws governing the physical unfolding of the computer. It may be that the moves that have happened physically ought not to have happened (chess-wise) due to a failure to run the program properly or to a failed program. Still, each chess move by the computer is identical to a physical state. 8. See Ralf Meerbote, “Kant on the Non-determinate Character of Human Actions,” in Kant on Causality, Freedom, and Objectivity, ed. William L. Harper and Ralf Meerbote (Minneapolis: University of Minnesota Press, 1984). For a further elaboration of this Davidsonian interpretation of Kant, see Hud Hudson, Kant’s Compatibilism (Ithaca: Cornell University Press, 1994).

210

freedom and morality

cal reason for Kant, then, is not an anomalous description of determined natural occurrences (which would make it determined too), but something undetermined and hence distinct (token-wise) from any natural occurrence, since nothing undetermined can be identical to a determined natural occurrence.9 Terrence Irwin and Henry Allison attribute to Kant a somewhat modified version of monism, according to which practical reasoning is a nonempirical (noumenal) description of natural occurrences. According to Irwin, Kant holds that free choices are noumenal events that are also phenomenal events.10 Irwin draws the consequence, correctly I believe, that therefore these noumenal events must be determined. As he says, “Now if an event is determined, it is true of it under all true descriptions [including nonempirical ones] that it is determined.”11 Allison, on the other hand, holds that a phenomenal determined event need not be determined under a nonempirical description.12 This, however, makes no sense if determination indeed pertains to events rather than to how they are described. The nonempirical nature of the description is simply beside the point. If a certain naturally determined event falls under the nonempirical description of being God’s favorite event, then it follows that God’s favorite event is naturally determined. My contention is simply that Kant is no monist at all regarding practical reason. Free or undetermined reasonable deliberation is the ground of certain empirical occurrences, but not identical to them.

2. open-ended reason and undetermined choice So far we have interpreted Kant’s remark that “It may be that all that has happened in the course of nature .l.l. ought not to have happened” as the claim that the laws of reason are distinct from any natural laws. This interpretation, as we have seen, doesn’t get us beyond a monism according to which cases of practical reasoning are identical to naturally determined occurrences. There is another way of interpreting Kant’s remark, however, that does, I believe, lead to the conclusion that practical reasoning is undetermined and hence cannot be identical to any 9. Kant, then, does not hold the “monism” component of anomalous monism. Nor is it clear that he holds the “anomalism” component. Although Kant did hold that there could be no science of psychology, by this he probably meant there could be no mathematically constructible system of psychological laws, not that there couldn’t be (low-level, but genuine) laws. 10. Irwin, “Morality and Personality,” 38. 11. Ibid. 12. Allison, Kant’s Theory of Freedom, 51–52.

Reason, Freedom, and Determinism

211

natural (and so determined) occurrence. Suppose that “all that has happened” leading to a particular action of mine includes considerations or reasons I went through in deciding to act that way. To say that nevertheless it “ought not to have happened” may mean that those considerations (unto a decision) are faulty for leaving out other pertinent considerations. To say, as Kant does, that this may always be the case is just to say that reasons for acting are always open-ended, or that any fixed sequence of reasons or considerations on any occasion are not (as far as reason goes) definitive. Such an open-ended nature of reasons for acting (on any occasions) may indeed be incompatible with the causality of reason in relation to actions being itself causally determined. Any causal determination of the practical deliberation will fix or determine a certain stock of reasons as decisive (as determining decisions and hence actions). But the open-ended nature of reasoning is just the idea that no fixed stock of reasons is decisive (conclusive, definitive), if indeed it is reason (and not extraneous foreign factors) that is the cause of the action. The point can be put by saying that it is a law of (practical) reason that on any occasion it is open-ended. The source of the incompatibility of such reasoning with causal determination is not just that the laws of reason are different from, say, physiological or psychological laws. Rather, the source is that this law of open-endedness implies that the operation of reason on any occasion is incompatible with the fixity of any causal determination of the reasons. Note that my contention is not that reason is open in some counterfactual way—namely, that if on an occasion I had been given other considerations I would have (or might have) decided differently. In this manner, indeed, my practical consideration would be reason-sensitive or open to influence by further reasons if these reasons had arisen. This reason-sensitivity is compatible with the causal determination of the actual episode of practical deliberation, since it is compatible with a fixed determined stock of reasons being, in actuality, decisive. However, this reason-sensitivity still leaves it that reason itself has not determined my action, since extraneous factors (the causal determinants) have fixed certain reasons as decisive, which, by the open-ended nature of reasons, aren’t decisive.13 Kant says, “The causality of reason we do not regard only as a co-operating agency, but as complete in itself ” (A555, B583, p. 477). It is the very open-endedness of reason that prevents it from being a merely “co-operating agency.” Reason cannot be a partial determinant because 13. For a compatibilist view of the reason-sensitivity or reason responsiveness that I am rejecting, see James Martin Fisher, The Metaphysics of Free Will: An Essay in Control (Oxford: Basil Blackwell, 1994).

212

freedom and morality

any other component that fixes the stock of reasons is incompatible with its being genuinely open-ended reason that is a partial determinant. Anything that fixes and hence closes off reasons has to be destructive to the nature of reason (as a determinant) itself. It is not just undue influence of desire or brainwashing that skews reasoning. Any determinant, including physiological or chemical determinants, “skews” reason simply for causally determining it and hence fixing it. If I am correct, then the very nature of practical reasoning as openended is incompatible with practical reasoning being causally determined, and hence practical reasoning cannot be identical to any causally determined (or, for Kant, natural) processes or occurrences. But why think that the nature of practical reasoning is in this manner open-ended? The open-ended nature follows from two plausible premises. First, the fundamental question or issue of practical reason is how I should live my life (or what conception of a good life it is that I should adopt), and this question is open-ended. There are no conclusive or definitive reasons for any specific conception of what life to live. The issue of the goodness or desirability of various conceptions of life is like the issue of the goodness or worthiness of a play or painting or novel. There are endless considerations, pro and con, that enrich the assessment, but no definitive or conclusive stock of reasons that settle the assessment once and for all. Coming to tentative decisions on the matter of what to do with one’s life is like coming to tentative appreciations of a work of art based on considerations or reflections upon it. Just as no aesthetic appreciation is definitive, so too no (tentative) appreciation for a certain way or certain ways of living is definitive. Second, this fundamental issue of practical reasoning governs, at least implicitly, all cases of practical deliberation as to what to do on various occasions. My decisions as to what to do are guided by (or at least must be consistent with) my conception of what I want out of life. Thus the open-endedness of the latter conception carries over to deliberations as to what now to do, in the sense that the reality of there always being further pertinent considerations regarding my life plan is also the reality of there always being further pertinent considerations regarding what now to do. Thus the reasoning that is germane to any of my practical deliberations is open-ended.14 If we accept these two premises, then it follows that practical reason is never definitive or conclusive. This doesn’t mean that it never leads to decisions, but rather that its leading to 14. For similar views on the open nature of life-decisions and practical deliberations, see Charles Taylor, “Responsibility for Self,” and Gary Watson, “Free Agency,” both in Free Will, ed. Gary Watson (New York: Oxford University Press, 1982).

Reason, Freedom, and Determinism

213

decisions must be compatible with its open-ended nature, and our contention, so far, has been that any causal determination of reasons unto a decision on an occasion is not compatible with this open-ended nature, since the causal determination fixes on a particular stock of reasons as (causally) decisive. For Kant, of course, moral considerations are predominant in practical reasoning and, for the most part, are not open-ended. Thus for Kant, that a particular action is a lie for the sake of my own advantage is a definitive reason for not doing it. This definitiveness of moral considerations, however, is compatible with the otherwise open-endedness of my conception of how to live. Morality, by itself, doesn’t settle the issue of what I should do with my life, and hence doesn’t settle particular occasions of practical deliberation. Even if I shouldn’t lie to a customer wanting to buy a used car, what should I do? Should I tell him the truth or should I change my occupation? Should I leave for Tibet to pursue my spiritual development? Recall our contention that for Kant morality pertains to the autonomous component of freedom, not to the undetermined or choice component; viz., only if my options are consonant with morality are they unpressured or nonconstraining options. This still leaves various options regarding which practical reasoning is openended, and so incompatible with being fixed by causal determinants.15 The open-endedness of reason, I now suggest, is consistent with decisiveness being fixed by undetermined choice. Reasoning, even though open-ended, must come to a close in a decision. If it is my undetermined free choice at any stage to either close off or else to continue deliberating, then reasons gone through never determine my choice. Thus, if I do choose to close off deliberation, it is nevertheless the case that I close off a genuinely open process, since it was possible as well for the deliberation to continue. Only undetermined choice can bring a genuinely open process to a close. Indeed, the process is genuinely open in precisely the fact that what brings it to a close is undetermined (by the process itself or by the process together with other operative factors). What I am suggesting, then, is that genuine practical reasoning must “realize” in the concrete the open-ended nature (the nondefinitiveness) of reasoning. This can only be so if the concrete process realizing it is a genuinely open process on the occasion, and this in turn can only be the case if the closure of the process is undetermined (whatever further factors are operative on the occasion). 15. In a somewhat grander vein, even morality is open to reasonable consideration for Kant. For example, without God and immortality moral laws might be otiose or empty. This doesn’t mean that moral laws are not definitively obligatory for Kant, just that they are not definitively closed to reasonable consideration.

214

freedom and morality

The open-endedness (nonconclusiveness) of reasons is just the fact that further reasons are always pertinent to any decision. If this is to be realized concretely, then the process of reasoning (to a decision) must be such that after any fixed set of reasons or considerations is gone through further reasons are causally open (this being the concrete pertinence of further reasons). But only if the process’s closure is undetermined (closed by undetermined choice) are further reasons still open after any fixed stock of reasons. It is true that some factor other than formulating reasons—namely, undetermined choice—operates in the concrete to bring deliberation to a close, so that it is not reasons themselves that are the source of the decision. However, the very nature of reasoning is to lead to a decision in such a way that the formulated reasons themselves don’t (on the occasion) fix the decision. The reasoning being closed by choice is a factor other than formulating reasons, but not a factor foreign to the process of reasoning, since reasoning itself demands that, in the concrete, its decisiveness is not determined but genuinely open. Free choice is not extraneous or alien to genuine reasoning, but the very way that genuinely open reasoning can be decisive. If this is correct, then we get the following “incompatibilist” account of genuine practical reasoning. Deliberation continues until a choice to close it off, which choice is, at the same time, the decision (the decisiveness) to act. The fact that the decision to act is the same as the choice to close off the deliberation implies not only that the practical deliberation process itself is not causally fixed, but also that it does not cause or determine the decision to act (since the deliberation isn’t a finished or complete event before the decision to act). What is true is that the choice to close off deliberation may become more likely as stronger reasons accumulate, but (consonant with the nondefinitiveness of reasons) this never determines the choice, and so the decision to act. Because the choice becomes more likely, we may say that practical deliberation guides the choice and so the decision. Our contention then can be put as follows: The open-endedness of reason requires practical deliberation itself to be causally undetermined (as to its closure),16 and in turn to guide, but not determine, the action.17 This con16. Practical reasoning may be undetermined in facets other than its closure (for example, as to which considerations are generated, as to how much relative weight is given to a consideration, etc.). For our purposes we stick to this one facet of closure (note, however, that since whether I deliberate further is undetermined, to this extent which considerations are generated—whether any further ones or not—is undetermined). 17. For other incompatibilist accounts of practical reason, see Robert Nozick, “Choice and Indeterminism,” in Agents, Causes, and Events, ed. Timothy O’Connor (New York: Oxford University Press, 1995), 101–14; Laura Waddel Eckstrom, Free Will: A Philosophical

Reason, Freedom, and Determinism

215

trasts with a “compatibilist” account according to which practical reasoning is both itself determined and causally determines action.18 The view I have sketched (if not all the details) is a development of what we may regard as Kant’s strategy against a compatibilist conception of freedom. The compatibilist contention is that freedom is equivalent to determination by unconstrained (unpressured, non-skewed) practical reasoning, and this in turn is compatible with practical reasoning itself being determined. As long as the chain of causal determinants of action goes through practical reasoning, that action is free. Any further issue of whether the agent “could have done otherwise” is basically beside the point, because being able to do otherwise now reduces to being able to act against or without reasoned decision, and being thus able is nothing worth having, and so no part of a valuable conception of freedom. Kant, by directly arguing from the nature of practical reason itself (rather than from a notion of being able to do otherwise) directly undercuts the compatibilist’s account of freedom. The import of being able to choose (and so perhaps to do) otherwise is that without it there is no genuine (i.e., open) practical reasoning at all. In other words, for genuine practical reasoning to be the source of my action, it must be open to me to close off deliberation or else go on with it (with the possibility that I come to alternative decisions as to what to do). In the incompatibilist view, either choice (to go on deliberating or to close it and act) is reasonable. Indeed, the very open-ended nature of reason is equivalent to the fact that reason doesn’t determine (or preclude) either choice. Being able to choose (and thus perhaps do) otherwise, therefore, is no longer being able to go against practical reason, but essential to genuine practical reason, which leaves open (by the very nature of reason) alternative choices.19 This, I believe, answers Galen Strawson’s objection against the indeterminist20 that if the agent has no principles of choice “governing what decisions it makes in the light of its initial reasons for action, then the decisions it makes are rationally speaking random; they are made by an agent-self that Study (Boulder: Westview Press, 2000), 106–9; and Robert Kane, The Significance of Free Will (Oxford: Oxford University Press, 1996). 18. Once the concrete realization of the open-endedness of reason is identified with an “open” process (closed by undetermined choice), the transference of the open-endedness of the issue of how to live our lives to the issue of what to do on an occasion is as follows: The choice to further consider our life plan or not is always implicated in our deliberations of what to do now, and so the latter deliberation is closed off or not by at least that choice. See Nozick, “Choice and Indeterminism.” 19. What we “miss” by our practical deliberation being determined to close up at a certain stage is having the reasonable alternative of continuing deliberation. 20. See Galen Strawson, “Libertarianism, Action, and Self-Determination,” in O’Connor, Agents, Causes, and Events, 27.

216

freedom and morality

is, in its role as decision maker, entirely non-rational ” (emphasis added). Only an implicit conception of reason as closed or as fixing outcomes would imply that free choice goes beyond reason. That there are no principles of reason governing choice (as to closing off deliberation) is demanded by the nature of reason itself. The choice I make is not random for being undetermined by reason, but reasonable by the openended nature of reason. In making either choice, then, I am expressing the realization of genuine open reason in my deliberation. Furthermore, this Kantian strategy of deriving indeterminism directly from the nature of practical reasoning allows the indeterminist to inherit the plausibility of the compatibilist’s conception of freedom, that a free act is an act due to my own (unconstrained, unpressured) practical reasoning, since the indeterminist accepts that conception when understood in a nondeterministic sense (as involving undetermined choice). Thus, for example, if I can identify myself with my practical deliberation (with being the author of my decisions), so that an act due to my practical deliberation is an act that is due to or is up to me (who I am), this is also the case when that deliberation involves undetermined choice. Hence, identifying myself with practical reasoning includes identifying myself with the choice that is integral to it. Further, if my being in control of my actions is a matter of their being due to my practical deliberation, then again, since choice is internal to such deliberation, the action is in my control despite the fact (indeed, because of the fact) that it is due to undetermined choice. Finally, this Kantian strategy, which argues for indeterminism directly from the nature of reason, allows the indeterminist to derive that responsibility pertains to free choice (and even to being able to do otherwise). Along with the compatibilist he may hold that we are responsible for our actions when they are governed by our own reasoned decisions. But for the indeterminist deciding to act is identical to choosing to close off deliberation, and so being responsible for one’s reasoned decision is being responsible for one’s choice to close off deliberation. Since it was open to continue (which opens the possibility of counterconsiderations arising, leading later to a choice of not so acting), one acts from a reasoned decision when one could have (decided and) acted otherwise.

3 . t r a n s c e n d e n ta l f r e e d o m It follows from our discussion that if determinism is true for natural occurrences, then no case of genuine practical reasoning can be identical to any natural occurrence, since no case of genuine practical rea-

Reason, Freedom, and Determinism

217

soning is determined. If determinism is true, then freedom or acting from reason does not exist in natural occurrences.21 Kant’s basic point in the Observation to the Third Antinomy is that nevertheless naturally determined actions can also be due to freedom or reason. I begin by presenting a model that I believe captures most of how Kant thinks he can reconcile determinism with causality by free reason. Suppose the writer-director of a play that is in rehearsal allows the actors to make choices regarding their roles. More specifically, they are to choose how their characters will act and for what reasons their characters will act in certain types of situations that are presented to them by the director. We may presume that these actors, when presented with these situations, deliberate about what their characters are to do. We can call this their out-of-play practical deliberations. Each actor then makes a series of out-of-play practical decisions regarding the actions of his character in various situations. Now the writer-director, we presume, must adjust the script into an overall dramatically inexorable play, where each character behaves in types of situations as the actor has chosen in his out-of-play deliberations. Further, the character in the play will go through the considerations that the actor went through in choosing the character’s action (for each particular type of situation). We can call this the in-play deliberation. The director’s overall script is to make not only the character’s actions but also these in-play deliberations part of the inexorable dramatic unfolding. The director is to make it, that is, such that, given the unfolding of the play to a certain stage, the in-play deliberation is completely “dramatically” determined. Everything that goes on in the play then is inexorably dramatically determined.22 Since the character’s reasonings are fixed and closed (after all, the script is written) these reasonings are contrary to the open nature of practical reasoning. Thus genuine open reasoning does not go 21. A fortiori, practical reasoning cannot be programmed into a deterministic machine. Dennett argues that it is plausible to hold that there is a randomness to the generation of reasons, which randomness can be achieved by indeterministic processes but also can be achieved by deterministic processes that are “random” relative to reasons. However, programming a machine with a deterministic randomizing device still determinately fixes or closes off the reasons or considerations the program will run through. Thus if the machine’s reasons go on until a tossed coin comes up heads, since it is determined which toss that will be, it is determined that the machine will only reason so far, which, we have argued, makes its deliberation a closed process (viz., makes those specific reasons absolutely decisive), in opposition to the open-ended nature of reason. See Daniel Dennett, “On Giving Libertarians What They Say They Want,” in O’Connor, Agents, Causes, and Events, 43–56. 22. Note that within the play everything proceeds in accordance with the compatibilist’s conception of practical reasoning; viz., the determination of our actions “goes through” the determination of our practical reasoning, which determines our actions.

218

freedom and morality

on in the play. Nevertheless, despite the fact that the character’s actions and reasonings are completely determined in the inexorable unfolding of the play, genuinely open out-of-play reason is causally efficacious in producing the character’s actions. Hence, complete in-play determinism is compatible with the efficacy of free, undetermined out-of-play reason. This latter has real efficacy, since the in-play determined deliberations and actions arise from the out-of-play deliberations. Now, let the writer-director be God (Kant’s transcendentally free cause), and the play be the entire inexorably determined series of natural occurrences of the world, and let us, the “in-world” deliberating characters, also be actors whose undetermined out-of-world choices are consulted by God in his writing of the inexorably unfolding script. Then our in-world actions and reasonings are completely naturally determined, and so genuine open-ended reasoning does not exist in the natural series. Nevertheless our actions as characters are also due to our free choices (as consulted actors determining our role as characters), and so undetermined, free, yet causally efficacious reason is compatible with natural necessity or determinism. Since anything in time is causally determined for Kant, we may equate natural determinism with in-time determinism, and hence our undetermined choices, as well as their causality, are out of time.23 Indeed, Kant repeatedly says that the causality of freedom is nontemporal (A540, B568, p. 468; A551, B579, p. 475). Just as the actor’s choices for his character have out-of-play causality (by contributing to the determination of which complete script, not to the inexorability of the action within the script), so too our timeless choices contribute to which entire series of natural necessity exists, not to the necessity within the series.24 In this manner, the causality of open reason is outside the series.25 Roughly, our undetermined choices are part of the transcenden23. Thus our interpretation of Kant is essentially that of Allen W. Wood in “Kant’s Compatibilism” in Wood, Self and Nature in Kant’s Philosophy, 73–101. 24. Ibid., 88. 25. See Immanuel Kant, Critique of Practical Reason, trans. Lewis White Beck (New York: Macmillan, 1993), 104, where Kant says that even though all our actions are naturally predictable, still “the entire chain of appearances .l.l. depends upon the spontaneity as a thing in itself ” (emphasis added). See also ibid., 121, where he says that as “pure intelligence existing without temporal determination” a person can be the “determining ground of natural causality itself.” This contrasts with Allison in Kant’s Theory of Freedom, who says, “Certainly we need not conceive of ourselves as literally timeless beings, in order to regard ourselves as rational agents.” Allison thinks that somehow our undetermined rationality can exist in a “moment of spontaneity” where we are not so completely determined that we cannot exercise spontaneity. It seems to me to make no sense whatsoever as a reconciliation of undetermined freedom and determinism to abrogate the determinism component.

Reason, Freedom, and Determinism

219

tal ground of why there is just this series of natural necessity. Thus transcendental freedom (the transcendental ground of the natural series, or God) is not just a “model” for our free practical reason. Rather, our free practical reason (as “consulted” by God) is literally a component of transcendental freedom. Equivalently, practical freedom shares in the intelligible causality by a transcendental object, in that God grounds a series of inexorable natural necessities in conformity with that free practical reason. Since the natural series contains within it no “sufficient reason” for why just this natural series, there is room for a transcendental ground, and since our free choices are part of this ground, they are part of the sufficient reason for why there is a series of nature inexorably leading to particular actions of ours. This makes these actions at once determined by natural necessity, but also (via grounding there being the natural series leading to that action at all) grounded in genuine, open nonnatural practical freedom.26 The view I am attributing to Kant is really a modification of Leibniz’s view. For Leibniz, God’s grounding of reality is adjusted to the nature of individual monads (to their complete concepts, which contain the law of their necessary unfolding). For Kant, this grounding is adjusted not to any necessary law of the monad’s unfolding but to the monad’s free choices; viz., Kantian “monads” (a-temporal intelligible beings), unlike Leibniz’s, have free and undetermined reason in their intelligible nature. This view also has affinities with Kant’s doctrine of the Ideal or the Highest Good, according to which (as we hope) God arranges or adjusts the causal series of natural necessity in accordance with happiness existing proportionally to morality (which makes justice a real determining ground of nature).27 A key component of Kant’s resolution of undetermined reason with natural causal determinism is his distinction between our intelligible character and our empirical character. The intelligible character, though having effects in the natural series, is itself, together with its causality, outside the series (A537, B565, p. 467; A540, B568, p. 468; A541, B569, p. 469). The intelligible character, I suggest, is the pattern of the set of free choices (involving when to close off deliberations) exhibiting the open-ended nature of practical reason. In our model it is the pattern of reasoned choices that the actor makes for his character’s actions, or the pattern of out-of-play deliberations. The intelligible character is not identical to the empirical character, but rather deter26. In this manner Kant avoids free reason being epiphenomenal with regard to our actions, despite these actions being naturally causally determined. 27. The view also has affinities to contemporary teleological views that “explain” Planck’s constant by saying that if it had been different, life could not have arisen.

220

freedom and morality

mines it (A551, B579, p. 475) or grounds it (A545, B573, p. 471).28 Further, I suggest, the empirical character is not constituted merely by desires and inclinations but includes empirically determined deliberations. The empirical character, Kant says, allows us to estimate the actions of reason by the intelligible character (A549, B577, p. 474). Indeed, it is the sensible sign (A546, B574, p. 472) or the sensible schema (A553, B581, p. 476) of the intelligible character. The empirical character, that is, is the temporal “expression” of reason’s causality. Now, only if that temporal character includes reasonings (albeit casually determined reasonings), I suggest, can the causality of free reason be expressed temporally in the empirical character. So, for example, if our empirical character included only desires, it would be no different from an animal’s behavior, and so no more a sign or schema of reason than an animal’s behavior. The clearest passage showing that Kant believes that there is reasoning in our empirical character is in the Critique of Practical Reason, where Kant says, “In the question of freedom .l.l. it is really not at all a question of whether the causality determined by a natural law is necessary through determining grounds lying within or without the subject, or whether, if they lie within him, they are in instinct or in motives thought by reason” (emphasis added).29 The empirical character, then, includes the pattern of deliberations according to the determinist’s conception of practical deliberation (viz., each such deliberation is fixed or causally determined) which itself is not genuine open-ended reason or deliberation. In our model, the empirical character is the pattern of in-play determined deliberations the character goes through. This understanding of the empirical character, I contend, is important in several respects. First, it shows that Kant is not arguing from the nonempirical nature of thinking of reasons or considerations to the nonempirical nature, and so naturally undetermined nature, of reason, since those thinkings are part of the empirical character. Rather, he is arguing directly from the undetermined (i.e., open-ended) nature of reason to its nonempirical nature (since no natural phenomena are “open”). Second, it shows why, though all appearances have a transcendentally free ground (God), only human action expresses (is a sensible sign or schema of) distinct free grounds (in addition to God). Only empirically deliberating beings are “signs” of genuine open reasoned deliberation. So, for example, if we are told that some of what goes on 28. Recall that this was our textual objection to a monistic interpretation of Kant according to which acting from reason is token identical to empirical causes or phenomena. 29. Kant, Critique of Practical Reason, 100–101.

Reason, Freedom, and Determinism

221

in a play expresses genuine open out-of-play deliberation on the part of actors, it is clear which parts of the play do—namely, the in-play deliberations of the characters.30 Finally, just as the empirical character includes closed (naturally determined) reasoning and not just desire, so the intelligible character includes goals and motives (not just sheer open reason). To return to our model again, the director offers the actor a scenario in which certain options satisfy certain needs and desires of the character. The actor can contemplate and reflect on (viz., deliberate about) the worthiness or reasonableness of fulfilling such desires, without having to feel them,31 just as in a more normal vein we can deliberate about what we would do if we were starving on a lifeboat, and can do this just after eating a large meal. Similarly, the noumenal character pertains to desires and needs even if, as noumenal character, they are not felt.32 The idea of God consulting our free choices and arranging the course of natural necessity accordingly is, for Kant, a regulative idea that governs our taking responsibility for our (determined) decisions and actions. Indeed, the import of the idea is that it shows that taking such responsibility is consistent with natural determinism. Consider the actor going through his character’s deliberation at a certain stage of the performance of the play. Despite the fact that this deliberation is inexorably determined by what has gone on previously in the play, the actor, in going through that deliberation, can properly take responsibility for his in-play character’s deliberation, since it was due to his out-of-play free and open reasonable deliberation. Similarly, as I now deliberate, my idea of God’s adjusting natural necessity to my free out-of-time choice is just the idea that my situation now is that of the actor’s deliberation at a stage of the play. Thus the idea properly allows me, as it allows the actor, to take responsibility for my deliberation (and so my decision and action). This idea, that is, allows me now to take the viewpoint of being “suspended” from the chains of natural necessity, within which viewpoint my free choice determines my deliberation. Note carefully that the practical vantage point and so the taking of responsibility happens 30. This answers Bennett’s objection that “the noumenal nature of freedom, according to the theory, implies there can be no empirical evidence as to when freedom is present and when it is not.” See Jonathan Bennett, Kant’s Dialectic (Cambridge: Cambridge University Press, 1974), 201. 31. This answers Allison’s objection against timeless agency that timeless noumenal beings couldn’t be sensuously affected. See Allison, Kant’s Theory of Freedom, 47. 32. If one objects that we would reason differently if we felt the need, then one is thinking of cases where desire skews reasoning, and hence not cases of acting that are determined even by empirical (closed or causally determined) practical reason. Such cases don’t express noumenal freedom at all.

222

freedom and morality

now in time as I deliberate, although the theoretical idea that makes the adoption of this viewpoint consistent with determinism involves only the idea of timeless choice (choice suspended from the series of time). Thus, the theoretical idea is regulative of taking responsibility in time. The theoretical idea does not involve the idea that I now have free undetermined choice (as to closing off or continuing my present deliberation). Kant clearly sees that this idea would be flat out inconsistent with a determinism that encompasses all temporal occurrences. However, as the model of the actor in regard to his in-play character shows, my now having free choice is not necessary for my now properly taking responsibility. In a word, the idea of timeless agency (without any idea of free choices in time) can be regulative of taking responsibility in time.33 When Kant turns “to illustrate this regulative principle of reason [that reason is a free timeless cause] by an example of its empirical employment” (A554, B582, p. 477) in the example of blaming an agent for a malicious lie, he says not that free choice exists in the moment but rather that “in the moment when he utters the lie the guilt is entirely his” (A555, B583, p. 478). Supposing now that it is proper to ascribe blame or responsibility only when it is proper for the agent to accept guilt or responsibility, Kant’s illustrative example shows that it is taking responsibility that is the empirical (in time) employment of the regulative idea of timeless reason, not having temporal free and open choices. Besides allowing for taking responsibility in time for particular deliberations, Kant’s idea of free reason allows us to take responsibility for improving ourselves. The writer-director may ask the actor to make a sequence of choices, where the scenarios may include making choices consequent upon preceding choices made. The actor may respond in one of two ways to a “bad” choice he makes. He may compound this by making consequent choices that are bad, or he may ignore the bad choice made and make consequent better ones. Suppose the former is the case. Then in the course of the play the actor can properly take responsibility for compounded bad decisions (due to bad deliberations), and so take responsibility for failing to “overcome” his past bad decisions by making better ones. Once again the unfolding of the pattern of his decisions (and so whether there is improvement or not) is inexorably determined within the play, but since it is due to the actor’s own 33. Kant’s idea of timeless agency, then, is not for the purpose of thinking of myself as now free, but for the purpose of thinking of myself as free in regard to what I do now, and hence thinking of myself as now responsible. Jonathan Bennett objects to Kant that “The normal manageable use of ‘reason’ .l.l. is a use which operates within the phenomenal realm” (Kant’s Dialectic, 203). But in Kant’s view free and open-ended reason can (theoretically) operate in regard to the phenomenal realm (not within it) giving us a “normal manageable” use of taking responsibility within the phenomenal realm.

Reason, Freedom, and Determinism

223

out-of-play free-choice sequence, he is responsible for the failure of improvement.34 Kant’s theory licenses, I believe, properly taking responsibility for the very same sorts of acts as on a determinist conception of practical reasoning; viz., I am responsible for acts due to my own deliberation. Thus if desire overwhelms me (and its overwhelming me is not due to previous exercises of practical reasoning), I am not responsible in either the determinist’s view or Kant’s view. In Kant’s view, in such a case God would have arranged natural necessity (the desire) against my free practical choice, and hence God (and nature), not I, would be responsible.35 Bennett objects that since in Kant’s view God adjusts the past series of natural necessity to my choice regarding a certain action, therefore in Kant’s view I freely choose the past.36 The response to Bennett is that although in Kant’s view I may indeed freely choose the past (freely choose in regard to what happens before the action I am choosing in regard to), I am only responsible for the foreseeable consequences of my choices, and these are the same as in any other (determinist) view of deliberation and responsibility. Thus the actor in our model chooses for his character in the light of what consequences are likely to ensue. The adjustment of the conditions in the play leading up to that choice by the director are, rather, the responsibility of the director.37 It is on this basis, I believe, that Kant can say that we blame the agent “just as if the agent in and by himself began in this action an entirely new series of consequences” that disregards the past series of conditions (A555, B583, p. 477, emphasis added).38 But there is one way in which Kant’s account of ascribing and taking responsibility differs from the deterministic account of practical reasoning; namely, in Kant’s view I take responsibility not only for my actions but for my deliberations as well. Kant, that is, has the “strong” indeterminist sense of responsibility. 34. Again unlike Leibniz’s view, where the unfolding of monads (and so improvement) involves no free choice on the part of the monads. 35. This also answers Bennett’s objection that in Kant’s view even insanity (since it has a noumenal ground) would be free and responsible. See Jonathan Bennett, “Kant’s Theory of Freedom,” in Wood, Self and Nature in Kant’s Philosophy, 102–12. 36. Ibid. 37. Allen Wood, in his classic paper “Kant’s Compatibilism,” says, “Even if my choice somehow issues in a world containing the First World War .l.l. it seems reasonable to hold me morally responsible only to those events which must belong to the actual course of things because I have the empirical character .l.l. that I do” (92). However, it may be that certain past events must belong to the actual course of things to lead up to my empirical character. The difference, rather, is foreseeability by the chooser. 38. One may object that it is only in an ad hoc way that Kant’s view can be tailored to familiar practical responsibility. This is to miss the point that what Kant is looking for (theoretically) is any view that can support practical responsibility consonantly with determinism.

224

freedom and morality

4 . n at u r a l i n d e t e r m i n i s m We have argued that if determinism for natural occurrences is true, then only in a Kant-like theory in which practical deliberations are distinct from any natural occurrences (and the effectiveness of each deliberation in grounding behavior is distinct from any natural effectiveness), can the open-ended nature of practical deliberation be reconciled with natural determinism. Further we have argued that Kant’s view permits taking responsibility within the natural course of events. Although Kant’s theory is theoretically consistent, it is evidentially empty, there being no theoretical reason whatsoever to believe that we make timeless noumenal choices to which God adjusts the course of nature. With the advent of quantum mechanics and its indeterminism regarding “natural” (or at least naturally evidenced) occurrences, there is the possibility of reconciling the open-endedness of practical reasoning not only with a theoretically consistent view of reality but with a theoretically well-supported view. We begin by presenting a simple model of how quantum mechanics could realize the open-endedness of practical reason.39 We suppose that our deliberating, or our consideration of reasons, pro and con, for acting, is some macroscopic process of the brain, and that this process is connected to a potential barrier toward which a quantum particle can be directed. As positive reasons accumulate, we suppose that this decreases the potential barrier (making it more likely that if a particle is directed at the barrier it will go through), while as reasons against acting arise the potential increases (making it less likely the particle will get through). We now presume that at a stage of the deliberation there arises the state of taking stock as to whether to close off deliberation and act or to continue deliberation. Equivalently we can call this the state of readying oneself for a decision. We presume that this state is identical to sending a particle through to the barrier. The choosing, then, is to be identified with the macroscopic interaction of the brain with the particle-barrier system, which “collapses” the system into either the particle having gotten through the barrier (which ends deliberation and makes the organism decisive as to going ahead to act), or else the particle having been blocked by the barrier (which makes the organism continue with further deliberation). In this model my deliberation is a genuinely open process, since it is genuinely open for the par39. This model essentially follows Robert Nozick’s suggestion in Philosophical Explanation (Cambridge: Harvard University Press, 1981), 298. For a different quantum mechanical model, see Kane, The Significance of Free Will, 128, 141.

Reason, Freedom, and Determinism

225

ticle to get through or not. We suppose further that if the deliberation continues and negative reasons come to dominate, the mechanism attaches instead to a second barrier system, which is such that if a particle is directed to this barrier, its getting through leads the organism to decide not to act, and its being blocked leads the organism to continue with further deliberation. In this way, the model allows that my reasons don’t determine the outcome (the decision to act or not), though they may, in changing the probabilities, guide the outcome. Recall now that the fact that it is free choice that closes off deliberation implies both that the deliberation is not causally determined and that the deliberation doesn’t causally determine the decision to act (the latter, since the deliberation is not closed off separately from the decision to act). Our model now satisfies both these features of free choice, and hence genuinely open practical reasoning.40 Notice that in this model the person’s character may influence (or even determine) which specific considerations arise and what their weights are, but it still doesn’t determine the person’s deliberation. The openness of deliberation “blocks” anything, including character from determining decision. Since, we may presume, the decision I do make (and what ensues on it in my life) can influence my future character, it follows that the openness of reason gives me the ability to “reform” my character (so that if I don’t re-form it that is my own free choice). Neither character nor deliberation determines my decision because nothing determines my decision other than making it (viz., other than the macroscopic brain interacting with the system after the particle is sent in).41 One might think that because nothing but free choice determines the decision, therefore the decision is beyond the governance of my reason, but this is to forget that being guided (but not determined) by reasons is what it is for a decision to be governed by genuinely open (nondefinitive) reason. Hence the decision is in the governance of my reasonable deliberation in the exact sense in which genuinely open reason can govern decision. If, as with the determinist, the decision is mine for being governed (determined) by my deliberation, this is equally so in the indeterminist account (viz., the decision is mine for 40. Our model leaves open again whether particular reasons arise deterministically or not, whether the weight of reasons is deterministic or not, whether a decision to act involves further indeterministic effort (trying) to act or not, etc. We are only attempting to model the open-ended nature of deliberation, that reasons are never definitive. 41. Compare this to Carl Ginet’s contention that when an event is one’s own action “then one’s determining it requires only that one perform it.” Applying this to our case we get that the choice (decision) is our own action. See Carl Ginet, “Reasons Explanation of Action: An Incompatibilist Account,” in O’Connor, Agents, Causes, and Events, 69–94.

226

freedom and morality

being governed—guided—by my deliberation). Further, if, as with the determinist, the decision is intentional for being formed as a result of my own deliberation, then likewise in the indeterminist account, the decision, whichever way it goes, is intentional for being formed in relation to my own reasonable considerations. Of course in the indeterminist case there are two possible outcomes (closing off deliberation or continuing), but both of these are results of nondefinitive reasoning. As an analogy, suppose I decide whom to telephone by blindly shuffling through my address book and pointing to a number. Whichever number I pick, I intentionally call that person. In understanding the nature of my decision process I understand that I am letting the blind selection form or determine my specific intention (not abandoning my intention to something else). Similarly, my understanding of the open, or nondefinitive, nature of practical deliberation is my understanding that I must leave my intention to be fixed by free choice. In general, in the determinist view, whatever factors go into a decision to make it my own reasonable and intentional decision equally go into a decision in the indeterminist view, since all the same factors are present. The only difference lies in the relation of those factors to the decision. Still, one wants to object that this difference in relation somehow puts the decision out of my control just at the point at which the decision is “given over” to pure chance. Once I ready myself for a decision, or am about to take stock by sending the particle to the barrier, at this very point the rest is pure chance. At this point I lose control and, so to speak, stand by helplessly awaiting the indeterminist outcome, like someone who has given over his decision to the spin of a roulette wheel then watches his fate helplessly. The difference, however, is that in the quantum-mechanical case I never give the decision over to a further process. Nothing happens between my giving the decision over (interacting with the quantum system) and the outcome (the particle being on the other side of the barrier or not). It is one of the “mysteries” of quantum mechanics that nothing happens for the particle to wind up on one side of the barrier.42 Thus, my (exercising) free choice and my specifically choosing (say, to close off deliberation) are not distinct. And so, unlike the case of the roulette wheel, there is nothing to lose 42. At least this is true in the classical interpretation of the collapse of the wave packet. In a decoherence view there is a “process” that constitutes the interaction leading to the particle being on one side or the other. But this process isn’t one that at any stage tends toward one state rather than the other (beyond the probabilities determined by the potential barrier, which in our case are determined by the weight of reasons). Hence the choosing (the still-free choice) can be identified with the entire decoherence process, so that it remains true that nothing intervenes between the free choice and the outcome.

Reason, Freedom, and Determinism

227

control of between the choosing and the specific choice or outcome.43 This I believe captures one aspect of making a choice that agent-causation theories44 have been anxious to capture, viz., that in my choosing I directly (not via some event or process) produce the decision or outcome. It captures it, however, not by a notion of agent-causation but by the fact that there is no event by which my free choice (the macroscopic interaction with the quantum system) produces the specific outcome. The event of freely choosing is simply not distinct from the event of choosing a specific outcome, since there is not first the interaction and then the outcome. Quantum-mechanical indeterminism, I believe, then gives a model (a concrete realization) of the open-endedness of practical reason. The basic strength of this model over Kant’s is that it gives real theoretical possibility to freedom, even in the sense that there could be empirical evidence for it, whereas in Kant’s view, by his own reckoning, freedom is consistent with theoretical reason, but not possible (in any real sense) theoretically. Note, however, that the quantum model may share with Kant’s model the feature that freedom goes beyond theoretical cognition. If we presume that the empirical evidence for quantum mechanics is sufficient for thinking that reality has quantum-mechanical structure, while at the same time we presume that we have no cognition of what the real individuals are that has this structure, then quantum reality will be beyond theoretical cognition. Of course, for Kant real theoretical (scientific) possibility cannot outstrip the limits of cognition, but this, I am suggesting, may be a mistake on his part. Our understanding of quantum-mechanical structure ontologically determining macroscopic reality may be enough to secure that this structure is real, while yet not enough to secure a cognition of individual elements that have this structure.45 If so, there is a possible wedge between scientific theorizing and cognition (even scientific cognition). Since the freedom of practical reason now would be on the side of the wedge beyond cognition, it would be the case that on the quantum mechanical model freedom, though really theoretically possible, is, as in Kant’s 43. Here is one way, then, that the mystery of quantum mechanics dovetails with the mystery of freedom of the will. 44. For such a theory, see Timothy O’Connor, Persons and Causes (New York: Oxford University Press, 2000), and Roderick Chisholm, “Freedom and Action,” in Freedom and Determinism, ed. Keith Lehrer (New York: Random House, 1966). 45. According to decoherence theory, one can give an account of how the quantum mechanical structure ontologically determines the macroscopic without giving an account (or interpretation) of what the reality of elements is that has this structure. If we presume that cognition requires the latter, while real theoretical possibility (even actual theoretical evidence of real structure) requires only the former, then decoherence theory is a case of real possibility going beyond cognition.

228

freedom and morality

view, beyond the limits of cognition. Of course this model violates Kant’s belief in causal determinism in the temporal realm, but whatever the cogency of Kant’s arguments for this determinism,46 it remains an enduring accomplishment of Kant’s discussion of freedom in the Third Antinomy that he derives undetermined choice directly from the nature of practical reason itself, since this cuts directly at the heart of the compatibilist conception of freedom. 46. It is perhaps interesting to note that the quantum mechanical violation of determinism also “violates” Kant’s grounds for holding determinism. Those grounds are, I believe, that causal determinism among phenomena is what “carries” or “realizes” the necessary time order (viz., that preceding times necessarily emerge into succeeding times). The quantum mechanical interaction from a superposition to some determinate state seems to leave it perfectly open how time “progresses” within it (whether time stands still, loops around, or goes from the preceding to the succeeding) in either the classical or the decoherence view.

12

Kant’s Formulations of the Categorical Imperative

1. introduction Kant’s second formulation of the categorical imperative commands respect for rational agents as ends in themselves. Korsgaard has interpreted this as commanding respect for rationality as the source of the value of our goals in life. She says, “But the distinctive feature of humanity as such is simply the capacity to take a rational interest in something: to decide under the influence of reason that something is desirable, that it is worthy of pursuit or realization, that it is deemed to be important or valuable not because it contributes to survival or instinctual satisfaction, but as an end for its own sake. It is this capacity that the formula of Humanity commands us never to treat as a means, but always as an end in itself.”1 In Korsgaard’s conception our rationality consists in setting ends, or in determining what ends to pursue in order to flourish or achieve well-being. However, this conception of our rationality as agents leads to a “universalization” problem. Even if it is granted that I must respect my own rational nature as the source of my goals, why must I also respect this capacity for setting ends in others? Couldn’t an egoist, using his rational capacity, come to the conclusion that certain of his goals were more important than other people’s? One may respond that in failing to respect the rational capacity of others, he can no longer value his own goals as deriving from the rational capacity in him. But the exercise of his reasoning tells him that his goals are more important than other people’s. Hence if he respects the reasoning capacity of others over his own goals, then he is disrespecting his own reasoning capacity. I do not want to suggest that the “universalization” problem is insoluble. I do suggest that in Korsgaard’s understanding of Kant it becomes the crucial issue in validating a categorical imperative in terms of the 1. Christine Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996), 114.

229

230

freedom and morality

formula of humanity.2 It is noteworthy that Kant himself sees no such problem or issue. Immediately, without any consideration, Kant equates respect for rationality in others with respect for one’s own rational nature. What is a thorny, difficult, and crucial issue for Korsgaard seems to be no issue at all for Kant. This suggests that Kant is working with a quite different conception of rationality, one according to which there simply is no “universalization” problem. Further evidence that the second formulation of the categorical imperative is not based on Korsgaard’s conception of rationality (as the source of our goals and values for flourishing in life) comes from Kant’s belief that the second formulation is straightforwardly equivalent to the first formulation in terms of universal willing. But now the first formulation (to allow only those maxims that can also be willed as universal laws) doesn’t on the face of it have anything to do with a conception of rationality as the source of our ends. It seems rather to turn on some sort of “contractualist” conception of rationality as universal agreement, or what is acceptable as a universal law. Since Kant believes that the equivalence of the two formulations is straightforward, it would seem that the second formulation in terms of humanity as an end should turn on the very same conception of rationality as the first does. Finally, Korsgaard’s view does not sit well with the fact that in passage after passage, in the Foundations and the Critique of Practical Reason, Kant characterizes setting goals or ends as being pathological or heteronomous in contrast to the autonomy of our rational nature. All of this suggests that the Korsgaard’s conception of rationality is simply not Kant’s conception at all. I wish to contend rather that Kant is a contractualist, and that contractualism is what is expressed in all three formulations of the categorical imperative. Further, Kant’s defense of contractualism depends on his denying Korsgaard’s view that autonomous rationality is the source of our goals for flourishing. These contentions, I hope to show, explain why there is no “universalization” problem in Kant’s second formulation, why he thought it was so obvious that the second formulation is equivalent to the first, and why he goes out of his way to contrast rationality with seeking well-being or happiness.

2 . t h e f i r s t f o r m u l at i o n The first formulation of the categorical imperative gives the content of morality (defines the difference between right and wrong), while the 2. Ibid., 122–23.

Formulations of the Categorical Imperative

231

second and third formulations give the motivational basis for being moral (in terms of respecting one’s own humanity and autonomy, respectively). In this section I shall briefly review the first formulation. There is nothing particularly new in this review, and many important issues are not discussed. The point is simply to show that the first formulation is basically a contractualist account of the content of morality, according to which an act is wrong if it violates a law that everyone can agree to, or contract for. On the specific version of contractualism pertinent to Kant, the basis of the universal agreement to there being a particular law is the “prudence” or self-interest of there being that law. Thus it is in everyone’s interest, including the thief’s, that there be laws against stealing. Therefore everyone, based on prudence or their selfinterest, will agree to there being a law against stealing. This defines stealing, in violating such a law, as wrong. In this version of a contractualist view, self-interest is involved in determining what can be universally agreed to, and hence is involved in the definition of what is right or wrong. It does not follow, however, that prudence or self-interest is the motivation for morality. Indeed, the motivation for morality would determine why one conforms one’s behavior to such laws even when it isn’t in one’s interest to do so (as when one can get away with violating the laws), and so the motivation cannot be one’s self-interest. In the Foundations of the Metaphysics of Morals (hereafter Foundations) the first formulation of the categorical imperative says that an act is wrong if its maxim cannot be willed to become a universal law (Foundations, Ak 422, p. 39).3 The maxim of an action includes within it inclination or private interest. The maxim is to act a certain way to further or promote a certain interest. If the maxim’s becoming a universal law contradicts this interest, then the law contradicts the maxim that contains the interest. In this way, willing of the maxim as a universal law would be literally inconsistent with the maxim. Thus, if the maxim is to steal in order to obtain goods which one can use securely, the universal law would contradict this interest, since with stealing as a universal law of behavior there would be no security in the use of goods. In Kant’s view, then, it is interests that determine what can be willed as universal laws, and hence what is or isn’t moral. If, as in Kant’s view, I cannot will stealing to be a universal law, that is because it is against my interest that others steal. But then it is in my interest that others don’t steal, in which case it is prudent for me to agree 3. Immanuel Kant, Foundations of the Metaphysics of Morals, trans. Lewis White Beck (New York: Liberal Arts Press, 1959), 39. “Act only according to that maxim by which you can at the same time will that it should become a universal law.” This appears on p. 422 of the Prussian Academy Edition, vol. 4, which, following Beck, I refer to as Ak 422.

232

freedom and morality

to a law against stealing. It is likewise prudent for anyone else to agree to such a law. Thus the same action of stealing to obtain goods that is wrong for Kant for one’s not being able to will it as a universal law is also wrong for violating what can be universally agreed to (what can be contracted for). Kant’s first formulation, thus, is a version of a contractualist theory in which self-interest or prudence is involved in the very “content” of morality (the very definition of what is right or wrong). It is the prudence or imprudence of there being certain laws (viz., the consistency or not of such laws with one’s interests) that defines what is or isn’t moral. In applying the principle of the categorical imperative, Kant characterizes certain maxims as being such that they “cannot be thought as a universal law of nature without contradiction, far from its being possible that one could will that it be such” (Foundations, Ak 425, p. 42). Thus, in thinking of making a promise with the intention of not fulfilling it, the universalization would supposedly make the promise itself impossible as an act. Kant seems to be saying that a condition for there even being an act of promising is that promises generally be made with the intention of fulfilling them, since otherwise they wouldn’t induce expectations in others. The impossibility of willing a universal law here is not just the imprudence of such a law, but the literal impossibility of it. This seems to give a test of consistency, and so morality, that is independent of whether or not it is prudent that there be a certain universal law. Note first, however, that it is not clear that Kant is even correct in his contention. If a promise is an undertaking of an obligation, it seems one can undertake it even if no one expects one to keep it. As evidence of this, suppose I am the kind of person who doesn’t keep his promises. Then when I “make a promise” it will not induce expectations in others. Nevertheless, surely I have still made the promise and can be held responsible for it. If so, then the act of making a promise does not entail that anyone expect it to be kept. Therefore, even if the universal breaking of promises would defeat any expectation of their being kept, this would imply only that promising would be otiose or useless, not that it would be impossible. Whatever the cogency of Kant’s contention that some actions would not even be thinkable as universal laws, the consistency of universal law with the action alone (without the purpose) is certainly not the generic principle of the categorical imperative. Even in the case of promising itself, Kant doesn’t stop at such supposed inconsistency, for he says the universal law would make “the end to be accomplished by it [the promise] impossible” (Foundations, Ak 423, p. 40). In other cases, such as with maxims of failing to assist others, Kant says, “although it is possible

Formulations of the Categorical Imperative

233

that a universal law of nature according to that maxim could exist, it is nevertheless impossible to will that such a principle should hold everywhere as a law of nature” (Foundations, Ak 424, p. 41). There is no impetus on Kant’s part, then, for putting sheer consistency with actions of maxims in place of consistency with inclinations expressed in maxims. Nor does Kant need to avoid all involvement of inclination in his definition of right and wrong in order to ensure that such interests form no part of the motive for doing right.4 Indeed, the first formulation doesn’t at all concern the motive for doing right (doing what can be willed as universal law). Kant says that the first formulation of the categorical imperative is also expressible as follows: “Act as though the maxim of your action were by your own will to become a universal law of nature” (Foundations, Ak 422, p. 39). In the Critique of Practical Reason (hereafter Critique) he calls this formulation a type of the moral law, which is to say it is something like a schema that makes the content of the moral law more intuitively applicable (Critique, Ak 69, p. 72).5 In the Critique Kant says the issue (of right and wrong) is whether a person would “assent of his own will to being a member of such an order of things where, for example, everyone deceives if they think it’s to their own advantage” (Critique, Ak 69, p. 72). An act is wrong, then, if one cannot assent to its being a universal law, or if it is imprudent that there be such a law. Kant goes on to say that in judging actions according to the imperative we all know full well that if we secretly act in a way that we cannot will as a law of nature, “it does not follow that everyone else will do so” (Critique, Ak 69–70, pp. 72–73).6 His response is that “this com4. Kant’s own stated purpose in making this distinction between what cannot be thought without contradiction, versus what cannot be willed consistently with the purpose in one’s maxim, is to distinguish strict duty from broad duty. What it distinguishes at best, I believe, are those duties that can be characterized as living up to expectations versus those that cannot be so characterized. This would cover strict duties such as keeping promises, telling the truth, and not stealing, but not duties such as refraining from killing or maiming. A universal law of maiming or raping would not make them impossible; it would only make them (contrary to inclination) rampant. I am following Korsgaard here (see Creating the Kingdom of Ends, chap. 3). 5. Immanuel Kant, Critique of Practical Reason, trans. Lewis White Beck (New York: Macmillan, 1993), 72. This appears on p. 69 in vol. 5 of the Academy edition. Again, following Beck, I refer to it as Ak 69. The reason that this “law of nature” variant is not an exact expression of the moral law is that adoption of maxims consistent with moral principle is a matter of freedom, not nature (see the Critique, 79). Whereas the first formulation itself is silent as to the motive of morality in freedom, the “law of nature” expression actually contradicts the motive (and so is only a type or schema for applying the law). 6. See also Foundations, 57 (Ak 439), “But a rational being, though he scrupulously follow this maxim, cannot for that reason expect every other rational being to be true to it.”

234

freedom and morality

parison of maxims of his action with universal natural law, therefore, is not the motive of the will” (Critique, Ak 70, p. 73).7 Hence, although one assents to a law based on whether it is prudent that there be such a law (whether one wants to live in such a world), one does not conform or subject oneself to the law based on whether it is prudent to do so. Prudence or self-interest, then, is involved in which laws can be willed, and so is involved in the very content or definition of morality. However, prudence is not therefore the motive for obeying the laws that can be willed, and so for doing what is right. As Kant sees in the current passage, making prudence or self-interest your motive would immediately lead to secretly violating the universal laws one wills when one can get away with it. In relation to the variant expression of the first formulation, then, Kant clearly distinguishes that which defines morality (right vs. wrong) from that which motivates it. He makes the same distinction in relation to the first formulation itself. He says in the Foundations that when we do something wrong we don’t will the maxim of our transgression as a universal law but rather we will that “the contrary of this maxim should remain as a law generally” (Foundations, Ak 425, p. 42). Here the point he makes in the Critique is reiterated—namely, that allowing ourselves exceptions to following the laws is not a violation of prudence, so long as they leave the universal law as a general law with only a few exceptions. This time, however, Kant is making the point in regard to willing universal law (not in regard to laws of nature) and so in regard to the original formulation of the categorical imperative. Kant ends the entire discussion of the first formulation of the categorical imperative by saying, “we have also clearly exhibited the content of the categorical imperative which must contain the principle of duty (if there is such).l.l.l. But we are not advanced far enough to prove .l.l. that there is a practical law which of itself commands absolutely and without any incentives” (Foundations, Ak 426, p. 43). In other words, the coherency of universal law with inclination gives the content of morality, but such inclination or prudence is not the incentive for siding with the universal law, and so is not the motive for acting morally. Indeed, Kant believes the imperative force must be categorical, which is to say that conformity to universal laws cannot he contingent upon my interests, and so upon whether it is prudent on the occasion to obey the law. To sum up, the first formulation of the categorical imperative is a contractualist account of the content or definition of morality; an act is 7. In Foundations, 57, (Ak 439), he says that even though we cannot count on its favoring our happiness, “Still the law .l.l. remains in full force.”

Formulations of the Categorical Imperative

235

wrong if it violates laws one can willingly contract to (assent to or will) based on its being in one’s self-interest that there be such laws. This view is compatible with prudence or self-interest being no part of the motive for behaving in accord with such laws. Kant does not consider the motive for conforming to universal law, and so for behaving morally, until the second formulation, to which we now turn.

3 . t h e s e c o n d f o r m u l at i o n After the discussion of the first formulation of the categorical imperative, Kant says that if judging actions by the imperative is to be a necessary law for rational beings, “it must be connected .l.l. with the concept of the will of a rational being as such” (Foundations, Ak 427, p. 44). This, in turn, he says, leads to a search for “objective ends, which depend on motives valid for every rational being” (Foundations, Ak 428, p. 45). In contrast to the discussion of the first formulation, then, Kant is now clearly concerned with the motive for submitting my actions to the test of the categorical imperative. Further, he is concerned with a motive that is independent of what is prudent according to inclinations. To get at this motive he says that man exists as an end in himself, that he must always be regarded as an end, and that he necessarily thinks of his own existence this way. Kant contrasts a person’s being something whose existence is an end in itself with the merely conditional worth of inclinations. He says, “The inclinations themselves as sources of needs, however, are so lacking in absolute worth that the universal wish of every rational being must be indeed to free himself completely from them” (Foundations, Ak 429, p. 46). I believe Kant is saying in these passages that the identity or personhood of a rational being consists in rationality governing his actions, as opposed to his actions being governed simply by inclinations. For a rational being, that is, rationality is not one capacity among others that can be weighed against other aspects of his being by any external or extrinsic standards. His rationality is not a means to anything else but is definitive of what he is. Kant talks of inclinations as having only conditional worth, while our own nature as rational beings has absolute worth. He goes on to say that without rational beings, “nothing of absolute worth can be found and if all worth is conditional and thus contingent, no supreme practical principle of reason could be found anywhere” (Foundations, Ak 430, p. 47). Kant does not say that only if some worth is absolute can there be any worth or value that is conditional. Nor does he say that rationality is the source or basis of all other (merely conditional) value. Indeed, he allows that if an inclination exists its object has worth or prudential value.

236

freedom and morality

Rather, what Kant says is that there can be no supreme practical principle (no way reason can be practical or motivating against all inclinations) if all worth is contingent (dependent on inclination). He is saying, that is, that there can be no categorical imperative, with a force independent of all inclination, if all worth is contingent upon inclination. On the other hand, if rationality has worth independent of inclinations, and if this rationality is constituted by submission to universal law, then we would have a motive over all prudential interests for conforming to the universal laws we will—the motive, namely, of respect for our own identity as rational beings. It is in the Critique that Kant makes explicitly clear that our rationality is, indeed, fully constituted by our submission to universal law (that is, to the categorical imperative as first formulated). Kant says, “Now this principle of morality on account of the universality of its legislation which makes it the formal supreme determining ground of the will regardless of any subjective differences among men is declared by reason to be a law for all rational beings in so far as they have a will; i.e., in so far as they are competent to determine their actions according to principles and thus to act according to practical a priori principles which alone have the necessity which reason demands in a principle” (Critique, Ak 32, p. 32, emphasis added). What it is to be rational, or to have a rational will, is exactly to submit to a priori practical principles that have necessity. Now, no principles that one makes for oneself as to one’s flourishing or happiness would be considered a priori or necessary for Kant. He clearly has in mind submission to the a priori principle expressed by the first formulation of the categorical imperative, so that in sum he is saying that to be rational is to submit one’s actions to universal laws (to what can be willed as universal law). Returning to the Foundations, Kant says that if we are to formulate a categorical imperative that incorporates a motive for conforming to the imperative, it will have to be an imperative restricting inclinations by “the conception of that which is necessarily an end for everyone because it is an end in itself ” (Foundations, Ak 429, p. 47, emphasis added). This is exactly the kind of remark that seems to call for filling in a gap in Kant’s argument from respecting my own rationality to respecting the rationality of others; perhaps filling it in by considerations of consistency, parity, impartiality, shared forms of life, etc. However, Kant says in defense of this simply that “every other rational being thinks of his existence by means of the same rational ground which holds also for myself ” (Foundations, Ak 429, p. 47, emphasis added). The point, I suggest, is quite simple. My rationality, for Kant, is exactly my submission to universal law (as in the first formulation of the categorical imperative).

Formulations of the Categorical Imperative

237

But the maxims that I can will as universal laws are exactly the same as any other human being can will as universal laws. Thus, for example, the fact that stealing to obtain goods cannot be willed a universal law without losing the secure use of goods (and so without contradicting the maxim) is true for every human being. Therefore, if I respect my own rationality, which is exactly to act in conformity with the categorical imperative as first formulated, then eo ipso I act in accord with the universal willing, and so with the rationality, of others. My actions, that is, do not violate what they can universally will, and so I respect their rationality. In other words, in respecting my own rationality, I automatically respect the rationality of every other human being, simply because the same universal willing that constitutes my rationality constitutes theirs. Both my existence and their existence is thought of “by means of the same rational ground” of conformity to universal law. Once it is understood that rationality is just submission to universal law, there is no gap between respecting my own rationality and respecting the rationality of others. The discussion so far shows that a person’s conformity to the categorical imperative, as first formulated, is equivalent to his respecting his rationality (and so his identity and his personhood). Thus there is a motive for conformity distinct from, and taking precedence over, all considerations of self-interest (satisfaction of inclinations). Further, respect for my own rationality, by such conformity to universal law, is identical to, and hence inseparable from, respect for the rational will of all other persons. It follows from all of this that it is possible to reformulate the categorical imperative in such a way as to include the motive for conforming to it, as Kant now does: “Act so that you treat humanity, whether in your own person or in that of another, always as an end and never as a means only” (Foundations, Ak 429, p. 47).8 This second formulation is exactly equivalent to the first formulation regarding the content of morality, viz., regarding what behavior is or isn’t commanded. Kant’s application of the second formulation to the case of promising is instructive in showing the equivalence of the formulations. If I make a promise that I don’t intend to keep, I use the other only as a means to my purpose: “For he whom I want to use for my purposes by means of such a promise cannot possibly assent to my mode of acting and cannot contain the end of this action in himself ” (Foun8. That the second formulation characterizes a motive for behaving in accord with the principle of the first formulation is clearly and well stated by Paul Guyer, “The Possibility of the Categorical Imperative,” in Kant’s Groundwork of the Metaphysics of Morals: Critical Essays, ed. Paul Guyer (Lanham, Md.: Rowman & Littlefield, 1998), 222.

238

freedom and morality

dations, Ak 430, p. 48, emphasis added). The inability of the other to contain the end of this action in himself is that he cannot will this maxim as a universal law, and hence cannot “contain it” in his rational being. It is not, per se, his inclination-based consent or acquiescence that is violated. Thus, even if on this occasion the person wouldn’t mind if I broke my promise (so that his inclination is to assent to it), still both he and I disrespect his rationality (I by breaking the promise, and he, afterward, by agreeing to it), since he cannot will as a universal law that people break promises when they think other people won’t mind.9 Hence the proscription against breaking promises is exactly the same as in the first formulation, viz., it cannot be willed as a universal law. The same exact equivalence applies to meritorious duties to others. Kant says that failing to contribute to the happiness of others or failing to further their ends is proscribed because “the ends of any person who is an end in himself, must as far as possible also be my end” (Foundations, Ak 427, p. 45). I contend that Kant’s point is not that another person’s having ends or goals they have devised regarding their happiness is part of what it is for that person to be an end in itself, so that those goals must be respected if I am to respect that person as an end in itself. What makes a person an end in itself for Kant is that the person makes consistency with willing universal law the principle of his will—not that he sets up or devises his own personal ends. That this is Kant’s point is made clear in the following passage: “Rational nature is distinguished from others in that it proposes an end to itself. This end would be the material of every good will. Since, however, in the idea of an absolutely good will without any limiting condition of the attainment of this or that end, every end to be effected must be completely abstracted (as any particular end would make each will only relatively good) the end here is not conceived as one to be effected. Now this end can never be other than the subject of all possible ends because this is at the same time the subject of a possible will which is absolutely good” (Foundations, Ak 438, p. 56, emphasis added). What makes rational nature an end in itself (or an absolutely good will) is that, independent of any end to be effected (any end it sets for itself in regard to its happiness), it sets up for itself a constraint on all ends in terms of consistency with universal willing (this being the possible will that is absolutely good). If this is 9. At least he cannot will it consistently with the maxim—I shall break a promise to achieve an end when another doesn’t mind. The universalization would mean that others would become tentative regarding his promising. His achieving the end of making a promise would be undercut, as others would first have to figure out whether he thinks they will mind or not. For the irrelevance of inclination-based consent, see Foundations, 48, n. 15 (Ak 430).

Formulations of the Categorical Imperative

239

correct, then for Kant I must foster others’ personal ends only because neither they nor I can will as a universal law the failure to contribute to, or to foster, the happiness of others. This is exactly the same consideration that matters in applying the first formulation to the case of meritorious duty to others. The difference is that it is now understood that the laws I can will are the same as any other rational being’s. Hence, in guiding my actions in terms of such universal willing, I am eo ipso respecting the laws (and so the rationality) of any other rational being. In sum, once we understand that in Kant’s conception rationality is being subject to principles or universal laws (as opposed to inclinations), and once we understand that the laws that can be universally willed are the same for all rational beings (the law either contradicting the interest in the maxim or not), then the equivalence of Kant’s second formulation to the first formulation (that they lead to the exact same restraints) is almost trivial. To respect another’s rationality (as per the second formulation) is to respect the universal laws they can will, which is the same as respecting the universal laws I can will (as per the first formulation). The difference in the formulations is that the second one embeds the motive for respecting humanity (personhood, identity, rationality) whether in myself or, equivalently, in others. Indeed, the second formulation is simply a re-characterization of the first (of what it is to will in accord with universal law), in terms of what rational willing is. In our understanding so far, the first formulation expresses a contractualist account of morality according to which an act is wrong if it violates universal agreement, where prudence or personal interest determines agreement (i.e., it either is or isn’t consistent with my self-interest for there to be certain universal laws). The second formulation expresses that the motive for keeping the agreement is not prudence or self-interest, but rationality. Thus, even if prudence tells me that I can get away with violating the agreement, I cannot get away with violating my own identity as a rational being—i.e., as a being who acts according to rational principle rather than mere inclination. The first formulation tells me that stealing to obtain goods is wrong because it is inconsistent with the prudential interest of obtaining goods that there be a universal law of such stealing. The second formulation tells me that even if I can steal and not be found out, so that my obtaining of goods in that situation is secure, I still violate my own rational nature by not conforming to the universal law. The view we are attributing to Kant in these two formulations, I suggest, is essentially Rousseau’s view in the Social Contract. For Rousseau the distinction between right and wrong is what conforms or not to the

240

freedom and morality

general will, which for Rousseau is unanimous or universal. Further, the basis of that universal will (of the unanimous agreement) for Rousseau is prudential interest. According to Rousseau the source of agreement to laws restraining behavior is the achievement of security. He says of the contract, “since there is no associate over whom he does not gain the same rights as others gain over him, each man recovers the equivalent of everything he loses, and in the bargain he acquires more power to preserve what he has.”10 Thus, for Rousseau, we agree to certain laws because it is in our interest that there be such laws. However, like Kant (but unlike Hobbes), Rousseau holds that private or personal interest is not the rationale of conforming to the laws one has agreed to. In chapter 8 of Book I Rousseau says that submission to the general will has “from a narrow stupid animal made a creature of intelligence and a man” in that such submission “makes reason rather than inclination the source of our actions.”11 The consequence of submission to the general will, then, is the rationality that constitutes our identity as human beings. Furthermore, for Rousseau the general will is indivisible; it is the same in me as it is in you.12 Hence respecting my rationality (in making conformity to the general will the principle of my behavior) is no different from my respecting your rationality (the general will in you). It is the same set of restraints (what can be willed unanimously) that constitute self-respect and also respect for the rationality of others. We see, then, that the second formulation, as well as the first formulation, evidences Kant’s own acknowledged debt to Rousseau. Our understanding supports Dieter Henrich’s claim that the categorical imperative was already formulated in the period (1765) in which Kant was influenced by Rousseau.13

4 . t h e t h i r d f o r m u l at i o n We turn now to the third formulation of the categorical imperative, which is that “all maxims are to be rejected that are not consistent with the universal lawgiving of will,” where the universal lawgiving of will signifies “the idea of every rational being as making universal law” (Foundations, Ak 431, p. 49). Whereas the second formulation turns on the idea that our rationality consists in submission of our will to universal 10. Jean-Jacques Rousseau, The Social Contract, trans. Maurice Cranston (Middlesex, Eng.: Penguin Books, 1968), 61. 11. Ibid., 64. 12. Ibid., 70. 13. Dieter Henrich, The Unity of Reason: Essays on Kant’s Philosophy, ed. Richard L. Velkley and trans. Jeffrey Edwards (Cambridge: Harvard University Press, 1994), 74, 83.

Formulations of the Categorical Imperative

241

law, the third formulation turns on the idea that our rationality is also the source of the universal laws to which it submits. The third formulation is equivalent to the first formulation. If I act on maxims that respect the lawgiving of other rational beings, then I act on maxims that I can will as universal law, since what I can thus will is exactly what other rational beings would make as law. Further, the third formulation is equivalent to the second formulation. A person is an end in himself exactly in making universal law (being rational). Therefore if I act on maxims in accord with rational will making universal law, I act on maxims that respect other people as ends (as rational beings). The third formulation again concerns the motive for conforming to universal law. As Kant says, he is here indicating in the imperative itself “the renunciation of all interests,” since “if we think of a will giving universal laws we find that a supreme legislating will cannot possibly depend on any interest for such a dependent will would itself need still another law which would restrict the interest of its self-love to the condition that (the maxim of the will) should be valid as universal law” (Foundations, Ak 432, p. 50).14 For example, if my motive for making a law is self-love or private interest, then I could make the law that no one except I should steal. However, if I am to make universal law, I cannot make this law, since I cannot universally will that people make themselves exceptions. The very idea, then, of legislating universal law goes against the idea of acting according to exceptions to the law based on self-interest. The third formulation, then, makes clear that in acting out of exceptions to universal law, I am not acting in accord with making universal law, and hence I am acting against my own legislative rational nature. In this way the third formulation expresses that the motive for conforming to the imperative is regard for my own rational legislative nature, not prudence or self-interest. Kant calls this third formulation of the categorical imperative “the principle of the autonomy of the will” (Foundations, Ak 433, p. 51) because it involves our being bound to laws that we ourselves make. These laws issue from our own will, indeed as constitutive of the rationality of our will, and so they require no other interest as a motive for obeying them. This idea of restraints on our behavior issuing from our own will is so important to Kant that he attributes the failure of all previous attempts to establish an airtight basis for morality to the failure to recognize this idea. “Man was seen to be bound to laws by his duty, but it was 14. Guyer again holds, likewise, that the third formulation continues the depiction of a motive for acting in accord with the first formulation. See “The Possibility of the Categorical Imperative,” 237.

242

freedom and morality

not seen that he is subject only to his own, yet universal, legislation, and that he is bound to act only in accordance with his own will” (Foundations, Ak 433, p. 51). Without this idea there is simply no motive for acting in accord with law that is unconditional or categorical (or that is not contingent upon private interests). Rousseau, likewise, sees this idea of rational nature giving law as fundamental to understanding how there can be legitimate, as opposed to forced, restraint on our behavior. He says, “How to find a form of association .l.l. under which each individual, while uniting himself with others, obeys no one but himself. This is the fundamental problem to which the social contract holds the solution” (emphasis added).15 His answer is essentially Kant’s—namely, that the laws of the social contract express our own wills (to be governed by rationality rather than inclination). He says, “no longer ask how we can be both free and subject to laws, for the laws are but registers of what we ourselves desire [as free rational beings].”16 Kant’s notion of an autonomous will as a will that is not subject to inclination, except as such inclination conforms to legislating universal law, follows Rousseau’s notion that in putting myself under the restraint of universal law I acquire “moral freedom which alone makes man master of himself, for to be governed by appetite alone is slavery, while obedience to a law one prescribes to oneself is freedom.”17 The essence of a contractualist theory is universal agreement on some basis despite the variation in, and even opposition of, private interests. I contend that Kant’s characterization of a realm of ends, which he derives from the third formulation, perfectly fits this characterization of a contractualist theory. He says, “If we abstract from the personal difference of rational beings and thus from all content of their private ends, we can think of a whole of all ends in systematic connection” (Foundations, Ak 433, p. 51), so that “there arises a systematic unity of rational beings through common objective laws” (Foundations, Ak 434, p. 52). The idea of rational beings all under a categorical imperative is the same as Rousseau’s conception of a people, so far as they are rational, forming a common general will by the social contract. Indeed, Kant’s depiction of a realm of ends is not far from Rousseau’s depiction of the social pact: “If, then, we eliminate everything that is not essential to it, we find it comes down to this ‘Each one of us puts into the community his person and all his powers under the supreme direction of 15. Rousseau, Social Contract, 60. 16. Ibid., 82. 17. Ibid., 65. For Kant’s equation of autonomy with freedom, see Foundations, 65 (Ak 447): “What else, then, can freedom of the will be, but autonomy; i.e., the property of the will to be a law to itself.”

Formulations of the Categorical Imperative

243

the general will; and as a body we incorporate every member as an indivisible part of the whole.’”18 For both Kant and Rousseau the harmony or unity of the association is that the very same universal laws are acceded to by everyone, so far as they are rational. Kant goes on to distinguish being a member of such a realm of ends (as being subject to the universal laws) from belonging to it as sovereign (as being the author of the universal laws), “when he, as legislating, is subject to the will of no other” (Foundations, Ak 434, p. 52). This again matches a distinction in Rousseau between “subjects in so far as they put themselves under the laws of the state” versus “citizens who share in the sovereign power”19 (who are authors or legislators of the general will). Rousseau’s contention that the general will “should spring from all and apply to all”20 is the same as Kant’s contention that the legislation that constitutes the realm of ends must be able to arise from the will of every rational being, and that the practical constraints of such a realm apply to each member in the same degree (Foundations, Ak 456, p. 75). A comparison of the passages we have just cited makes it hard to believe that Kant did not have Rousseau’s text in mind when describing the realm of ends. Kant’s third formulation, together with its attendant notion of a realm of ends, expresses the essence of a contractualist view—namely, that different wills are harmonized or brought under unanimous agreement. In the Critique Kant criticizes making the desire for happiness the supreme practical principle precisely because it doesn’t make wills harmonious. He says, “the wills of all do not have one and the same object, but each person has his own (his own welfare)” (Critique, Ak 28, p. 27). A supreme practical principle should harmonize all wills. Now, although the private welfare of one person can contradict that of another, the autonomy of one person (being a legislator of universal law) can never contradict the autonomy of any other (being a legislator of the same universal laws). As Kant puts it, “every will, even the private will of each person directed to himself, is restricted to the condition of agreement of rational being (his own and others) namely, that it should be directed to no purpose which would not be possible by a law which could issue from the will of the subject who is the passive recipient of the action” (Critique, Ak 88, p. 91). In summarizing the three formulations of the principle of morality (Foundations, Ak 436–37, pp. 54–55), Kant says that although they all formulate the same law, there is a difference in them that “is more sub18. Ibid., 61. 20. Ibid., 75.

19. Ibid., 62.

244

freedom and morality

jectively than objectively practical.” I contend that he means that the second and third formulations have more of a grip on us, since they explicitly incorporate the motivational ground of acting morally (of conforming to universal law). As Kant says, these formulations enable one to “gain a hearing for the moral law.” Kant says, however, that in moral evaluation, or in determining what is right or wrong, the “rigorous method” is to use the original formula. In our understanding, there is simply no bypassing the first formulation in evaluating whether an action is right or wrong, since the other two formulations contain nothing more than the first as regards the definition of what is right or wrong (as regards the content of morality). For example, if I try to judge whether an action such as stealing treats another as an end or only as a means (the second formulation), I cannot just say to myself that the other person wouldn’t like it. It is his rational assent alone that is at issue. But that assent, in turn, reduces to whether my action violates the universal laws he, as a rational agent, accedes to. But those are the very same laws that I, as a rational agent, can accede to. Hence what I have to do is to judge whether it is consistent with the inclination of the action (to obtain goods for use by stealing) to will the action as a universal law. The “objective” evaluation or determination, then, does not circumvent the first formulation. Similar remarks apply to the third formulation in relation to the first. Hence, the first formulation is the “objectively practical” (i.e., content-determining) component of all the formulations. The rest of the second section of the Foundations basically goes over how the view developed in this second section is consonant with “the common rational knowledge of morals” Kant had set out in the first section. It ends, finally, with the claim that autonomy of the will is the foundation of ordinary morality or of “the universally received concept of morals” (Foundations, Ak 438, p. 56). Once again this follows Rousseau, for whom the transition from natural (animal) to human freedom is the foundation of all legitimate restraint on inclinations. For both Kant and Rousseau what makes freedom or autonomy the foundation of morality is not that without it we cannot hold others responsible for their actions. Rather, without autonomy there is no prescriptive force to moral laws that can stand up against the incentives of inclination. It is the respect for (and satisfaction in) the dignity of our own humanity as autonomous authors of our own wills that overrides the pull of inclinations. In the Critique Kant puts this point as follows: “Thus we can understand how the consciousness of this capacity of a pure practical reason through a deed (virtue) can produce a consciousness of mastery over inclinations and thus independence from them

Formulations of the Categorical Imperative

245

[which brings forth] contentment whose source is contentment with one’s own person” (Critique, Ak 119, p. 125).

5. the heteronomy of happiness We have argued that for Kant the nature of morality is the submission of happiness (private ends) to universal law. Since such submission expresses our rational identity, there is a motive for morality so understood. Kant holds that our rational autonomy exists only in the submission of our wills to universal law, not in the setting out of ends or goals for our happiness. Happiness, for Kant, is consciousness of the “agreeableness of life” (Critique, Ak 22, p. 20). It pertains to “our whole existence and our satisfaction with it” (Critique, Ak 62, p. 64), where this satisfaction is determined by the “greatest possible and most lasting satisfaction of our inclinations” (Critique, Ak 147, p. 154). In other words, happiness is something like being pleased with our lives in regard to the overall satisfaction of our inclinations. Since different biological endowments and different circumstances lead to different inclinations differing in strength, happiness is not determined by the same set of activities or the same way of living for different people. Further, for Kant, no person has a “definite and certain concept” (Foundations, Ak 399, p. 15) of what will maximally satisfy their basic feelings of pleasure and displeasure. Hence Kant holds that there is no fixed or final concept of happiness. The role of reason in regard to a person’s happiness is to “think out for itself the plan of happiness and the means of attaining it” (Foundations, Ak 395, p. 11). Reason, then, does not only determine means but it sets ends as well, in the sense that it determines what activities to engage in or what way of living to aim for. But this setting of ends, for Kant, is always in the service of the maximal satisfaction of the inclinations I am endowed with. Kant believes that reason serves “only the purposes [of satisfying sensuous needs] which, among animals, are taken care of by instinct” (Critique, Ak 62, p. 64).21 Given this conception of happiness, and of the role of reason in guiding a person toward happiness, it is not surprising that Kant says, “The principle of one’s own happiness, however much reason and understanding may be used in it, contains no other determination for the will than those which belong to the lower faculty of desire” (Critique, Ak 24, p. 23). There is always the stock of inclinations (the lower faculty of desire) that reason is responsible to and conditioned by in setting ends 21. He even says (Foundations, 11 [Ak 393]) that nature would have done better to leave things to instinct. Kant underestimates the utility of versatile reasoning for organisms in complex and variable niches.

246

freedom and morality

for what sort of life to live. Hence such reason “as merely regulating the interest of the inclinations” is always “pathologically determined” (Critique, Ak 121, p. 127).22 Even though Kant’s account of happiness involves reason’s setting ends, it is not Korsgaard’s conception of rational autonomy as the source of our values. For Kant the ultimate source of our values for flourishing or for happiness is inclination. It is this that makes rationality about our happiness (about how to flourish in our lives) something heteronomous for Kant, since we have no real authorship over our own reasoning about happiness so long as it is basically determined in relation to inclinations which that reasoning doesn’t determine or create. We do not reason ourselves into the pleasant feeling that a warm bath provides. It is for this reason that Kant says “The inclinations themselves as sources of needs, however, are so lacking in absolute worth that the universal wish of every rational beings must indeed be to free himself from them [so far as this being’s rational nature is concerned]” (Foundations, Ak 428, p. 46). It is also for this reason that Kant can say, “But still he is not so completely animal as to be indifferent to everything that reason says on its own and to use it merely as a tool for satisfying his needs as a sensuous being. That he has reason does not in the least raise him above mere animality if reason serves only the purpose which, among animals, are taken care of by instinct” (Critique, Ak 446, p. 64). In his conception of the highest good Kant does allow that a rational being wants to be happy in accord with virtue or moral uprightness. This implies that a rational being with inclinations values the satisfaction of inclinations that is consonant with morality. It does not imply, however, that his rationality is the source of the value of his inclinations. The “goodness” of the satisfaction of his inclinations per se is in his animality, not his rationality. The only role of reason above animality is to “de-value” the satisfaction of inclinations that conflict with the moral law. For Kant, in the sheer seeking of happiness we are slaves to inclination, not masters of our ends. I believe that the passages I have cited show that Kant goes out of his way to deny the conception of autonomous goal setting that Korsgaard attributes to him on the basis of the second and third formulations of the categorical imperative. In our interpretation it is clear, indeed, why Kant must hold that happiness is heteronomous. Figuring out our personal happiness can conflict with conformity to universal law. If the figuring itself were autonomously rational, then I would not lose my au22. Reason in regard to happiness is pathological for Kant, even if what is agreeable to us (what satisfies our inclinations) are things like knowledge, art, or friendship, as opposed to food, sex, and titillation.

Formulations of the Categorical Imperative

247

tonomy by following the dictates of deliberation about my happiness when it goes against acting in accord with universal law. Indeed, in such cases I would be at a standstill, having to lose autonomy either way I acted (whether against my own personal happiness or against universal law). I should probably conclude that at least by following the dictates of deliberation on my own happiness I don’t sacrifice my own happiness along with a portion of my autonomous rationality. In sum, the autonomous rationality of conformity to universal law would no longer serve as a motive for overriding conflicting deliberations about my own happiness. The heteronomy of deliberations regarding happiness, then, is utterly central to the defense of morality (conformity to universal law) as against the dictates of happiness. Far from the second and third formulations of the categorical imperative expressing that personal goals and ends derive from autonomous rational nature, the very significance of these formulations demands that rationality about personal ends or goals be heteronomous.

concluding remarks We have argued that for Kant rationality is exactly a matter of submission to universal law. The only way that autonomous rationality is involved in our deliberations on our personal happiness is as a negative restraint, limiting that deliberation by conformity to universal law. In particular, for Kant, there is no positive role for rationality of being an autonomous source of ends or goals for flourishing in life. The basis of this contention of ours is, first, that it makes the equivalence of the three formulations of the categorical imperative straightforward (to respect someone else as an end in himself and as a lawgiver is exactly to respect what I can universally will). Second, it ensures that there is no gap in the second formulation between respecting my own rationality and respecting the rationality of others (it is respect for the exact same universal laws that can be willed). Third, it explains why Kant thinks the first formulation is the “rigorous method” for determining right and wrong (the second and third formulations don’t replace the conception of universal law—they simply add motivational status in terms of rationality). Fourth, it makes clear Kant’s own acknowledged debt to Rousseau, since Rousseau likewise holds right and wrong to be determined by the general (universal) will, and the motive of morality to be that conformity to the general will makes us human, free, and rational. Fifth, it explains why in passage after passage Kant goes out of his way to deny that devising personal ends for happiness is an expression of rational autonomy. The heteronomy of personal ends ensures the “advan-

248

freedom and morality

tage” of morality (autonomous submission to universal law) over all personal ends. Although I believe I have presented strong evidence that Kant’s moral theory is a contractualist one that denies autonomy to deliberation about personal ends, I don’t believe that Kant is right about the supposed heteronomy (even “pathology”) of happiness. Although it is a fairly subtle issue going beyond the scope of this paper, I believe that Korsgaard is correct against Kant in holding that there is an autonomously rational component to acting from personal goals.23 Kant himself cannot defend the heteronomy of setting personal goals by some sort of generic naturalism, according to which what appears to be autonomous deliberation is, and can be, nothing more than maximizing inclinations. The reason is that Kant himself denies such a naturalism in the case of acting from conformity to universal law. In this case Kant allows that inclinations or feelings are determined by reason, as opposed to reason being driven by inclination. If in the case of conformity to law the feeling or the inclination can be determined by the independent recognition of rationality, then Kant has no grounds for excluding the same idea in regard to the setting of personal ends. We can understand happiness “nonpathologically,” then, as a feeling of satisfaction with our lives based on independent rational reflection upon our lives (not upon other inclinations).24 Suppose, contra Kant, that deliberation over personal ends is a case of rational autonomy. Then the problem for Kant’s view would be that a conflict of such personal ends with morality could no longer be resolved by morality’s expressing rational autonomy. This, however, is also a problem for a view such as Korsgaard’s that makes respect for others’ rationality in setting goals the ground of morality. If the particular reasoning that leads us to our ends or goals seems correct or without flaws, but also conflicts with respect for the rationality of others, then we are at a stalemate where there is no good reason for preferring respect for others (morality) over our own personal ends. The only solution, I believe, is to show that there cannot be such a conflict—i.e., to show that the very cogency of our specific reasoning about which goals or ends to adopt automatically entails morality (respect for other people). This is what is required to rationally justify morality if we give up Kant’s doctrine of the heteronomy of happiness. 23. See Korsgaard, Creating the Kingdom of Ends, 114ff. 24. This conception is not so far from Kant’s own conception of aesthetic pleasure in the third Critique. Note that in this conception of happiness no satisfaction of inclination (no pleasure) itself is valuable except as reflection makes it so.

13

The Rational Justification of Morality

1. introduction Kant attempted to derive morality from our nature as rational beings, rationality signifying for Kant universality, which was the form of moral law. Korsgaard has modified this Kantian program by making rationality the source of our goals and morality a matter of respecting or valuing our goals by respecting our rational nature as such.1 Supposedly a derivation from our rational nature would explain why it is that morality matters or that we should be moral. Even if it might promote the satisfaction of our goals to behave contrary to morality, it would be irrational to do so. If one is asking for a reason to be moral, the fact that it is unreasonable not to be is pertinent. However, whether rationality is in this way a conclusive or dominating reason to behave morally as against my goals is not obvious. For Kant, rationality involves universality of will, and what I violate by pursuing my goals against the moral law is just universality of will. It would seem that I have two considerations to weigh—achieving my goals vs. being rational in the sense of conforming to universal will. It is not clear why the latter should dominate the former. For Korsgaard, more subtly, rationality is the source of my goals. What I violate by pursuing my goals against morality is respect for the very source of my goals themselves. This still leaves it open that I think these specific goals are the best ones for me to have or the most conducive to my happiness. Hence there are still two considerations to weigh—achieving my happiness vs. being rational in the sense of respecting the faculty or capacity that is the source of my goals. If there is a conflict, the latter supposedly overrides the former, since if I cannot respect the reasoning capacity that is the source of my goals, then I cannot respect the reasoning that leads me to think that my goals are the ones conducive to my flourishing. This may be so, but it does not imply 1. Christine Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996). See also Alan Gewirth, Reason and Morality (Chicago: University of Chicago Press, 1978).

249

250

freedom and morality

that the specific reasoning to these goals is bogus or faulty. I am left, then, with a certain standoff—either I disrespect my reasoning capacity as to its being an inherently public or common one, or else I disrespect the upshot of my otherwise unimpugned reasoning that certain goals will issue in my happiness. Rationality will automatically override or dominate achieving my goals only if it pertains to reasoning about the viability of the specific goals themselves. The rational assessment of goals in regard to their promoting my happiness directly dominates rationality in achieving my goals, because the worth of achieving goals depends directly on the worth of the goals to be achieved. If morality is to automatically override the achievement of my goals when these conflict, then morality must be derived from being rational as to having these specific goals in the first place. I am not contending that universality of will or respect for public rational nature are not parts of being rational. I am contending rather that those aspects of rationality must enter directly into the nature of being rational about which goals to have to be happy, because it is only rationality about the cogency of our goals themselves that automatically dominates achieving of our goals. In what follows I try to show how morality can be grounded in being rational in the very procedure of devising and assessing goals as conducive to happiness. We shall set out a definition of happiness and an axiom regarding it, from which it will follow that the reasonable pursuit of happiness requires morality.

2. the reasonable pursuit of happiness To begin with, happiness is connected to being satisfied or content with one’s life, or with how one’s life is going. Equally it has to do with a positive evaluation of one’s life or an appreciation for how one’s life is. Happiness, that is, involves a positive attitude toward one’s life. This is compatible with one’s being happy in some ways and not in others, and with happiness coming in varying degrees. Now, I suggest that although contentment with one’s life is an essential component of happiness, it is not by itself happiness. A person may be content or satisfied with his or her life because the person’s outlook is so limited that the person cannot even imagine alternatives. The contentment, that is, is not based on informed reflection on one’s life. Such a person, we might say, believes he or she is happy but really isn’t. Any connection of the notion of a happy life to the notion of a worthwhile life, I believe, requires this distinction between merely believing that one is happy versus really being happy. Happiness, then is not sheer contentment but rather “jus-

Rational Justification of Morality

251

tified” contentment. One is happy, that is, if one’s satisfaction with one’s life is justifiable, supportable, defensible, etc. Equivalently, one’s positive evaluation of one’s life must be informed or reasonably made. One’s appreciation must derive from some suitable understanding of how one could have lived differently, what the options for human living are, and so on. In a certain sense the issue of happiness requires a stance toward our lives similar to the stance that a critic might take toward a work of art. There is a difference between simply immediately just liking a painting versus appreciating it in an informed and considered way. Similarly, there is a difference between merely liking one’s life and happiness, the latter being a suitable or informed or considered appreciation of it. We can sum this up by saying that being happy is being justifiably or suitably content with one’s life. So far we have a somewhat vague and merely formal definition of happiness; i.e., there is no indication as yet as to which activities or pursuits make a person happy or contribute to a person’s being happy. I am going to take it as an axiom that there is no definitive or conclusive justification in regard to being content with one’s life that it include any fixed list of pursuits or activities. Equivalently, there isn’t any provable or unrevisable basis for positively evaluating one’s life for containing any specific stock of achievements, goals realized, etc. For example, no one can prove or definitively establish that being healthy is something for a person to be content about. It might be that there is justification for being satisfied with having good health, but, according to the axiom, there is no proof or definitive justification for being so satisfied. Again, an analogy to aesthetic appreciation is helpful. Although one’s appreciation of the Mona Lisa may be informed or suitable, there are no “material” elements (aspects of the painting) that “prove” or conclusively justify a positive assessment. The defense of our axiom is simply the implausibility of its denial. There always seem to be further considerations or reflections that are pertinent both to weakening and to strengthening one’s positive evaluation of one’s life for including any particular achievements, states of being, or activities. There is the consideration, for example, of how much one has to give up to be healthy, how much enjoyment one may have passed over, how much danger and excitement one may have missed out on, etc. The justification is never a closed, unrevisable affair, because there always remain pertinent considerations (both pro and con) to keep the issue open-ended. So long as there are always pertinent reflections remaining, there is no definitive justification. Similarly, the appreciation of Hamlet or the Mona Lisa is open-ended. There are

252

freedom and morality

always further pertinent reflections that can strengthen or weaken our appreciation. We can sum up our axiom as follows: People do not have knowledge (conclusive justification) of what constitutes their own happiness. Armed with our definition and our axiom, we can turn now to what constitutes reasonableness in devising and assessing our goals. If there were a definitive material conception of one’s happiness, then reasonableness about one’s goals would be a matter of coming to that conception or getting it right about what one’s happiness is. I believe that in Plato’s view being reasonable meant having the right opinion of the truth of what one’s happiness is. Since by our axiom there is no conclusive truth, being reasonable is not a matter of true opinion but rather of one’s conception being well considered and well informed. Since there is always further potentially relevant consideration and information, the reasonableness of one’s conception involves being open to further consideration and information. Consider a scientific researcher who has no knowledge of a cure for cancer. It would be unreasonable of him to close himself off from other researchers’ thoughts or ideas. In particular it would be unreasonable or incompatible with his having a well-considered opinion for him to prevent or interfere with other researchers’ devising and developing their own thoughts or ideas. He would cut himself off from those thoughts or considerations that are potentially relevant to his own thinking about the issue. As one case, his interfering with someone else’s thinking on the matter by feeding that person false information or by sabotaging that person’s application for a research grant might prevent that person from coming up with objections that would show the faultiness of his own research. The interference thus makes his own pursuit of the issue unreasonable. It would likewise be unreasonable for him to interfere with other researchers’ testing out their thoughts and ideas, since he himself is then cut off from the information gotten from that testing, information that might be relevant to his own thinking on the issue. For example, his sabotaging someone else’s experiments might close off information that would show that his own theory needs revision. This sort of interference with testing out ideas makes his own pursuit of the issue unreasonable. Let us apply this result to the subject of one’s own happiness. Here again, since I have no knowledge, it is unreasonable for me to interfere with others’ devising and developing their own thoughts and ideas about what their happiness is, for those thoughts and ideas are potentially pertinent to my own thinking. One’s preventing or interfering with other people’s devising and considering their own goals cuts one off as well from those considerations. For example, if I brainwash some-

Rational Justification of Morality

253

one to think only of living in a way that I find suitable, I close myself off from his having thoughts about how to live that might expose the flaws in my own conception. Likewise, it would be unreasonable for me to prevent or interfere with other people’s pursuing or living out and so testing their goals. Abstract consideration of the pros and cons of various goals goes only so far, since not everything about the cogency of goals can be anticipated by sheer consideration. The assessment of goals also requires information that is gotten only by seeing how they pan out in practice. In preventing or interfering with other people’s acting on their goals I interfere with the testing out of their goals in practice, and so again cut myself off from potentially pertinent information. For example, if fame is one of my goals, how it has worked for others who have pursued it and how they think about it throughout their actual pursuit is relevant information. In sum, being reasonable about the topic of my happiness requires not interfering with others’ coming up with and considering their own goals and, as well, not interfering with their testing out those considerations by acting in pursuit of their goals. In the scientific research case, being open to considerations and information generated by others, and thus coming to a considered and well-informed opinion, may result in finally reaching definitive knowledge. When the topic is my happiness, our axiom implies that there is no such definitive knowledge to reach. In this case, being reasonable in the sense of coming to considered and well-informed opinions is not a means to reaching definitive knowledge. Nevertheless, such reasonableness is internally related to happiness. The more informed and considered one is about one’s goals, the more one can have informed support for contentment upon reflecting on one’s life of pursuing or achieving those goals. The same reflection that goes into devising and supporting goals is available for the justification for being content with living a life in terms of those goals. The goal is not knowledge but richness of appreciation (of one’s life), and being open to considerations and information is exactly being open to what can enrich our reflection on our lives. One may, of course, reasonably pursue happiness and fail to be content with one’s life and so fail to be happy. However, without the reasonable pursuit, one is assuredly not going to be suitably content. Hence the reasonable pursuit is necessary for happiness (for suitable or justifiable contentment). As an analogy, consider the case of being reasonable about how to understand or interpret Shakespeare’s Hamlet for stage production. In this case, as in the case of happiness and unlike the case of scientific research, there is no definitive or conclusive knowledge or understanding

254

freedom and morality

to be had; viz., there is no such thing as the true interpretation. Once again, a director is reasonable in staging a production to the extent that his thinking on the matter is well considered and well informed, and once again this involves being open to how others think about the matter and how the thinking of others has panned out in actual productions. In this case, again as in the case of happiness, such well-considered and well-informed opinion is not a means toward progressing to some true or perfect production, but nevertheless it is pertinent to the goodness of the production exactly because that goodness is the reflective satisfaction with the production on the part of the director and the audience. The less well considered and well informed the thinking that goes into the production, the more impoverished the nature of the satisfaction upon reflection or the less suitable the appreciation.

3. the principle of noninterference Let us say that a person is acting out of a reasonable pursuit of happiness if the person’s actions are consonant with his conception of happiness and his conception is well considered and well informed; i.e., he does not close his conception off from consideration and information. We can say as well that such a person is acting out of values that are open to reason. Such a person’s actions are guided by his reasonable conception of happiness or what he values in his life in the sense that if he thought that what he was doing contradicted that reasonable conception, he wouldn’t do it. Thus, in particular, such a person is not overcome by desires that conflict with his values and is not irretrievably fixed to these values so that they are closed to further consideration or information. Such a person need not at all be constantly reconsidering his life and may be deeply committed to a certain way of living. It is just that he retains the capacity to rethink his life and does not close himself off from consideration and information. A director may be deeply committed to certain ideas in staging Hamlet without being closedminded or fanatical. Nor need such a person be constantly gathering or amassing considerations and information. Indeed, such amassing, to the extent that it undercuts his living his life, is inconsistent with his reasonably pursuing happiness. The point is to come to well-considered, well-informed decisions as to how to proceed in life, not to theoretically amass considerations and information. I suggest that people for the most part in their daily lives conform to this characterization of acting out of a reasonable pursuit of happiness. Our result so far may be stated as follows: It is always unreasonable for anyone to interfere with other people’s acting out of a reasonable

Rational Justification of Morality

255

pursuit of happiness. Equivalently we may say that it is always unreasonable for anyone to interfere with others’ acting out of values open to reason. Such interference can take two forms. One may interfere with another’s devising or assessing values or goals, or one may interfere with another’s acting out of those values or goals. In the first case one is interfering with another person’s considerations regarding his life, which is unreasonable in closing oneself off from those potentially relevant considerations. In the second case one is interfering with another person’s pursuing his goals or testing out his values, which is unreasonable in closing oneself off from potentially pertinent information as to how certain values pan out in certain circumstances. Thus interfering with someone else’s reasonable pursuit of happiness is closing oneself off from considerations and information, which is incompatible with one’s having a well-considered and well-informed conception of one’s own happiness. We note that interference with the reasonable pursuit of happiness is always unreasonable, no matter what one’s goals. In the scientific research case, if one’s goal is not just to come up with the answer but to win a Nobel prize, then it might not be unreasonable to interfere with other researchers by feeding them false information or sabotaging their experiments. Indeed, the pure goal of knowledge in this case is optional, and hence my unreasonableness in interfering with others’ pursuit of knowledge is conditional upon my adopting the goal of reaching knowledge per se. In the case of one’s own happiness, however, the goal is not likewise optional. In Kant’s terms happiness or satisfaction with one’s life is an apodictic end,2 since happiness is just a name for being suitably satisfied with pursuing and achieving one’s goals. In this regard it is flat out unreasonable or categorically unreasonable to interfere with another’s reasonable pursuit of happiness, no matter what my personal goals are. The scope of the unreasonableness of interfering extends to others who to my way of thinking are pursuing silly or unlikely goals. In devising our goals, considerations of culture, society, and biological nature are important. Still, in accord with our axiom, no one has definitive opinion that these are predominant or overriding components of happiness. Nonconformists, loners, ascetics, and others live lives that are full of potentially pertinent information, and their own considerations of their lives are potentially pertinent considerations. This is just to say that neither culture nor biology is a source of absolute value or a defin2. Immanuel Kant, Foundations of the Metaphysics of Morals, trans. Lewis White Beck (New York: Liberal Arts Press, 1959).

256

freedom and morality

itive component of happiness. Thus if someone’s goal in life is to collect jelly beans rather than to have a family or develop her natural talents, there are all sorts of considerations that make such a life pertinent to my own reasonable pursuit of happiness. At the very least a mirror is held up to me that may give me some larger perspective from which I see that my own culturally and biologically based goals may lack a sensitivity to the absurdity of things, or may lack imagination. Far-out or highly nonconformist ways of living play a role similar to the role of farout productions of Hamlet. Perhaps there is not enough in them for me to adopt that way of staging, but there is enough in them to color my production. Our cultural and biological milieu, I suggest, plays a heuristic role similar to a principle of conservatism in scientific research. This principle, which perhaps would have favored Lorentz over Einstein, is not so strong as to override open-mindedness. Likewise the culturally or biologically strange or frivolous, though violating a certain heuristic, is not to be precluded or interfered with. Further, the fact that at some level another person is pursuing a kind of life that has been tried many times by others doesn’t imply that there is nothing to learn from his life. The circumstances and natural endowments of people are varied enough for each new person to be a twist on even a very common theme. Thus neither the radicalism nor the conservatism of a person’s goals makes it reasonable to interfere with his considerations or actions. There is one sort of value or pursuit of happiness, however, that is beyond the scope of our contention that interference is unreasonable—namely, the person whose goals themselves are unreasonable. If a person’s conception of happiness involves enslaving or tyrannizing women, for example, then that conception, in interfering with a woman’s pursuit of happiness, is definitively unreasonable since it involves closing off considerations and information from that woman. The fact that our axiom contends that no one has definitive knowledge of what his happiness is, is compatible with the fact that we have definitive knowledge that being closed-minded by interfering with the woman’s life is unreasonable. One might object that this person with his goals is a source of information about what it is like to lead a life that includes lording over a woman. Maybe this leads to feelings of power and well-being. Therefore, interfering with this person would close off such information and so would indeed be unreasonable. Recall, however, that true happiness is a matter of satisfaction or positive evaluation consequent on well-considered, well-informed reflection upon one’s life; not a matter per se of feelings or of what it is like. Although the tyrant’s life might provide information about what it is like

Rational Justification of Morality

257

to live a life of tyrannizing a woman, it provides no information at all about what might constitute well-considered, well-informed reflection upon such a life. We know that reflection upon such a life is reflection that has closed off considerations and information. Hence any positive assessment consequent to that reflection is impoverished or stilted and so only a “false” happiness. I may perfectly well grant “positive” feelings of power and well-being to a life of tyrannizing others. Still, I cannot have satisfaction consequent to reflection upon achieving those feelings that way, and so those feelings achieved that way cannot be part of my true happiness. Certain goals, then, automatically are and are known to be incompatible with reflective satisfaction and hence incompatible with real happiness. If I interfere with someone’s pursuing or even devising those goals, I would not be losing considerations or information relevant to or compatible with my happiness. For this reason it is not unreasonable in my pursuit of happiness to interfere with a person whose goals are themselves unreasonable for already interfering with still others’ reasonable pursuit of happiness. Our contention, then, is not that it is unreasonable to interfere with other people’s pursuit of happiness, but only with other people’s reasonable pursuit of happiness. Not only is it unreasonable to interfere with the reasonable pursuit of happiness of others, but it is also unreasonable to fail to promote it in certain respects. Consider again the pursuit of a cure for cancer. Not only is it unreasonable to prevent another person’s considerations, but in certain circumstances it is also unreasonable to fail to help him along by providing him with one’s own considerations. Not only is it unreasonable to sabotage another’s experiments, but in certain circumstances it is also unreasonable to fail to promote a research grant for him to carry those experiments out. The basis of the unreasonableness is exactly the same. Failing to promote the reasonable pursuit of a cure by others closes oneself off to potentially relevant considerations and information, and where one has no knowledge, this is unreasonable. Similarly, then, since I have no definitive knowledge of my happiness, it is unreasonable in certain circumstances to close myself off from considerations and information pertinent to that issue by failing to promote the consideration or devising of goals by others or failing to promote the acting out of goals by others. Although the basis or rationale of promoting the reasonable pursuit of happiness by others is the same as the basis or rationale of not interfering with it, the exact implications for behavior, we shall see, are much less clear.3 3. See the end of section 5 below.

258

freedom and morality

4 . a c at e g o r i c a l i m p e r at i v e Our principle of noninterference4 is a principle guiding action or behavior toward others and indeed is a categorical principle in the sense that it is not conditional upon one’s achieving one’s goals. Whether or not interfering with another leads to the promotion of one’s personal goals, it is unreasonable to do so. If we formulate the principle as an imperative—“Don’t interfere with someone else’s rational pursuit of happiness”—then what we have in Kant’s terms is a categorical imperative, as the force or basis of the imperative is not hypothetical upon what will satisfy one’s goals or what one’s goals are for their happiness. Further, this principle has a force overriding prudence or the achieving of one’s goals. Indeed, the supposed rewards of violating the principle by interfering are rewards only to the extent that they fit into a well-considered and well-informed conception of what counts as a reward. Interfering, in blocking such a conception, divests the supposed reward of the status of something for me to reasonably pursue to be happy. It is not as if the reward stands on its own independently as a factor competing with noninterference and so having to be weighed against it, for the noninterference is the necessary condition for any cogency in thinking of the reward as a reward (as something that meshes with a well-considered and well-informed conception of how to live). This categorical and predominating5 nature of our principle of behavior in relation to the satisfaction of our personal goals means that the principle can serve formally as a principle of morality, since one formal feature of morality is precisely that it may conflict with prudence but is overriding when there is a conflict. Our principle of behavior is also categorical with respect to whether or not one can get away with violating it. Even if one isn’t found out, one still loses considerations and information by interfering. By analogy, again, sabotaging another cancer researcher’s project secretly is no less unreasonable than doing it publicly, as far as developing and coming to reasoned opinions is concerned. Once again this makes our principle formally suitable to be a principle of morality, since right and wrong should not be determined by whether or not one is found out. 4. The discussion of the present section applies as well to the principle of the promotion of the reasonable pursuit of happiness of others. 5. Kant emphasizes the categorical nature of morality. The overriding or predominant nature is set out by Plato in Book 2 of The Republic, when he has Glaucon say that if morality is to be proved as something other than what is good for its consequences or good for avoiding harm by others by merely appearing to be moral, then we must show that a life of perfect justice with dire consequences is still better than a life of injustice with “good” consequences.

Rational Justification of Morality

259

A third formal aspect of morality is that it involves a deference to or respect for other people. Indeed, one aspect that makes morality problematic is exactly that at times it puts deference to other people over and above our own satisfaction of our goals. In this respect morality stands against selfishness, if selfishness is defined as exclusive concern for the satisfaction of one’s goals. Why should other people, even other people whose happiness emotionally is not connected to mine, matter more than my satisfying my goals? In our view respect or deference is due to other people because my reasonable pursuit of happiness is inseparable from theirs. To interfere with someone else’s reasonable pursuit of happiness is automatically to interfere with my own reasonable (well-considered and well-informed) pursuit of happiness. One might object that the “value” or import of other peoples’ lives (their reasonable pursuit of happiness) is as a means to our own lives, which accords no intrinsic value to others. Morality then becomes a more sophisticated version of selfishness,6 where other people have only instrumental value or worth. This objection fails to emphasize the dependence of my life on other people. I, as against other people, cannot be a reasonable pursuer of happiness and so in this regard the deepest nature of my life is inseparable from others. If selfishness involves an antagonistic relationship of putting myself above other people, then our principle of deference or noninterference is better called self-regarding than selfish, because it recognizes that my own existence as a reasonable pursuer of happiness is internally connected to the similar existence of others. There is no separable self as reasonable pursuer of happiness to be selfish about. My reasonable pursuit of happiness is internally bound up with others and so my deference to or regard for others is as deep as my deference to or regard for myself. In sum, the objection notes that in our view others have value in relation to my reasonable pursuit of happiness, but fails to note that my reasonable pursuit itself has no existence apart from others. Equivalently, the value of others in our view is not merely instrumental. If others were means to my reasonable pursuit where that pursuit was definable independently, then indeed others would be mere means. The pursuit of happiness by others has a status closer to being a component of, rather than a means to, my reasonable pursuit, since my reasonable pursuit involves respect for others for its very existence. In this regard the respect for others is no mere extrinsic means to my reasonable pursuit of happiness. A fourth aspect of morality is that it has a certain breadth of scope 6. Tantamount in this way to Hobbes’s theory.

260

freedom and morality

that goes beyond special relationships such as love, friendship, kinship, etc. The distinction between right and wrong extends to people we dislike, to strangers, and to foreigners. In a certain sense this is a principle of parity or equality. Whether or not we are fond of a person, whether or not that person shares our religious and cultural values, whether or not that person is of our race or gender, he or she is due our deference or noninterference simply as a reasonable pursuer of happiness. If we say that someone’s humanity is that part of him or that capacity in him that is unique to human beings, and if we hold that being a reasonable author of goals for life is uniquely human, then our principle of noninterference is a principle of deference to the humanity of others, independent of special relationships in which they may stand to us. Our principle implies that I cannot detract from or fail to respect the humanity of others without at the same time detracting from or failing to respect my own humanity. If we say that someone’s freedom or authority over her own life is her acting out of a reasonable pursuit of happiness,7 then our principle of noninterference is a principle of deference to the freedom of others, implying that I cannot detract from or fail to respect the freedom of others without detracting from or failing to respect my own freedom. Our principle, then, has a scope or force that goes beyond special relationships among people, being based on people’s humanity and freedom, not their kinship. This makes our principle at least potentially, if not actually, universalist in the following sense—that our principle applies as widely as behavior can influence others’ lives and as far as others’ lives can be communicated to us. That an ethical view in fact extends only to beings whom my behavior can influence is not surprising. There are no prohibitions or permissions where there is no possibility of influence. There may have been times when the scope of morality was only one’s group or proximate groups, for the sufficient reason that the nature of transportation made any interaction with remote groups impossible. In our view people who are so remote that all communication of their lives to us is impossible are not in fact sources of potential information or considerations to us and so likewise are beyond the scope of morality. Once again there may have been times when the scope of morality was limited thus by the nature of communication. Nowadays, with almost universal transportation and communication, the scope of our principle approaches true universality. Our principle, 7. Acting from a reasonable pursuit of happiness contrasts with being overcome with desire, with others being in control of our actions, etc. Because it thus contrasts with the various ways of failing to be free, it is plausible to identify such acting with being free.

Rational Justification of Morality

261

then, is “conditionally” universalist—it extends to all people per se on the condition that we can, however indirectly, act upon them and hear from them. Although this isn’t quite a Kantian universality that extends to all rational natures whether we can be in interaction with them or not, nevertheless, since the scope is not restricted by kinship or culture per se, our principle does have potentially universal import. In sum, our principle of noninterference has the formal features of being categorical (independent of advantages achieved by violating it and independent of whether or not one gets caught violating it), of involving deference to other people (respect for their lives, their humanity, their freedom), and of being universal (extending beyond kinship, cultural affinity, etc.). These features, I suggest, are sufficient to characterize our principle as a principle of morality. Roughly, a person has specifically moral ideals if these ideals are categorical in their lives, if they involve deference to other people, and if they are not specifically limited in range to kinship or group. Whatever the exact content of a person’s ideals (even if they are hideous ones), if they satisfy these formal features they are moral ideals. Not only is our principle of noninterference a principle of morality, but our argument for it establishes that any morality that contradicts it is unreasonable. Moral ideals that contradict our principle license at least some behaviors that violate the principle and so license flat-out unreasonable behavior. This makes them unreasonable ideals. The justification of our principle, then, is a justification of a common core of all “true” (i.e., reasonable) moral ideals.

5 . o r d i n a ry m o r a l i t y In this section we outline what actions are consistent or inconsistent with our principles of noninterference and promotion of the reasonable pursuit of happiness by others. If we define the difference between right (permissible) and wrong (impermissible) in terms of whether an action is consistent or not with our principles, then we get the result that doing wrong is flat-out unreasonable. We shall proceed by considering what I believe is the common core of “ordinary” morality and try to establish that this core coincides with our principles. What is ordinarily reckoned as wrong at the core is exactly what is inconsistent with our principles. If so, then following Rawls our principles systematize the core of ordinary morality, and besides they justify that core as necessary to any reasonable morality.8 We shall first consider the principle 8. See John Rawls, A Theory of Justice (Cambridge: Harvard University Press, 1989).

262

freedom and morality

of noninterference by itself and then consider how the principle of promotion of the happiness of others figures in. To take someone’s life for the sake of personal advantage (such as money, fame, peace of mind, etc.) is to interfere, since it clearly blocks any further considerations of goals or acting out of goals on the part of the person. Such actions, which are clearly wrong on ordinary reckoning, are equally clearly violations of our principle. By contrast, killing somebody in self-defense is not ordinarily wrong in the sense that it would ordinarily be thought to be permissible. It is true that in killing in self-defense I interfere with a person’s pursuit of happiness. I claim, however, that I do not interfere with that person’s reasonable pursuit of happiness, since his coming to kill me is flat-out unreasonable by his acting in a way that closes off considerations and information. Thus this action is not a violation of our principle. The same pertains to the case of my killing a second party to save the life of an innocent third party. One may object that although such killing doesn’t violate our principle it is still as unreasonable as not saving the third party’s life, because I lose information and considerations on the killer’s part. Note, however, that the killer is going down a road that is known to be flat-out unreasonable (closing off information and considerations), in line with the case discussed above of tyrannizing a woman. As in that case, there is no question of my reasonably considering the merits of such a life that includes killing, since I cannot ever subsequently be reflectively satisfied in any broad or clear way with having adopted that life. I am choosing, then, between losing no information that can be incorporated into happiness vs. losing information (from the third party, who isn’t unreasonable) that may well be incorporated into happiness. It follows that it is reasonable to protect considerations and information from reasonable people against those who are acting unreasonably. The same sort of reasoning that applies to the taking of life applies also to the infliction of bodily harm and more generally to the violation of a person’s own use of their body, as in kidnapping, false imprisonment, rape, etc. To interfere with the workings of someone’s body by maiming it or restraining it is clearly to interfere with his acting out of values that are open to reason, since the body is essentially involved in thus acting. Once again, the particular imperative not to inflict harm, although categorical, is not without exceptions. To maim a person in order to stop that person from killing another is neither a violation of our principle nor wrong in ordinary thinking. In our ordinary moral thinking, lying to someone because it puffs us up or because we stand to gain position, love, power, etc. is to do something that is clearly morally wrong. Now, a person’s acting out of values

Rational Justification of Morality

263

or out of a conception of happiness requires a basis in factual information for his planning or his considering his ends and the means toward them. As an extreme example, a person who believes that if he practices hard enough he can train his arms to be wings that will enable him to fly, and makes this one of his goals in life, is not acting out of values open to reason because of the falsity in the basis of his thinking. If I lie to someone, then via the false information I potentially interfere with the reasonableness of his planning or his authorship of plans and goals and so with his acting out of values open to reason. Lying to save an innocent life, by contrast, is a clear ordinary case of not doing what is wrong. Suppose that a murderer is pursuing an innocent person who has come to my house. The murderer then comes to the door and asks if I know where the person is. If I lie to him I interfere with his pursuit, but since the pursuit is unreasonable in its attempt to interfere with the innocent party’s life, I do not interfere with his acting out of values open to reason. On the other hand, if I do not lie, then I aid his unreasonable interference and so interfere with the innocent person, which is unreasonable (and clearly wrong in ordinary thinking). Little “white lies” told to spare a person’s feelings still interfere with the person’s factual basis for planning and considering. Although this liar respects the person’s feelings, he does not respect the person’s acting out of values open to reason. Telling such lies is paternalistic, since my reckoning of what is best for the other person guides my action without the participation of his reckoning of what is best for himself. In my act I leave out his being in charge of his life and, on the whole, our principle of noninterference is incompatible with paternalism. But if a person has Alzheimer’s and asks whether he can come home tomorrow and will forget whatever answer I give him, then there is nothing clearly wrong in lying to him to make him feel better momentarily. The person no longer has the capacities involved in reasonably pursuing happiness, so there is no interfering with his acting out of values open to reason. We may say in this case that lying is a custodial rather than a paternalistic act because it is directed toward taking care of a person’s life who can no longer take care of his own. Breaking a promise to someone is disappointing an expectation about my behavior that my promise induces in him. To break a promise for my own advantage (I am too tired to keep it, something better has come up, I will make a lot of money by breaking it, etc.) is something that is ordinarily thought to be clearly wrong. Now, for a person to set goals and make plans to achieve them requires not only a factual basis of what goes on in the natural world, but also facts regarding what is so in the behavior of other people. Without a stock of social expectations

264

freedom and morality

that promising is designed to augment, a person cannot rationally be in charge of his life. For me to break a promise, then, is to interfere by disappointing expectations regarding my behavior that are part of another’s having true social expectations that are required for his planning. If by contrast I promise to return a gun I borrow from someone who in the interim goes mad, his going mad destroys his rational agency. Hence in breaking my promise to him I am not interfering with his rational planning or acting out of values open to reason. Although expectations regarding others’ behavior are essential to rational agency, it is both clearly wrong in ordinary moral thinking and unreasonable according to our principle for me to expect another to fulfill any expectation I might happen to have regarding him. If I expect him to give me money simply because I want it so much, it is my expectation itself that is unreasonable, in that any action or exertion of pressure to promote its fulfillment is interference with his rational agency. Therefore, in disappointing my expectation he does not interfere with my acting out of values open to reason. Similarly, if you expect me to conform to your religion and I disappoint you, I am not doing you a wrong. In this case your very expectation that I live my life according to your values is unreasonable, since there is no way of my fulfilling that expectation without my abandoning my reasonable pursuit of happiness. To steal someone’s property for the sake of having it oneself is wrong in ordinary thinking. Stealing presupposes that things are owned. The concept of ownership is based on social agreement. I am in a relationship to something of owning it because of a general implicit or explicit agreement that anyone in that relationship is to be left to dispense with the thing as he pleases. In this conception of private property stealing is interfering with a person’s social expectations regarding other people’s behavior. Since once again acting from a reasonable pursuit of happiness requires a background of fulfilled social expectations, stealing is interfering with someone’s reasonable pursuit of happiness.9 Cheating on an exam to get good grades or cheating in a game of checkers because it makes me feel good to win are clear cases of doing what is ordinarily thought to be wrong. Cheating disappoints expectations, since people enter into a competitive enterprise according to ex9. Note, however, that we are not basing morality on social contracts. Rational agency and the entailed unreasonableness of interfering with it in others does not itself derive from social agreements. Some cases of immorality (maiming, killing) don’t violate rational agency by violating agreements, and those cases of immorality that do violate them are wrong or unreasonable because they thereby interfere with the reasonable pursuit of happiness. Although this sort of trust in others is integral to morality, it is not definitive of it and is itself accounted for by a principle of noninterference binding people together.

Rational Justification of Morality

265

pectations that others will conform to the rules of the enterprise. Since the cheater, by entering the enterprise, gives it to others to understand that he shall conform to the rules, he is implicitly saying or even promising that he will behave in certain ways, so that his cheating is lying or even breaking a promise. Although cheating is thus interfering with someone’s acting out of values open to reason, winning at the cost of someone else’s achieving their goals is not. In competitive enterprises each person pursues his values in the light of his accepting that he may lose. This is part of what he takes into reasonable consideration when he enters the contest. In winning, then, I am interfering with someone’s achieving a certain goal but not with any rational or well-considered pursuit of the goal, and so I am not interfering with his acting out of values open to reason. The person in authority who practices job discrimination is cheating in the enterprise of getting jobs. In a society where expectations are that certain groups will be discriminated against, discrimination no longer involves a violation of expectations in competitive enterprises of seeking opportunities. The expectations themselves, however, interfere with the members of the target group rationally having plans and values that involve those enterprises, and so the expectations of discrimination themselves are unreasonable. If anyone not in the target group banks on such expectations, my disappointing them by giving the minority person the job doesn’t interfere with the first person’s acting out of values open to reason, since his expectations are flat-out unreasonable. Those who go along with these interfering social expectations are participating in a tribal pact that puts their group above being rational in pursuing one’s happiness. Enslaving or brainwashing someone is clearly wrong in ordinary thinking, and each is an extreme case of putting myself in charge of someone’s life. We can define undue influence on another’s life as influence that sidesteps or circumvents his own rational consideration. If the person somehow likes or values being enslaved, he is giving up his reasonable authorship of his life to another and so is not reasonably pursuing happiness. If I enslave him I am an accomplice in this interference with his reasonable pursuit of happiness. We can define brainwashing as the instituting of values or a way of life that insulates those values from further effective consideration. It might seem that this could supply me with a set of lives in others to function as test cases for how values pan out. I produce in someone a baker’s life, in another a butcher’s life, and thereby get information about such lives. But the most crucial part of the information to be gotten from others’ pursuits is their opinion and assessment of their lives or their own considerations about how things are panning out, and this cannot survive insulat-

266

freedom and morality

ing their values. Brainwashing is distinct from socialization, which is a necessary process of maturing to adulthood with a stock of preferences and “borrowed” values. However, so long as the socialization does not destroy the ability to rationally reconsider these values, it is distinct from brainwashing and not a case of interference. So far we have considered a principle of noninterference that systematizes and rationalizes only that part of morality that proscribes certain acts or restrains our behavior. Besides proscriptions against initiating harm, our ordinary moral thinking encompasses not letting harm happen and, as well, helping. Thus if I can save a person from drowning with only minimal risk to myself, it would be wrong to let the drowning happen, and if small effort on my part can turn a drug addict’s life around, it would seem wrong not to help. These kinds of cases correspond to Kant’s imperfect duties to others,10 and to our principle that it is unreasonable to fail to promote the reasonable pursuit of happiness of others. The reason why this is a principle of imperfect duties is that one’s fulfilling or pursuing one’s own personal goals is relevant to how much it is reasonable to do for others. If my considered goals are perpetually interrupted and thwarted by helping others whenever they need it, then clearly I am violating my own acting out of values open to reason or my own reasonable pursuit of happiness. In thus acting, even though I am adding a stock of considerations and information, I am doing so at the cost of ever acting out of the values they determine, and so I make those considerations irrelevant. What this principle does importune, however, is that I should consider carefully before having goals and pursuits that by their nature preclude my helping others. What it implies besides is that only a serious interruption can excuse me,11 and that not helping, or letting harm come, is not an explicit value in its own right that I can hold as one of my pursuits or goals in life.12 Besides latitude in how much to divert myself from my goals, there is also latitude in regard to the distribution of my benevolence or in whom to help. Even here, however, there are extreme cases where the helping or the not letting harm happen seems required.13 It is the latitude in con10. The sketch that follows borrows heavily from Marcia Baron, Kantian Ethics Almost Without Apology (Ithaca: Cornell University Press, 1995). 11. In particular, if I am walking down the beach and someone is drowning I cannot reasonably fail to help him because it takes fifteen minutes out of my day, and if without great risk I can defend an innocent party against his killer I cannot reasonably let the killing happen. 12. One cannot reasonably have a value, that is, that others under all circumstances should be self-reliant. 13. It is unreasonable to let the drowning person die even if he is a stranger, even if he is an acquaintance I don’t like, and even if he is a competitor for a job I am after.

Rational Justification of Morality

267

forming to this second principle or in obeying this second imperative that distinguishes imperfect duties from perfect ones. The conformity to the second principle, then, is not in all cases categorical with respect to advantages, since a person, while conforming to the principle, may say to himself that he will help in a certain way unless it diverts him too much from achieving certain of his goals. The conformity to the principle is, however, categorical with respect to reputation. Just as being found out or not in breaking a promise is irrelevant to the interference with reasonable pursuit of happiness, so too having it be known or not that I am helping is irrelevant to the promotion of reasonable pursuit of happiness. Hence the principle per se does not justify a person’s helping only if others will know that he is. We note that although the second principle has latitude, this is not because promoting considerations and information is less a component of being reasonable about an open issue than not interfering with considerations and information.14 Indeed, the balance of reason can sometimes lie with promotion at the cost of interfering. An imperfect duty can override perfect duties, as in the case of a doctor who breaks a promise to meet me so that he doesn’t let a person die who needs surgery.15 In sum, the integrity of our own projects vis-à-vis the morality of helping others is justified, in our view, by that integrity being constitutive of the reasonable pursuit of happiness which is the grounding or justification of that morality.

conclusion The discussion of the previous section shows that reasonably pursuing happiness, in the sense of being open to considerations and information about what goals to pursue, entails a basic core of ordinary moral behavior. In this way we have a rational justification of ordinary morality. Our conception of rationality comprises the contract conception of universal agreement, since everyone must agree, first, that no one’s reasonable pursuit of happiness is to be interfered with, and, second, that they are to promote the reasonable pursuit of others. Our conception of rationality on the subject of our happiness makes clear a 14. What explains the latitude and so the “imperfection” of the second principle, I believe, is simply that not letting something happen and helping are open-ended as to time, effort, etc., and thus can upset my being a pursuer of goals at all, whereas not interfering is usually much more circumscribed. 15. Because imperfect duties themselves have latitude and because they can override perfect duties, morality based on our principles is not just a matter of applying formulas to situations. The balancing or weighing of factors in regard to reasonably pursuing happiness has no cut-and-dried formula. Intuition, feeling, imagination, and even arbitrary choice may all be essentially involved.

268

freedom and morality

basis of universal agreement, or what all reasonable people will agree to. Unless the universal agreement account incorporates this specific basis of the agreement, it will leave open the potentially opposing consideration of what I think is reasonable in obtaining my specific goals, which may not coincide with what can be universally agreed to. Thus either the contractual notion of rationality as universal agreement “reduces to” our notion of rationality over what goals to pursue, or it fails to rationalize morality definitively. Our conception of rationality also comprises regard for rationality as the public and shared capacity that in me is the source and wellspring of my setting goals and acting, since our conception entails respecting the exercise of this capacity in others. Our conception makes clear that we are respecting it for its actual employment in coming to well-considered and well-informed opinions, not for its sheer existence. Unless the conception of rationality as an indissolubly public and shared capacity incorporates this exact employment, it will leave open the potentially opposing consideration of what I think is reasonable in obtaining my goals. My reason may tell me that certain goals are the best for me to have even if such goals conflict with regard for the public and shared nature of rationality, and in such cases, either way, I must lose regard for my goals. Thus either the notion of rationality as the indissolubly public source of my goal setting “reduces to” our notion of coming to well-considered, well-informed opinions, or else it fails to rationalize morality definitively. In general, to the extent that any conception of rationality fails to incorporate our conception of coming to well-considered and well-informed opinions about what goals to pursue, it potentially leaves being thus reasonable about the content of our happiness as a counter-consideration. Thus, only a conception that explicitly incorporates ours is a definitive rational justification of morality. In this regard our justification is not one among others, but a necessary component of such a justification.

Bibliography

wo r k s b y k a n t Critique of Pure Reason. Trans. Norman Kemp Smith. Boston: Bedford, 1965. Foundations of the Metaphysics of Morals. Trans. Lewis White Beck. New York: Liberal Arts Press, 1959. Critique of Practical Reason. Trans. Lewis White Beck. New York: Macmillan, 1993. Kant’s Gesammelten Schriften. Prussian Academy Edition. Berlin: Walter de Gruyter, 1929. Vols. 4, 5, 17, 18. Kant’s Philosophical Correspondence. Ed. Arnulf Zweig. Chicago: University of Chicago Press, 1967. Kant’s Inaugural Dissertation of 1770. Trans. William J. Eckoff. New York: Ams Press, 1970. Metaphysical Foundations of Natural Science. Trans. James Ellington. Indianapolis: Bobbs-Merrill, 1970.

o t h e r wo r k s Adickes, Erich. Kants Lehre von der doppelten affection des Ich als Schluessel zu seiner Erkenntnistheorie. Tübingen: J. C. B. Mohr, 1929. Allison, Henry. Kant’s Transcendental Idealism. New Haven: Yale University Press, 1983. ———. Kant’s Theory of Freedom. Cambridge: Cambridge University Press, 1990. ———. Idealism and Freedom. Cambridge: Cambridge University Press, 1996. Ameriks, Karl. “Kant’s Transcendental Deduction as a Regressive Argument.” In Kant’s Critique of Pure Reason: Critical Essays, ed. Patricia Kitcher, 85–102. Lanham, Md.: Rowman & Littlefield, 1998. Anscombe, G. E. M. “Causality and Determination.” In Causation, ed. Ernest Sosa and Michael Tooley, 88–104. New York: Oxford University Press, 1993. Aquila, Richard. Matter in Mind. Bloomington: Indiana University Press, 1989. Baron, Marcia. Kantian Ethics Almost Without Apology. Ithaca: Cornell University Press, 1995. Baumanns, Peter. “Kants transzendentale Deduktion der reinen Verstandesbegriffe (B).” Kant-Studien 82 (1991): 329–48, and 83 (1992): 60–83. Beck, Lewis White. Commentary on Kant’s Critique of Practical Reason. Chicago: University of Chicago Press, 1960. Bennett, Jonathan. Kant’s Analytic. Cambridge: Cambridge University Press, 1966. 269

270

bibliography

———. Kant’s Dialectic. Cambridge: Cambridge University Press, 1974. ———. “Kant’s Theory of Freedom.” In Self and Nature in Kant’s Philosophy, ed. Allen W. Wood, 102–12. Ithaca: Cornell University Press, 1984. Buroker, Jill Vance. Space and Incongruence: The Origin of Kant’s Idealism. Dordrecht: D. Reidel, 1981. ———. “On Kant’s Proof of the Existence of Material Objects.” In Proceedings of the Sixth International Kant Congress, ed. Gerhard Funke and T. M. Seebohm, 183–98. Washington, D.C.: University Press of America, 1989. Chisholm, Roderick. “Freedom and Action.” In Freedom and Determinism, ed. Keith Lehrer, 11–44. New York: Random House, 1966. Davidson, Donald. “Causal Relations.” In Donald Davidson, Essays on Actions and Events, 149–62. Oxford: Clarendon Press, 1980. ———. “True to the Facts.” In Donald Davidson, Inquiries into Truth and Interpretation, 37–54. Oxford: Clarendon Press, 1984. ———. “Truth and Meaning.” In Donald Davidson, Inquiries into Truth and Interpretation, 17–36. Oxford: Clarendon Press, 1984. Dennett, Daniel. “On Giving Libertarians What They Say They Want.” In Agents, Causes, and Events, ed. Timothy O’Connor, 43–56. New York: Oxford University Press, 1995. Descartes, René. Meditation on First Philosophy. Trans. Laurence J. Lafleur. Indianapolis: Bobbs-Merrill, 1960. Devitt, Michael. Realism and Truth. Princeton: Princeton University Press, 1984. Dummett, Michael. “Truth.” In Michael Dummett, Truth and Other Enigmas, 1–24. Cambridge: Harvard University Press, 1978. ———. “What Is a Theory of Meaning?” In Mind and Language, ed. Samuel Guttenplan, 97–122. Oxford: Clarendon Press, 1975. Eckstrom, Laura Waddel. Free Will: A Philosophical Study. Boulder: Westview Press, 2000. Ewing, A. C. Kant’s Treatment of Causality. London: K. Paul, Trench, Trubner, 1924. Fair, David. “Causation and the Flow of Energy.” Erkenntnis 14 (1979): 219–50. Fisher, James Martin. The Metaphysics of Free Will: An Essay in Control. Oxford: Basil Blackwell, 1994. Friedman, Michael. Kant and the Exact Sciences. Cambridge: Harvard University Press, 1992. Gewirth, Alan. Reason and Morality. Chicago: University of Chicago Press, 1978. Ginet, Carl. “Reasons Explanation of Action: An Incompatibilist Account.” In Agents, Causes, and Events, ed. Timothy O’Connor, 69–94. New York: Oxford University Press, 1995. Guyer, Paul. “Kant’s Intentions in the Refutation of Idealism.” Philosophical Review 92 ( July 1983): 329–83. ———. Kant and the Claims of Knowledge. Cambridge: Cambridge University Press, 1987. ———. “The Possibility of the Categorical Imperative.” In Kant’s Groundwork of the Metaphysics of Morals: Critical Essays, ed. Paul Guyer, 215–48. Lanham, Md.: Rowman & Littlefield, 1998. Henrich, Dieter. “The Proof Structure of Kant’s Transcendental Deduction.” Review of Metaphysics 22 (1969): 640–59. ———. The Unity of Reason: Essays on Kant’s Philosophy. Ed. Richard L. Velkley and trans. Jeffrey Edwards. Cambridge: Harvard University Press, 1994.

bibliography

271

Hudson, Hud. Kant’s Compatibilism. Ithaca: Cornell University Press, 1994. Hume, David. A Treatise on Human Nature. Ed. L. A. Selby-Bigge. Oxford: Oxford University Press, 1888. Irwin, Terrence. “Morality and Personality: Kant and Green.” In Self and Nature in Kant’s Philosophy, ed. Allen W. Wood, 31–56. Ithaca: Cornell University Press, 1984. Kane, Robert. The Significance of Free Will. Oxford: Oxford University Press, 1996. Kemp Smith, Norman. A Commentary to Kant’s Critique of Pure Reason. 2d ed. London: Macmillan, 1918. Kitcher, Patricia. Kant’s Transcendental Psychology. New York: Oxford University Press, 1990. Korsgaard, Christine. Creating the Kingdom of Ends. Cambridge: Cambridge University Press, 1996. Lovejoy, A. J. “On Kant’s Reply to Hume.” In Kant: Disputed Questions, ed. M. S. Gram, 284–308. Chicago: University of Chicago Press, 1967. Lewis, David. “Are We Free to Break the Laws?” Theoria 47 (1981): 113–21. Mackie, J. L. The Cement of the Universe. Oxford: Clarendon Press, 1974. McTaggart, J. M. E. The Nature of Existence. Cambridge: Cambridge University Press, 1927. Mehlberg, Henry. Time, Causality, and the Quantum Theory. Dordrecht: D. Reidel, 1980. Mellor, D. H. The Facts of Causation. London: Routledge, 1995. Melnick, Arthur. Space, Time, and Thought in Kant. Dordrecht: D. Reidel, 1989. ———. Representation of the World: A Naturalized Semantics. New York: Peter Lang, 1996. Meerbote, Ralf. “Kant on the Non-determinate Character of Human Actions.” In Kant on Causality, Freedom, and Objectivity, ed. William L. Harper and Ralf Meerbote, 138–64. Minneapolis: University of Minnesota Press, 1984. Nozick, Robert. Philosophical Explanation. Cambridge: Harvard University Press, 1981. ———. “Choice and Indeterminism.” In Agents, Causes, and Events, ed. Timothy O’Connor, 101–14. New York: Oxford University Press, 1995. O’Connor, Timothy. Persons and Causes. New York: Oxford University Press, 2000. Omnes, Roland. The Interpretation of Quantum Mechanics. Princeton: Princeton University Press, 1994. Putnam, Hilary. Realism and Reason. Vol. 3 of Philosophical Papers. Cambridge: Cambridge University Press, 1985. Quine, W. V. Word and Object. Cambridge: MIT Press, 1960. ———. “Ontological Relativity.” In W. V. Quine, Ontological Relativity and Other Essays, 26–68. New York: Columbia University Press, 1969. Rauscher, Frederick. “Kant’s Conflation of Pure Practical Reason and Will.” In Proceedings of the Eighth International Kant Congress, ed. Hoke Robinson, 579–86. Milwaukee: Marquette University Press, 1995. Rawls, John. A Theory of Justice. Cambridge: Harvard University Press, 1989. Reichenbach, Hans. The Direction of Time. Berkeley and Los Angeles: University of California Press, 1956. Rescher, Nicholas. Kant’s Theory of Knowledge and Reality. Washington, D.C.: University of America Press, 1983.

272

bibliography

Rousseau, Jean-Jacques. The Social Contract. Trans. Maurice Cranston. Middlesex, Eng.: Penguin Books, 1968. Salmon, Wesley. “Causality, Production, and Propagation.” In Causation, ed. Ernest Sosa and Michael Tooley, 154–71. New York: Oxford University Press, 1993. Spinoza, Benedict. Ethics. Trans. R. H. M. Elwes. New York: Dover, 1955. Strawson, Galen. “Libertarianism, Action, and Self-Determination.” In Agents, Causes, and Events, ed. Timothy O’Connor, 13–32. New York: Oxford University Press, 1995. Strawson, Peter. Individuals. London: Methuen, 1959. ———. “Truth.” In Truth, ed. George Pitcher, 32–53. Englewood Cliffs, N.J.: Prentice-Hall, 1964. ———. The Bounds of Sense. London: Methuen, 1966. Stroud, Barry. “Transcendental Arguments.” In Kant on Pure Reason, ed. R. C. S. Walker, 117–31. Oxford: Oxford University Press, 1982. Suchting, W. A. “Kant’s Second Analogy of Experience.” In Kant Studies Today, ed. L. W. Beck, 322–40. La Salle, Ill.: Open Court, 1969. Swing, T. K. Kant’s Transcendental Logic. New Haven: Yale University Press, 1969. Taylor, Charles. “Responsibility for Self.” In Free Will, ed. Gary Watson, 111–26. New York: Oxford University Press, 1982. Vaihinger, Hans. Commentar zu Kants Kritik der Reinen Vernunft. Berlin: Union Deutsche Verlagsgesellschaft, 1892. Van Cleve, James. “Four Recent Interpretations of Kant’s Second Analogy.” Kant-Studien 64 (1973). ———. “The Ideality of Time.” In Proceedings of the Eighth International Kant Congress, ed. Hoke Robinson, 411–22. Marquette: Marquette University Press, 1995. ———. Problems from Kant. New York: Oxford University Press, 1999. Ward, Andrew. “Kant’s First Analogy of Experience.” In Kant-Studien 92 (2001): 387–406. Watson, Gary. “Free Agency.” In Free Will, ed. Gary Watson, 96–110. New York: Oxford University Press, 1982. Westphal, Kenneth R. “Affinity, Idealism, and Naturalism: The Stability of Cinnabar and the Possibility of Experience.” In Kant-Studien 88 (1997): 139–89. Wolff, Robert Paul. Kant’s Theory of Mental Activity. Cambridge: Harvard University Press, 1963. Wood, Allen W. “Kant’s Compatibilism.” In Self and Nature in Kant’s Philosophy, ed. Allen W. Wood. Ithaca: Cornell University Press, 1984.

Index

Adickes, Erich, 165 Aesthetic, 3–11, 21–24, 105, 106, 107, 118, 125–26, 149, 164, 170 Allison, Henry, 26, 56, 99, 125, 158–59, 206, 208, 210, 218, 221 Ameriks, Karl, 39 Analogies, 15–16, 20, 105, 117, 141 Anscombe, G. E. M., 121 Antinomies, 26–27, 150, 164, 169, 170 Appearances, 53, 58, 60, 71, 73, 78, 97, 101, 102, 112, 115, 118, 128, 134, 149, 150, 151, 153, 154, 155, 156, 157, 159, 161, 164, 167, 169, 170 Apperception, 12, 35–37, 41–45, 48, 50, 54, 57, 59, 60, 62, 63, 64, 68, 70, 116, 126, 128, 155, 165, 168 Aquila, Richard, 64 Aristotle, 57, 199 Assertabilist Semantics, 197–99, 201 Autonomy, 205, 206, 213, 230, 241, 242, 243, 244, 245, 246–47, 248 Baron, Marcia, 266 Baumanns, Peter, 47 Beck, Lewis White, 206, 231, 233 Bennett, Jonathan, 99, 125, 135, 221, 222, 223 Bohr, Niels, 195, 196, 197, 198 Buroker, Jill, 133, 134 Categorical Imperative, 234, 236, 242, 247, 258, 261, 262, 267; First Formulation, 230–35, 236, 237, 238, 239, 241, 244, 247; Second Formulation, 229–30, 235–39, 241, 244, 247; Third Formulation, 240–44 Categories, 12–13, 15, 39–40, 44, 50, 58, 63, 71, 116, 117, 118, 134, 147, 151, 154, 168, 169 Causation, 18–20, 53, 54, 61, 95–97, 100–101, 103, 107–9, 113, 114, 115,

116, 117, 118, 119–22, 140–41, 154, 155–56, 158, 161, 186–87, 189 Chisholm, Roderick, 227 Cognition, 35, 38, 39–40, 43, 47–50, 53, 54–55, 57–58, 59, 60, 62–63, 64–65, 67, 69–74, 76, 89, 90, 91, 92, 117, 147, 148, 149, 151, 152, 153, 154–55, 157, 161, 162, 165, 169, 173, 190, 200, 227; Causal Theories, 69, 75, 89–91; Descriptive Theories, 48, 69, 75, 88–89, 91, 152, 156, 172 Constructivism, 4, 23, 25, 30–31, 42–43, 84, 89, 91, 92, 130–31, 150, 165–66, 167, 173, 174, 176, 183 Continuity, 4–6, 22–23, 42, 49, 70, 105, 126, 131, 150, 165, 170 Contractualism, 230, 231, 232, 234, 239, 242–44, 248, 264, 267–68 Copernican Revolution, 115, 159, 162, 185, 190, 195, 200 Davidson, Donald, 87, 185, 186, 187, 209 Dennett, Daniel, 217 Descartes, Rene, 131, 135, 136, 137, 139, 140, 143 Devitt, Michael, 90 Double Affection, 161, 163 Dummett, Michael, 85, 92 Eckstrom, Laura, 214 Empirical Character, 209, 219–21 Empirical Idealism, 132 Ewing, A. C., 104 Facts, 185–90, 199–200, 201 Fair, David, 120 Falkenstein, Lorne, 26 First Analogy, 16–19, 71–73, 109–13, 114, 117, 125–28, 129, 132, 135, 141 First Antinomy, 173–78, 179, 183 Fisher, John Martin, 211

273

274

index

Freedom, 205, 206, 207, 213–16, 225, 226, 227, 228, 260; Transcendental, 205, 209–10, 217–23 Frege, Gottlob, 76 Friedman, Michael, 70, 84, 91 Geometry, 7–9, 84 Gewirth, Alan, 249 Ginet, Carl, 225 Guyer, Paul, 19, 113, 134, 135, 179, 237, 241 Hallucinations, 77, 138, 139, 141, 143 Happiness, 230, 243, 245–47, 248, 250–51, 252, 255; Reasonable pursuit of, 250–54, 255–58, 259, 262, 264, 265, 266, 267 Henrich, Dieter, 64, 240 Hobbes, Thomas, 240, 259 Hudson, Hud, 209 Hume, David, 90, 97, 107–9, 115, 118, 122, 187 Illusions, 78 Incompatibilism, 207, 208, 211–16, 222 Infinity, 174–75, 179; Potential, 160, 161,174, 175, 176, 177, 179, 184 Instrumentalism, 194–95, 196, 197, 198 Intelligible Character, 209, 219–21 Internal Realism, 192, 193, 195, 198 Irwin, Terrence, 206, 210 Judgment, 42, 55, 58, 59, 60, 63, 67, 71, 123 Kane, Robert, 215, 224 Kemp Smith, Norman, 3, 56, 104, 114, 160 Kitcher, Patricia, 4 Korsgaard, Christine, 229, 230, 233, 246, 248, 249 Lambert, Johann, 21, 24, 25, 26 Leibniz, Gottfried Wilhelm, 6, 22, 27, 35, 40, 91, 103–4, 159, 160, 161, 219, 223 Locke, John, 96 Logical Functions, 47, 54–56, 57–58, 60, 62, 63, 64–66, 67, 68, 123, 151, 182 Lovejoy, A. J., 98, 115 Mackie, J. L., 119, 120, 122 McTaggart, John, 120 Meerbote, Ralf, 209 Mehlberg, Henry, 97

Mellor, D. H., 120, 121 Metaphysical Deduction, 54, 55, 57–58, 60–61, 63, 64, 65, 68 Metaphysical Realism, 190, 192, 194, 195, 196, 197, 198, 199, 200 Mill, J. S., 187 Morality: Definition of, 232, 233, 234, 244; Justification of, 247, 248, 249–50, 261, 264, 267–68; Motivation for, 231, 233, 234, 235, 236, 239, 241, 242, 244, 247, 249; Ordinary, 261–67; Principle of, 258–61, 266–67 Motion: intrinsic, 28–29, 180–81; of the Subject, 6, 13, 23, 27, 29, 43, 50, 70, 105, 116, 126, 147, 164, 165 Newton, Isaac, 6, 7, 23, 142 Noumena, 152, 155, 157, 158, 162 Nozick, Robert, 214, 215, 224 Objectivity, 41, 47–48, 53, 57–58, 62, 64, 97–98, 101, 114, 152–52, 154, 155, 156 O’Connor, Timothy, 227 Omnes, Roland, 193, 199 Phenomena, 148–49, 152, 162 Phenomenalism, 53, 109, 115, 118, 149, 154, 158, 161, 169, 182, 183 Plato, 252, 258 Practical Reasoning, 205, 206, 207–9, 210–16, 223, 224–26, 227, 228 Pragmatism, 196, 197 Putnam, Hilary, 185, 186 Quantum Mechanics, 121–22, 185, 190–95, 196, 197, 198, 199, 200, 201, 205, 224, 226, 227, 228; Decoherence theory of, 193–95, 198, 199, 201, 226, 227 Rauscher, Frederick, 206 Rawls, John, 261 Refutation of Idealism, 125, 128, 129, 131, 134, 135–39, 141, 142 Reichenbach, Hans, 119 Relativistic Physics, 28–29, 84, 92, 119, 181 Rescher, Nicholas, 155 Resposnsibility, 205, 207, 216, 221–23 Rousseau, Jean-Jacques, 239–40, 242–43, 244, 247 Rules, 13–15, 41–43, 48, 50, 54, 57, 59, 63, 65, 70–71, 74–75, 89, 91, 97–98,

Index 116, 117, 126, 128, 148, 149, 151, 152–53, 154, 161, 162, 165, 158, 169, 171, 175 Russell, Bertrand, 78 Salmon, Wesley, 120 Schemata, 47, 59–63, 67, 68, 103, 123, 151, 182 Schematism, 59–63, 68 Schopenhauer, Arthur, 114 Schroedinger, Ernst, 193 Second Analogy, 18–20, 53–53, 97–107, 112, 113–14, 118, 140, 141, 155 Second Antinomy, 160 Space: Constructivist, 4, 45, 49, 69–70, 126, 129, 159, 160, 164, 167, 169, 181, 184; Ideality of, 10, 20, 28, 31, 130, 132, 147, 150, 151, 164–65, 167, 180; Metaphsical Exposition of, 4–7; Pure Intuition, 3–4, 42, 49, 69–70, 126, 130; Transcendental Exposition of, 7–9 Spinoza, Benedict, 95 Strawson, Galen, 95 Strawson, Peter, 3, 30, 39, 40, 44, 55, 89, 98, 115, 179, 180, 186 Stroud, Barry, 118 Substance, 16–19, 52–53, 65, 71–73, 80–82, 91, 92, 95–97, 109, 111, 112, 113, 114, 115, 116, 117, 122–24, 128, 134, 135, 138, 155, 158, 160, 161, 168–69, 170, 181, 182, 183 Suchting, W. A., 104 Swing, T. K., 3, 106 Tarski, Alfred, 87 Taylor, Charles, 212 Things as they Appear, 152, 153, 156–58, 159, 161, 163 Things in Themselves, 147, 151, 152, 153, 154, 155, 157–58, 159, 161, 171 Third Analogy, 114 Thoughts, 74–76, 84–85, 90, 91, 92, 147, 148, 188–89, 190 Time: Causal Theory of, 19–20, 96, 97, 102–5, 106–7, 118, 119, 120; Constructivist, 10–11, 17, 23–24, 43, 45, 49, 51, 69–70, 71–72, 105, 106, 109,

275

110, 112, 120, 122, 123, 126, 127, 137, 164, 166, 167, 169, 181, 184; Ideality of, 17, 20, 26, 27, 30, 31, 106–7, 124, 130, 132, 164–65, 167; Necessary Advance of, 18–20, 53–54, 96, 99–102, 103–5, 106, 108–9, 112, 118, 119, 122, 140–41, 155; Pure Intuition, 49, 51, 69–70, 95, 105, 109, 110, 118, 126, 130; Past, 16–19, 51–52, 72–73, 91, 111–12, 113, 124, 127, 128, 136, 139, 166, 168–69, 171, 177 Transcendental Acts, 30–31 Transcendental Affection, 154, 161 Transcendental Arguments, 118 Transcendental Deduction, 11–13, 15, 35–46, 47–50, 52, 54, 62, 67, 115–18, 126, 152–53, 156, 165; in B-Edition, 47, 54, 62–67, 68; Subjective, 45 Transcendental Idealism, 10, 118, 134–35, 149, 159, 164–70, 173, 179, 180, 181, 182, 183 Transcendental Illusion, 27, 171 Transcendental Imagination, 12, 13, 42–44, 45–46, 49–50, 61–62, 65–66, 70, 116, 117, 126, 151, 165, 168 Transcendental Logic, 56 Transcendental Object, 152, 153, 155, 158, 159, 161, 162, 171, 219 Transcendental Psychology, 31, 45–46 Transcendental Realism, 132, 164, 170–73, 174–78, 179, 180, 182, 183 Trendelenburg, Friedrich, 21, 26, 27, 30, 171 Truth, 54–55, 84–88, 92, 188, 189, 190, 200 Vaihinger, Hans, 3, 26, 171 van Cleve, James, 21, 99, 155 verificationism, 91–92, 183 Ward, Andrew, 18 Watson, Gary, 212 Westphal, Kenneth, 118 Wolff, Robert Paul, 3, 39, 106 Wood, Allen, 218, 223 Zeno’s Paradoxes, 6, 7, 165