Kant's third analogy of experience

Kant-Studien

Volumn 88, Issue 4, 1997, Pages 406-441

  Watkins, Eric ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0013539391     PISSN: 00228877     EISSN: None     Source Type: Journal    
DOI: 10.1515/kant.1997.88.4.406     Document Type: Article

References (102)

1. To avoid confusion, I shall use 'Principles' when I refer to Kant's 'Grundsätze des reinen Verstandes' in the first Critique's Transcendental Analytic
2. Especially given the close connection Kant sees between the Principle of Succession (the pre-Critical counterpart of the Second Analogy) and the Principle of Coexistence (the preCritical counterpart of the Third Analogy) as early as the Nova dilucidatio (1755)
3. Peter Strawson, The Bounds of Sense. An Essay on Kant's Critique of Pure Reason, pp. 139-140 (Methuen, London, 1966)
4. Jonathan Bennett, Kant's Analytic (Cambridge University Press, Cambridge, 1966, p. 219): "The third Analogy, concerning, community, does not require extended discussion. It says: 'all substances, insofar as they can be perceived to coexist in space are in thoroughgoing reciprocity.' Kant is here making the interesting and rather Spinozist claim that we could not know that two things coexisted in the same universe unless they had causal commerce (= community) with one another. His attempt to prove this, however, is a failure which is not even incidentally valuable except for a few flickers of light which it throws on the second Analogy." Although this attitude is typical, I hope to show that it is mistaken
5. Henry Allison, Kant's Transcendental Idealism. An Interpretation and Defense (Yale University Press, New Haven, 1982)
6. C. D. Broad, Kant. An Introduction (Cambridge University Press, Cambridge, 1978)
7. A. C. Ewing, Kant's Treatment of Causality, 1924, reprinted 1969 by Archon Books
8. Herbert James Paton and Norman Kemp Smith (prominent examples of less recent commentators) are much more systematically complete in this respect as each devotes more than a brief dismissal to the Third Analogy. I should note that the literature in German pays even less attention to the Third Analogy than does the Anglo-Saxon literature. Two exceptions are Jeffrey Edwards ("Der Ätherbeweis des Opus postumum und Kants 3. Analogie der Erfahrung", in: Übergang. Untersuchungen zum Spätwerk Immanuel Kants, (Ed.) Forum für Philosophie Bad Homburg, Vittorio Klostermann, Frankfurt am Main, 1991, pp. 77-104)
9. and Hermann Schmitz (Was wollte Kant? Bouvier, Bonn, 1989)
10. Arthur Melnick, Kant's Analogies of Experience (University of Chicago Press, Chicago, 1972)
11. Paul Guyer, Kant and the Claims of Knowledge (Cambridge University Press, Cambridge, 1987). All references to Guyer will be from this work
12. Guyer, as far as I can tell, is responsible for this name
13. Actually, the problem of time-determination is generally conceived of as including objective temporal properties in general and not only objective temporal relations. While much of the following discussion applies to the problem as the general conception poses it, I am concerned with only the more narrow conception
14. Guyer adds the further aspect to the problem of time-determination that one cannot take for granted "that the order of perceptions is known independently of anything else" (p. 256). In other words, even "the subjective order of time" is not given. However, this added aspect is not directly involved in the argument of the Third Analogy
15. And even if it did, there is still a jump from these properties to their interpretation as the object's objective temporal properties
16. In a sense, even a clock does not have listed on it its temporal properties, since the clock's state must be interpreted, i. e., is not immediately given. Further, it is only due to its causal mechanism that we are inclined to view a clock as a reliable indicator of "objective" time
17. I would not want to take up a defense of the claim that the other Principles state rules for this procedure as well. (Cf. Guyer pp. 184 f., 194, and 202 for the difficulties such a defense would have to overcome.)
18. Cf. Bernhard Thöle's discussion of the problem of time-determination in Kant und das Problem der Gesetzmäßigkeit (De Gruyter, Berlin, 1991), pp. 128-150
19. Properly speaking, of course, it is not the relational categories per se, but rather the schemata of the relational categories that must be employed in Kant's answer in the Analogies, since the categories proper (i. e., unschematized) have no temporal meaning
20. Guyer raises this objection on p. 178: "Kant now writes as if the argument for the principles must go back to the very foundations of the deduction itself - as if it must restart the argument of the deduction - rather than just apply the conclusions already reached in the deduction and the schematism". He then weakens the objection as follows: "When he comes to the central paragraph of this introduction to the principles, Kant does not write as though there had been no transcendental deduction at all. But he certainly does write as though the deduction had established at most the very general point that experience or empirical knowledge requires the concept of an object, and as if the schematism had added only the equally general point that this must be the concept of an object capable of being given by means of the kind of sensibility we possess". This passage expressing the weaker objection seems to me to contradict his general explanation of the Analogies in which he interprets the Analogies as being necessary for experience of independent objects at all, p. 209
21. In yet other words, it is the Transcendental Deduction that distinguishes the categories from the regulative Ideas of reason, which are, like the categories, necessary, but, unlike the categories, not epistemically justified. But see Kant's transcendental deduction of Ideas (A 336/B 393)
22. Again, Guyer levels this charge on p. 180: "There is no suggestion that we know that we can make just these three forms of time-determination because we independently know that there are three categories of relation, or that we must be able to make these three kinds of time-determination because only thus will we be able to use or lend objective reality to such categories. ... Kant goes directly from the modes of time which have to be determined because of the structure of time itself and the possible relations among objects in it to the existence of principles of the understanding, without either an advance road map or detour"
23. I should note that what I identify as the three modes of time does not necessarily agree with everything Kant says about the modes of time. However, Kant does speak of the three modes of time at B 219
24. It should be noted that Kant understands the disjunctive form of judgment to express not only 'not both a and b', but also 'at least a or b'
25. This follows due to the (somewhat peculiar) way in which Kant understands the disjunctive form of judgment. One might object that the second item related in a disjunctive form of judgment need not exist at all and thus not simultaneously. However, I take it that a non-existent item would have a temporally distinct feature, namely that of not existing at the time when the first item exists
26. This justification ought to reveal the nature of Kant's argumentation here. He cannot be presenting rigorous arguments for the necessity of the three modes of time. Rather, I believe he is presenting his rationale for thinking that there are three modes of time. This rationale need not take on a deductively valid form
27. Again, Kant ought to be taken here as suggesting that logical priority applied to time is naturally to be taken as temporal priority or succession. But it is not absolutely necessary to do so
28. This objection has the disadvantage that it cannot explain why there shouldn't be more or fewer Analogies. For example: Why isn't there an Analogy enabling judgments that 13 substances exist three seconds apart? The account I propose has a natural explanation why the number of Analogies ought to be three
29. In my dissertation (Kant's Third Analogy of Experience, University of Notre Dame, 1994, pp. 1-75), I consider the two editions' formulations of the Principle of the Third Analogy and show that they are not in fact distinct. Rather, the second edition formulation expresses more clearly the actual intent of the first edition formulation
30. Occasionalism is also under attack here, but during Kant's philosophical activity there were certainly fewer proponents of occasionalism than of either the physical influx or the pre-established harmony theories. This opposition may also begin putting the very term "mutual interaction" into its proper perspective. For in light of this context, mutual interaction may be taken to mean reciprocal causality (or real connection) whose contrast term is some, according to Kant "inexplicable" subspecies of interaction. The standard view takes the Third Analogy's central concept of mutual interaction to be most closely related to Newton's law of universal gravitation. See my discussion in my disssertation (pp. 187-319) of the relation of the Third Analogy to Newton's law of universal gravitation
31. See, e. g., §§ 8, 14 in Leibniz's Discourse on Metaphysics
32. The manner in which Leibniz or a Leibnizian view is Kant's target here is far from straight-forward. But even here one can note that at least for Leibniz each substance does depend in some manner on God at each and every moment so that he could say that all created substances stand in a direct causal relation with God and thus indirectly with each other. (Granted, this causal relationship falls short of mutual interaction, since it is a merely one-way relationship.) The issue of simultaneity of substances in Leibniz is also somewhat complex, since for Leibniz time is merely a well-founded phenomenon, not one of the irreducible properties of substances. Thus, Leibniz might respond by claiming that, properly (i. e., metaphysically) speaking, substances do not exist simultaneously. However, such an interpretation assumes that what is derivative (time) is not real (presumably for that very reason), a less than obvious principle, especially in light of Leibniz's claim that time is a well-founded phenomena. Fortunately, the Third Analogy does not depend in any essential way on these issues. For further discussion of related issues, see my "Kant's Theory of Physical Influx" in the Archiv für Geschichte der Philosophie 77 (1995), pp. 285-324
33. The warrant for this epistemic qualification will be provided shortly
34. A 212-213: "Nehmet nun an: in einer Mannigfaltigkeit von Substanzen als Erscheinungen wäre jede derselben völlig isoliert, d. i., keine wirkte in die andere, und empfinge von dieser wechselseitig Einflüsse, so sage ich: daß das Zugleichsein derselben kein Gegenstand einer möglichen Wahrnehmung sein würde, und daß das Dasein der einen, durch keinen Weg der empirischen Synthesis, auf das Dasein der anderen führen könnte. Denn, wenn ihr euch gedenkt, sie wären durch einen völlig leeren Raum getrennt, so würde die Wahrnehmung, die von der einen zur anderen in der Zeit fortgeht, zwar dieser ihr Dasein, vermittelst einer folgenden Wahrnehmung bestimmen, aber nicht unterscheiden können, ob die Erscheinung objektiv auf die erstere folge, oder mit jener vielmehr zugleich sei. Es muß also noch außer dem bloßen Dasein etwas sein, wodurch A dem B seine Stelle in der Zeit bestimmt, und umgekehrt auch wiederum B dem A, weil nur unter dieser Bedingung gedachte Substanzen, als zugleich existierend, empirisch vorgestellt werden können. Nun bestimmt nur dasjenige dem anderen seine Stelle in der Zeit, was die Ursache von ihm oder seinen Bestimmungen ist. Also muß jede Substanz (da sie nur in Ansehung ihrer Bestimmungen Folge sein kann) die Kausalität der anderen in sich enthalten, d. i. sie müssen in dynamischer Gemeinschaft (unmittelbar oder mittelbar) stehen, wenn das Zugleichsein in irgendeiner möglichen Erfahrung erkannt werden soll. ... Also ist es allen Substanzen in der Erscheinung, sofern sie zugleich sind, notwendig, in durchgängiger Gemeinschaft der Wechselwirkung untereinander zu stehen."
35. In all of the following, when I speak of two substances, I mean two non-identical substances
36. Although Kant uses the term 'empirical synthesis' here, what he means by it is something more like Hume's stream of consciousness. Kant is more explicit about this point in § 26 of the second edition Transcendental Deduction when he refers to what results from the synthesis of apprehension as the cognitive process underlying apprehension
37. This statement ignores the difficulties raised above in footnote 33 for such attributions to Leibniz. For Leibniz does not always prefer to speak properly (i. e., metaphysically)
38. One could also circumvent this objection by reformulating C1 as C1: If substances isolated from each other exist simultaneously, then their simultaneity cannot be known. But still, C1′ below would seem to best capture Kant's intent
39. Cf. The discussion of the Principles and the Principle of the Third Analogy above
40. Kant does argue in the Nova dilucidatio that isolated substances could not change, but it is not obvious that Kant is reusing that argument here (except in the minimal sense that both involve consideration of isolated substances). For the earlier argument is metaphysical in nature, the later one epistemological
41. This difficulty is also noted by Paton (V. 2, p. 298)
42. Broad (p. 177)
43. and later Schmitz (p. 267). I am assuming, as they do, that permanence just means "existing at all times". Naturally, if permanence does not have this meaning, this difficulty does not arise
44. 1. I take it that this point is distinct from that of not being able to perceive time itself or absolute time. For this amounts to the claim that if there were an absolute time, substances would not relate to it directly
45. In short, our "knowledge" of the states of these two substances would be temporally indeterminate
46. The general logical structure of this step of the argument is approximately as follows: P5: Y only if Z P6: X only if Y C2: X only if Z. This reconstruction is, however, only approximate because the terms Y have been treated as if they were exactly identical, though they are not. The Y as it occurs in P6 considers the determination of the place in time of two substances, whereas the Y as it occurs in P5 considers the determination of the place in time of a single substance. However, this difference is not important since the Ys could be made equivalent by formulating a complicating premise P7: The place in time of two substances can be determined only if substance S determines the place of substance S′ in time and substance S′ determines the place of substance S in time (i. e., a two-fold application of the antecedent of P5). (This premise can be safely suppressed because P5 justifies P7, that is, how the place in time of a single substance is determined applies, mutatis mutandis, to how the place in time of two substances can be determined.)
47. If P5 should not follow from C1′, there are two other possible justifications. First, in the second edition Kant restricts the conclusion of the argument to spatial substances. If P5 holds only for spatial substances, this restriction becomes intelligible. Why would P5 hold only for spatial substances? Kant may have held, along with other moderns, that matter is inert, i. e., causally inefficacious. Since Kant views matter as spatial substance, P5 is justified and historically uncontroversial. A second, more contentious justification lies in Kant's description of causality in the Second Analogy as being the relation between a condition and the conditioned. It may be the case that for finite substances it is impossible for a substance to be the condition of itself. (Perhaps considerations similar to Descartes's in the Third Meditation are relevant here, according to which if a finite substance were the cause of itself, i. e., of its substantiality (i. e., were to condition itself), it would make itself perfect, i. e., non-finite, which is contradictory. However, this kind of argument does not seem to hold for mere properties of substances.)
48. It should be noted that direct perception is a technical term that includes any of the following: intuition, sensation, the laws of the association of ideas, the flow of consciousness and Kant's judgments of perception
49. Cf. Terence Wilkerson, Kant's Critique of Pure Reason: A Commentary for Students (Oxford University Press, Oxford, 1976)
50. and Strawson, The Bounds of Sense
51. P5″ is not positing the material existence of a plurality of substances, but rather advancing the hypothetical thesis that if judgments of coexistence are to be made, a plurality of substances exist
52. Cf. Guyer, pp. 269-271. But Guyer is neither the only nor the first one to make this point. I would perhaps only want to restate the point such that reversibility is part of what coexistence means to human beings (or to beings for whom there is a distinction between subjective and objective time). Thus, any Critical account of coexistence must also account for reversibility
53. In fact, Kant seems to think that the subjective order of our representations must be derived from the objective order of objects. Cf. A 193/B 238
54. Guyer (p. 270) argues against the plausibility of such an argument, although he grants that its possibility remains. Guyer is certainly correct to note that Kant does not provide one
55. The German text reads as follows: "Zugleich sind Dinge, wenn in der empirischen Anschauung die Wahrnehmung des einen auf die Wahrnehmung des anderen wechselseitig folgen kann, (welches in der Zeitfolge der Erscheinungen, wie beim zweiten Grundsatze gezeigt worden, nicht geschehen kann). So kann ich meine Wahrnehmung zuerst am Monde, und nachher an der Erde, oder auch umgekehrt zuerst an der Erde und dann am Monde anstellen und darum, weil die Wahrnehmungen dieser Gegenstände einander wechselseitig folgen können, sage ich, sie existieren zugleich. Nun ist das Zugleichsein die Existenz des Mannigfaltigen in derselben Zeit. Man kann aber die Zeit selbst nicht wahrnehmen, um daraus, daß Dinge in derselben Zeit gesetzt sind, abzunehmen, daß die Wahrnehmungen derselben einander wechselseitig folgen können. Die Synthesis der Einbildungskraft in der Apprehension würde also nur eine jede dieser Wahrnehmungen als eine solche angeben, die im Subjekte da ist, wenn die andere nicht ist, und wechselweise, nicht aber daß die Objekte zugleich seien, d. i. wenn das eine ist, das andere auch in derselben Zeit sei, und daß dieses notwendig sei, damit die Wahrnehmungen wechselseitig aufeinander folgen können. Folglich wird ein Verstandesbegriff von der wechselseitigen Folge der Bestimmungen dieser außer einander zugleich existierenden Dinge erfordert, um zu sagen, daß die wechselseitige Folge der Wahrnehmungen im Objekte gegründet sei, und das Zugleichsein dadurch als objektiv vorzustellen" (B 257)
56. One rather substantial assumption would be that time itself must contain all of the properties true of all substances at any moment. Another is that human beings have the capacity to perceive properties of independent substances at a single moment (so that a judgment, which can only occur in a single moment, can take place). Finally, one would have to be able to assume that the judgment that we are perceiving time itself is objectively valid
57. Consider, for the sake of comparison, Guyer's reconstruction of this premise: "Coexistence cannot be inferred directly from any immediate knowledge of the reversibility or even actual reversal of representations ... There must therefore be some sort of rule from which the coexistence ... can be inferred" (p. 271). This reconstruction has the advantage that it need not encounter the difficulties involved in linking the Third Analogy up with the appropriate category, but at the same time the disadvantage discussed above in the first section, namely that the argument would appear to proceed independently of the Transcendental Deduction
58. In a sense the first edition's argument showing that the mere existence of substances does not establish the possibility of judging their coexistence fits in here because the existence of these substances alone does not (as PB4 claims) allow one to infer Reversibility
59. Of course, some differences are possible, but some can be excluded on the basis of being non-Kantian in nature. What this exclusion means in effect is that the Third Analogy relies on the success of the Transcendental Deduction in the following manner. The Transcendental Deduction is to show that the categories are indispensable for knowledge. If the Third Analogy were successful without the use of any category, then it would, in effect, violate the conclusion of the Transcendental Deduction. So the question is how the Third Analogy can incorporate its corresponding category. In the strongest (and most faithful) version of a reconstruction of Kant's project, the Third Analogy must display an epistemic function in the particular case of time-determination which only a category can fulfill. This is what PB4 must do. This interpretation obviously does not solve all of the difficulties of the Transcendental Deduction, but it does at least make sense within the Kantian project and render the text more coherent
60. Cf. ch. 4 (esp. pp. 104-123) in Robert Pippin, Kant's Theory of Form (Yale University Press, New Haven, 1982) for a discussion of Kant's theory of empirical concepts
61. It should be noted that PB4 does not rest on the success of the Metaphysical Deduction in the sense that the Metaphysical Deduction is intended to establish a complete or exhaustive table of categories. PB4 requires only a much more modest claim, namely that there really be a category of community with the meaning Kant suggests. Whereas the completeness of the table of categories is rarely granted, it seems likely that the more modest claim should be. Cf. R. Brandt, Die Urteilstafel; Kritik der reinen Vernunft A 67-76; B 92-101 (Felix Meiner, Hamburg, 1991)
62. It is at this point that the lack of completeness of the table of categories becomes problematic. For if there are indefinitely many categories, such a process could never be completed
63. Cf. my discussion in my dissertation (pp. 187-319) on the relation between the first Critique and the MFNS
64. Cf. Michael Friedman's interesting work on Kant's derivation of the law of universal gravitation in Kant and the Exact Sciences (Harvard University Press, Cambridge, 1992)
65. That such an assumption is not warranted can be illustrated through a related example. From a lack of effects (or change), one cannot infer the lack of forces. For example, the book on the table is not changing, but it is exerting a force on the table. Similarly, even if Mars is not successful in moving the table toward it, it is exerting a force of attraction (i. e., its "half" of the mutual interaction)
66. In this respect, the Third Analogy's causal ties may be unlike the Second Analogy's, though Kant does not explicitly claim in the Second Analogy that the succession of states must be a successive change of states
67. Robert Audi has introduced a valuable distinction between positive and negative epistemic dependence and explained it as follows: Certain beliefs depend positively on the processes that produce and sustain them, yet they are only negatively dependent on what would undermine these processes under certain counter-factual circumstances. The point for the present example is that whether Mars could have acted differently and caused the table to have different properties (through which we know the table at all) does not change the fact that knowing that, e. g., Mars in fact produces a gravitational effect on the table is required for judging that the table and Mars exist simultaneously
68. Let me qualify this claim. The argument does imply Reversibility, which, I take it, must be cashed out in terms of counterfactuals. (Had I been looking there instead of here at that time, I would have perceived x, and had I then looked here at the later time, I would have perceived y.)
69. Kant seems to have discovered this difference between logical and real opposition sometime before 1763, since in his An Attempt to Introduce the Concept of Negative Quantities into Philosophy he introduces the distinction between logical and real opposition. Logical opposition is based on the 'law of identity' and is purely formal, whereas real opposition is characterized as "taking place only insofar as of two things qua positive grounds one suspends the consequence of the other ... Thus, in real opposition both predicates must be positive yet such that in their connection the consequences in the same subject are reciprocally suspended" (2:175-176)
70. And Kant says about real opposition: "The one suspends what the other one posits; only the consequence is something (cogitabile). The moving force of a body in one direction and an equal striving of the same in the opposite direction do not contradict each other, and are simultaneously possible as predicates in one body. The consequence of it is rest, which is something (repraesentabile)" (2:171)
71. See Klaus Reich (Die Vollständigkeit der kantischen Urteilstafel, Berlin, Schoetz, 1948)
72. Lorenz Krüger ("Wollte Kant die Vollständigkeit seiner Urteilstafel beweisen?", Kant-Studien 59 (1968), pp. 333-356) and R. Brandt for a discussion of these issues
73. Melnick, pp. 96
74. 1)
75. Melnick seems to admit that a single causal tie can establish only that the effect occurs after the cause
76. After all, the world of which a, b, and c are a part might split up into two distinct worlds (with b in one and c in the other) such that we can still observe the motion of both b and c. If this were to occur, a and b would not be simultaneous, since they occupy distinct worlds, which would have distinct times and thus no possible relation of simultaneity, since simultaneity presupposes a single time
77. Actually, I do not think that taking the subsequent motion of a will in fact be sufficient to determine the simultaneity of b and c, but this insufficiency is due to the transient nature of impact
78. It should be noted in this context that this dynamical law is not uniquely described. First, it seems to contain some kind of 'causal field', when Melnick lists a number of conditions and then adds 'and so on'. And a causal field may involve mutual interaction implicitly. Second, the dynamical law contains both the acting force of a on b and the acting force of a on c, but Melnick does not explicitly consider the possibility that these two acting forces may be related in an important way. Melnick does isolate the impact of a as the (initiating) condition that allows us to apply the law, but, again, it is not obvious how this is to be interpreted. Mathematically speaking, of course, there is no difficulty in calling the impact of a an initiating condition, but one would need to determine what metaphysical assumptions this mathematical way of speaking has, an issue Melnick does not consider
79. Although the argument that I have reconstructed above helps us to see where Melnick's criticism goes awry, it is clear that Melnick's criticism is inadequate even on his own terms. He does not show that his dynamical law does not imply mutual interaction and he seems to admit that the causal ties simpliciter establish for Kant only that the one state occurs before the other
80. Guyer, p. 271
81. Guyer's use of the term 'actual' is ambiguous. It might mean either that Kant in fact held the weaker position or that Kant is only rationally justified in accepting the weaker position (although he tries to hold the stronger position). I suspect Guyer's considered position is the latter
82. Whether Kant in fact allows the cause and the effect to be simultaneous is a contentious point, but it is not disputed that he does not require such simultaneity
83. Another difficulty, perhaps not fatal, is that Guyer's one-way causal relationship requires that this causal tie act faster than the speed of light, since the cause and the effect must be simultaneous in such a way that if the cause and effect are not contiguous, the cause must act immediately (i. e., faster than the speed of light) on the effect, which violates special and general relativity theory
84. By place in time here I mean something like McTaggart's B-series of substances being located earlier, simultaneous, and later than other substances
85. And Guyer would seem to accept the first step of the argument, at least in principle, since he accepts the problem of time-determination. Thus, while the argument proposed in section II helps us to see the difficulty with Guyer's objection, Guyer's objection fails on its own terms
86. In fact, P5‴ is almost a quotation of Guyer's criticism of Kant's notion of mutual interaction
87. One might postulate a third entity that determines the states of S and S′ in virtue of two causal ties. But this maneuver will not be successful as was shown in the discussion of Melnick's interpretation nor will it allow for Guyer's alternative, since Guyer wants to postulate a single causal tie
88. Unless Guyer can show that a substance can act on itself only if it also acts on another substance, but Guyer provides no such argument and such a claim does not seem intuitively plausible
89. I should note that B. Jeffrey Edwards presents an interpretation of the Third Analogy according to which an inconsistency arises between the Third Analogy and the general project of the first Critique: Edwards states: Through his attempts at a Critical justification of the Principle of Community as a rule of the unity of experience Kant makes implicitly clear that the presence of a comprehensive dynamic field of material activity is to be understood as a condition of our experience of the empirically accessible whole of objects (or substances). Thus the argument against empty space aims to show that the unity of objective experience is neither rooted only in the formal determinacy of human sensibility nor proceeds only from the objectively constitutive epistemic achievements of the pure understanding. The proper meaning of the argument thus seems to me to lie in the fact that a theory of a priori knowledge of objects cannot be adequately constructed on the sole foundation of a transcendental formalism of understanding and intuition. (p. 88) In a footnote Edwards goes on to explain that this "transcendental formalism of understanding and intuition" contains (or at least implies) Kant's idea that the first Critique can anticipate only the form of experience a priori, not any of its matter, which contradicts the Third Analogy on Edwards's interpretation, according to which the Third Analogy establishes a comprehensive dynamic field of material activity (i. e., the matter of experience) as a condition of our experience of an empirical world of substances. Two points about Edwards's interpretation seem subject to criticism. First, even if his interpretation of the Third Analogy were correct, it would not seem to violate the first Critique's limitation to merely formal knowledge. For the Third Analogy (even on Edwards's interpretation) does not claim to be producing a priori any 'material' knowledge. Rather, it is making the hypothetical claim that if one is to have knowledge of coexistence, then one must experience mutual interaction. This conditional claim is not identical with the categorical claim that there is a dynamical field of substances in mutual interaction (and even if it were, it is still not obvious that anything 'material' has thereby been proved, since the form-matter distinction has not been explained sufficiently). Second, Edwards's interpretation of the Third Analogy's intent is not uncontroversial. Edwards wants to read the Third Analogy as establishing a dynamical field of substances that prohibits the possibility of a void. He takes the first step of the argument (what we above called the isolated substances argument) to show that a void is not possible. However, such a reading is not necessary. As I suggested above, what is important in the argumentative structure of the first step of the Third Analogy is a lack of causal influence of one substance on another, regardless whether anything occupies the space in between these two substances (i. e., regardless whether there is a void). Since neither a void nor a causally inert body lying in between two substances would contribute to the determination of a substance's place in time, the void/plenum issue is not directly relevant to the Third Analogy. (The Third Analogy could very well obtain for a Swiss cheese universe, one with voids interspersed throughout.) Accordingly, on the above reading, the Third Analogy could in principle allow for empty gaps between mutually interacting substances; rather, other arguments in the Critique of Pure Reason (e. g., the Anticipations of Perception) aim to exclude the possibility of a void
90. 2 in A; this causally necessitates a later alteration in B; and so on". Italics added. Broad finds textual support in Kant's distinction between mediate and immediate community and the example used to illustrate it, namely that of the light mediating between celestial bodies and our eyes. This example need (or ought) not be taken to show anything other than that mediating entities may be required for distant objects to interact mutually. The italicized phrases indicate how Broad's model differs from the one just developed. On Broad's model, one never knows, it would seem, that two substances coexist in Kant's sense, since one would never know how both substances exist at any single given time (where the phrase 'how the substances exist' refers to the distinction drawn above)
91. I take this independence to be what caused the difficulty in the previous model
92. Causal in the broad sense of determination: one substance determines (some part of) the other's state
93. It should be noted that this explanation of mutual interaction implies that the mediacy/immediacy distinction illustrated by the example of the light as a medium between 'celestial bodies' and my eyes cannot be taken temporally the way Broad does (cf. fn. above). That is, the causal ties cannot obtain successively, but rather must obtain simultaneously
94. Thus, the two causal ties are temporally interdependent, but not necessarily ontologically or causally
95. One could weaken Kant's model of mutual interaction, so that gaps could arise, but then one would have to maintain the 'strong' interpretation of 'thoroughgoing'. Thus, seen strategically, one's choice is between a strong model of mutual interaction along with a weak interpretation of 'thoroughgoing' and a weak model of mutual interaction along with a strong interpretation of
96. Why does Kant not answer this question in the Third Analogy? I would like to suggest that Kant is not attempting to state sufficient conditions for judging objective coexistence. Rather, he is attempting to establish that mutual interaction is a necessary condition for such judgments. The same goes, mutatis mutandis, for Kant's Second Analogy. (This also explains, I think, why Kant's reply to Hume puzzles so many readers. Their conception of Hume's problem, namely that of specifying sufficient conditions for causal sequences, differs from Kant's conception of Hume's problem and thus also from the nature of Kant's undertaking in the Second and Third Analogies.)
97. See my discussion of this issue in my dissertation (pp. 187-319)
98. One important issue that has been discussed in the literature on the Second Analogy is whether the Second Analogy implies the existence of particular causal laws. Allison argues that Kant's Second Analogy does not require causal laws, whereas Guyer and Friedman argue that the Second Analogy does require causal laws. The same issue must arise for the Third Analogy as well. I maintain that even if Kant's intention was that causal laws are required, he does not provide any argument to back this intention up, perhaps because he did not distinguish clearly between causality and necessity. That is, even if Kant does sometimes seem to say that the Second Analogy requires particular causal laws, I do not see that the argument of either the Second or the Third Analogy establishes such a claim. These arguments are supposed to show that one must apply the category of causality (or mutual interaction) in order to make a judgment of objective succession (or coexistence); Kant is not concerned with the conditions under which one is justified in applying these categories. Rather his concern is to show that these categories are the conditions under which one is justified in making judgments of time-determination. Perhaps some argument could be made in order to establish that the categories of causality and mutual interaction can be applied only under certain conditions (and I think Kant is in a position to provide such arguments as I have indicated in the previous paragraph), but such an argument is simply not present in either of these Analogies
99. I should like to thank Hud Hudson for pressing this point
100. Arthur Schopenhauer, Die Welt als Wille und Vorstellung, V. I, p. 544, from Sämtliche Werke, V. II, Wiesbaden 1972
101. Interestingly, Margaret Morrison ("Space, Time and Reciprocity" in Proceedings of the Eighth International Kant Congress V. II, pp. 187-195 (Marquette University Press, Milwaukee, 1995)) notes the same difficulty, but draws a rather different conclusion. "If A causes B then B must be later than A and if B also causes A then A must be later than B. It is impossible for A to be both later and earlier than B, therefory they must be simultaneous". I conclude that A and B (qua determinate events) cannot be the basic constituents of Kant's causal ties
102. I should like to thank Karl Ameriks, Jeffrey Edwards, Hud Hudson, Lynn Joy, Lorenz Krüger, and an anonymous referee for this journal for helpful comments on earlier versions of this paper or for discussion of its content