Book Search:  


Google full text of our books:

Knowledge, Reason, and Taste:
Kant's Response to Hume
Paul Guyer

Book Description | Endorsements

COPYRIGHT NOTICE: Published by Princeton University Press and copyrighted, © 2008, by Princeton University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. Follow links for Class Use and other Permissions. For more information, send e-mail to

This file is also available in Adobe Acrobat PDF format


IN THE Prolegomena to Any Future Metaphysics that could come forth as Science that he published in 1783 in the hope of both defending and popularizing the Critique of Pure Reason that he had published two years earlier, Immanuel Kant famously wrote “I freely admit that the Erinnerung of David Hume was the very thing that many years ago first interrupted my dogmatic slumber and gave my investigations in the field of speculative philosophy a completely different direction” (Prolegomena, Preface, 4: 260). I have left Erinnerung, which typically means “recollection” or “remembrance,” but could also mean “reminder” in the sense of “admonition,” untranslated in order to sidestep a scholarly debate about chronology—did Kant mean that it was only the recollection of Hume some years after he had first read the early German translations that interrupted his dogmatic slumbers, or did he mean that the admonition of Hume had interrupted his dogmatic slumbers as soon as he read him, many years before writing the Critique of Pure Reason and the Prolegomena?1 What is important for my purposes in this collection of essays is that, however he got there, by the time Kant reached his philosophical maturity at least one way he conceived of his philosophical task was that of putting key principles on more secure foundations than all those earlier ones that Hume had shown to rest not on reason but at best on experience and custom, and at worst on mere dogmatism.

What principles? Kant repeatedly stated that Hume had cast doubt on whether the concept of causation expresses a genuine necessity that is “thought through reason a priori . . . and has an inner truth independent of all experience” (Prolegomena, Preface, 4: 249), and settled instead that it expresses nothing more than our own feeling of necessitation in response to the frequent association or repetition of impressions of external objects. He clearly thought that Hume had raised a genuine problem about the real foundations of the concept of causation and the necessary truth of both the general principle that every event has a cause as well as particular causal laws, and thought that Hume had not sufficiently resolved this problem, although Hume himself may have been content with his solution. But Kant also held that Hume had put his finger on a more general problem without realizing what he had done:

Thus I first investigated whether Hume’s objection could be made general, and I soon found that the concept of the connection of cause and effect is far from being the only one by means of which the understanding thinks the connections of things a priori for itself, rather that metaphysics consists entirely of such concepts. I sought to assure myself of their number, and since this succeeded according to my wish, namely from a single principle, I went on to the deduction of these concepts, from which I was assured that they were not derived from experience, as Hume had worried, but had arisen from the pure understanding. This deduction, which seemed impossible to my acute predecessor, and which no one other than him had ever even thought about, although everyone had confidently used the concepts without asking on what their objective validity is grounded, this, I say, was the most difficult thing that could ever be undertaken in behalf of metaphysics; and what is the worst thing about it is that metaphysics, no matter how much of it is everywhere available, could not give me the least assistance, since this deduction must first establish the possibility of a metaphysics. Now since I was successful in the solution of the Humean problem not just in a special case but with regard to the entire faculty of pure reason, I could thus take sure although always slow steps in order finally to determine the entire scope of pure reason in its boundaries as well as to determine its content completely and in accordance with universal principles, which was then the very thing that metaphysics needed in order to execute its system in accordance with a secure plan. (Prolegomena, Preface, 4: 260–61)

In this statement and others like it (Pure Reason,B 19,A 764– 68/B 792–96; Practical Reason, 5: 13, 5: 50–57), Kant claims that the doubts Hume had raised about the existence of an a priori concept and principle of causation were only an example of the kind of doubts that could be and indeed should be raised about the previously merely dogmatic foundations of all the central concepts of metaphysics, and that none of these concepts and principles could be secure until they had all been given a proper foundation or “deduction” by Kant himself. As Kant rightly points out, Hume had raised no objection to our ordinary use of causal concepts and beliefs and, by implication, our ordinary use of the other concepts and beliefs that are in the same boat, nor did Kant himself think that scientists in their laboratories or craftsmen in their shops have to suspend all their activities until their key concepts had been put on a sound footing. But in his view Hume had without realizing it raised a challenge for all of metaphysics that had to be answered before philosophy could proceed. Moreover, Kant held that unless the metaphysical concepts at stake were both properly founded and properly limited, that is, restricted to the properly demarcated sphere of human experience, perfectly reasonable doubts about their cognitive value beyond this sphere could end up undermining our confidence in their use within this sphere, and thus cast doubt about our use of these concepts for the purposes of ordinary cognition and ordinary science after all. Thus, in a crucial passage in the Critique of Pure Reason, Kant writes about Hume, whom he calls “perhaps the most ingenious of all skeptics” (A 764/B 792), that:

The skeptical aberrations of this otherwise extremely acute man, however, arose primarily from a failing that he had in common with all dogmatists, namely, that he did not systematically survey all the kinds of a priori synthesis of the understanding . . . had he done so . . . He would thereby have been able to mark out determinate boundaries for the understanding that expands itself a priori and for pure reason. But since he merely limits our understanding without drawing boundaries for it, and brings about a general distrust but no determinate knowledge of the ignorance that is unavoidable for us, by censuring certain principles of the understanding without placing this understanding in regard to its entire capacity on the scales of critique, and, while rightly denying to understanding what it really cannot accomplish, goes further, and disputes all its capacity to extend itself a priori without having assessed this entire capacity, the same thing happens to him that always brings down skepticism, namely, he is himself doubted, for his objections rest only on facta, which are contingent, but not on principles that could effect a necessary renunciation of the right to dogmatic assertions. (A 767–68/ B 795–96)

Kant recognized that Hume had been satisfied with his own explanation of key concepts and principles as resting on “nothing but a custom arising from its experience and its law,” and thus as “merely empirical, i.e., intrinsically contingent rules, to which we ascribe a supposed necessity and universality” (A 765/B 793), and he recognized that Hume had used his empirical derivation of such concepts and principles to argue that we could not apply them to objects of which we have no experience. Thus, Kant recognized that it was the intended payoff of Hume’s philosophy that we cannot use our empirically grounded principle that every event has a cause to infer that the whole world has a unique cause of a sort that we have never directly experienced, namely God.2 However, Kant also believed that without a clear demarcation between the realm of human experience within which the principle of causation and all the other fundamental principles of human thought apply and the realm beyond experience where we can doubt that those principles apply, there is no barrier to prevent our skepticism about the validity of those principles in the latter realm from splashing back and undermining our confidence in the validity of those principles in the former realm, and with no foundation for those principles within the former sphere but mere “custom and experience,” Hume would have no way to resist this gastric reflux of doubt. Thus, although Hume advocated only “mitigated scepticism” as “the limitation of our enquiries to such subjects as are best adapted to the narrow capacity of human understanding” (Enquiry I, Section 12, Part 3, p. 120), Kant nevertheless felt justified in calling him the “most ingenious of all skeptics.”

As we will see in chapter 2, Hume had raised a more general challenge to our most fundamental concepts and principles than Kant seems to have realized. Kant’s acquaintance with Hume’s works during the crucial years of his own philosophical development was incomplete: Hume’s Enquiry (originally Philosophical Essays) concerning Human Understanding, first published in 1748, was translated into German in 1755,3 and Kant is known to have owned this early translation of the first Enquiry at the time of his death and reasonably presumed to have read it much earlier in his life,4 in all likelihood very soon after it came out. But under the rubric of “Sceptical Doubts concerning the Operations of the Understanding,” Section 4 of the first Enquiry focuses almost exclusively on Hume’s worries about causation, and under the title of “Sceptical Solution of these Doubts”—a title to which Kant could certainly have appealed for his own characterization of Hume as a skeptic—the first Enquiry provides only Hume’s empirical account of our belief in causation. Hume’s original Treatise of Human Nature, by contrast, very clearly raises doubts not only about our concept of and beliefs about causation but also about our concepts of and beliefs in external objects and an enduring self, and moreover notoriously regards those concepts and beliefs as much more problematic than the concept of and belief in causation, with his account of which Hume was entirely satisfied. However, the Treatise was not translated into German in its entirety until long after Kant had completed his work on the Critique of Pure Reason.Atthe time that he wrote the Critique, Kant is thought to have had anything approximating firsthand knowledge of the Treatise only through Johann Georg Hamann’s translation of Book I, Part IV, Section 7 of the Treatise, published in a Königsberg newspaper in 1771.5 In this section, Hume gives his famous argument that while “reason is incapable of dispelling” the “clouds” of skeptical doubts, “nature herself,” in the form of riverside walks and nice evenings of dinner, backgammon, and conversation, “suffices to that purpose,” but he does not restate the particular skeptical doubts about self and object as well as about causation that he had earlier raised.6 Nevertheless, I will propose, the philosophical approach Kant developed for showing that our concept of and beliefs about causation have a foundation that Hume denied they have also provides Kant with an approach for addressing the concerns Hume raised about external objects and the self—so even though Kant did not know that Hume had generalized his skeptical doubts about causation as Kant thought he should have, the general approach to grounding metaphysical concepts and principles Kant developed in response to Hume’s worries about causations does address the other problems that Hume himself had raised. Thus Kant was wrong to think that Hume had not generalized his problem about causation, but right to think that he himself had developed a general method for addressing the generalization of Hume’s problem.

Beyond showing that Kant did indeed generalize Hume’s problem about causation and was stirred by his Erinnerung of that problem to develop a general foundation for other theoretical concepts such as those of self and object, I will also suggest that much of Kant’s philosophy beyond theoretical metaphysics can be read as a response to Hume, specifically that important elements of Kant’s moral philosophy, his aesthetics, and his teleology can also be fruitfully read as responses to Hume. By saying this I by no means intend to say that in all these other parts of his philosophy Kant exclusively or even foremost intended to respond to Hume, any more than I mean to suggest that Kant was concerned with Hume alone in this theoretical philosophy. While he was not a learned historian of philosophy, Kant was broadly acquainted not only with the German philosophy of his own century but also with a vast array of European philosophy, science, and thought of both the seventeenth and eighteenth centuries. Thus, his interlocutors and targets in theoretical philosophy include Descartes, Leibniz, Wolff, Baumgarten, Crusius, and Mendelssohn as well as Locke, Berkeley, and Hume; his targets in moral and political philosophy include ancient Stoics and Epicureans, Wolff and Baumgarten, and also Montaigne, Hobbes, Mandeville, and Shaftesbury, Hutcheson, and Adam Smith as well as Hume (see e.g., Practical Reason, 5: 40); his targets in aesthetics include Baumgarten, Meier, Mendelssohn, Lessing, and Herder as well as Burke, Gerard, and Hume; and his targets in teleology include Spinoza, Wolff, and Herder as well as Hume—so it would always be a grievous error to reduce Kant’s targets in any area of his work to a single figure, no matter how important. Kant’s career-long focus on causation, for example, has to be understood as a response to debates within German rationalism that began with Leibniz’s claim that genuine substances are “windowless monads” that cannot actually cause changes in each other but merely represent changes in each other because of God’s beneficent selection of a coherent set of actual monads from among all those possible.7 Nevertheless, Kant does mention the name of Hume not only at crucial moments in his theoretical philosophy but also in his moral philosophy (again, Practical Reason, 5: 13–14, 50–56), his aesthetics (Judgment,§34), and his teleology (Judgment,§80), and I will propose that quite apart from any debate about the historical influence of Hume on Kant or Kant’s intentions to respond to Hume, it is nevertheless illuminating to think about the ways in which these parts of Kant’s philosophy can also be considered as responses to challenges that Hume raised. Just as in the case of theoretical philosophy, where Kant by no means rejected, indeed endorsed Hume’s project of criticizing the use of our fundamental concepts and principles in dogmatic metaphysics while nevertheless holding that these concepts and principles required a more secure foundation than Hume had given to them, Kant’s relations to Hume in moral philosophy, aesthetics, and teleology are also complex. In the case of moral philosophy, the difference between a philosopher who held that the use of reason is never more than merely instrumental to the realization of goals set entirely by sentiment and one who held that the fundamental principle of morality must be founded in pure reason is obvious, but I will argue that there are also important affinities between Hume’s and Kant’s models of motivation and their uses of these models in their opposed moral theories. In the case of aesthetics, I will suggest that Hume profoundly influenced Kant’s conception of the problem of taste, although once again Kant strives for an a priori rather than merely empirical foundation for our claims to agreement in judgments of taste. In the case of teleology, I will argue that Kant fully endorses Hume’s criticism of the constitutive or dogmatic use of teleological principles within both natural science and teleology, but also may well derive his conception of the heuristic use of teleology within our investigation of nature from Hume, while he at the same time argues that our naturally teleological conception of nature itself has a use in morality, specifically in moral theology, that Hume entirely failed to recognize. In all these cases, I suggest, reading Kant’s philosophy as a response to Hume is a way to elucidate, through both their similarities and their differences, some of Kant’s deepest philosophical assumptions and ambitions.


Having stated my intention to look at Kant’s treatment of causation as a response to Hume, at his theoretical philosophy as a whole as a response to further problems about objects and the self that Hume had raised even though Kant did not know that, and even at further domains of Kant’s philosophy as if they were responses to Hume, I should say something here about some recent arguments against overemphasizing Kant’s intentions to respond to Hume in his theoretical philosophy. (Neither of the arguments I am about to discuss go beyond Kant’s theoretical philosophy.)

Gary Hatfield has argued that in his few references to Hume in the first edition of the Critique of Pure Reason, Kant actually enlisted him as an ally in his own critique of traditional metaphysics, although he also held that Hume did not have a systematic criterion or method for distinguishing between illegitimate pretensions and legitimate claims of reason,8 and therefore did not just suggest a useful “skeptical method” for scrutinizing dogmatic metaphysics but risked lapsing into actual skepticism. According to Hatfield, in the first edition of the Critique “Kant did not treat Hume’s account of causation as a genuine threat to natural science or ordinary knowledge,” but saw “Hume’s own attempts as directed primarily against the pretensions of reason,” although he also saw “Hume’s failure” as lying “in not curbing those pretensions once and for all.”9 Correspondingly, Hatfield sees Kant’s own aim in the first edition of the Critique not so much as demonstrating the reality of synthetic a priori knowledge and principles in the face of Hume’s doubts about that, but instead as providing “an explanation of the possibility of synthetic a priori cognition in order to be able to assess its possibility and impossibility in various metaphysical domains,”10 thereby making precise and conclusive the criticisms of traditional metaphysics that Hume had left indeterminate and therefore possibly too broad. As Hatfield sums up, “Given that Kant does not mention any skeptical threat to ordinary experience, mathematics, or natural science in the A Critique, that he evaluates the skeptical method positively, that he describes the one crucial function of the Deduction and Analytic of Principles as preparatory to limiting the understanding to experience, and that he singles out the possibility (or impossibility) of transcendental metaphysics as his main quarry, there seems little basis for arguing that instead Kant was really out to refute the skeptic and save ordinary knowledge.”11 Hatfield then continues that in the Prolegomena Kant “did take a new interest in Hume and skeptical idealism” in response to the charge of the notorious “Garve-Feder” review of the Critique that there was no difference between Kant’s “transcendental idealism” and the idealism of Berkeley and Hamann’s comparison of Kant’s skeptical conclusions about metaphysics to those of Hume, but that “a close reading of the Preface” to the Prolegomena “reveals that he did not present Hume’s problem or Hume’s doubt as a challenge to anything but metaphysics. He certainly did not portray Hume as presenting a skeptical challenge to natural science or to ordinary knowledge of objects.”12 Only in the second edition of the Critique (1787), Hatfield concludes, do Kant’s added remarks about Hume show that “He related Humean skepticism to ordinary experience; [that] he suggested that his Deduction avoids skepticism”; and did he intend “the latter claim to cover Humean skepticism.”13 But, Hatfield also suggests, Kant’s new emphasis on refuting Hume’s skepticism about our ordinary and scientific knowledge of causation rather than refining Hume’s skeptical method in the critique of traditional, dogmatic metaphysics, is a response to external factors, an increased interest in skepticism and charges that Kant’s own transcendental idealism leads to skepticism, in the later half of the 1780S, rather than an accurate indication of Kant’s original concerns in the Critique of Pure Reason. So, Hatfield concludes, for a historically reliable interpretation of Kant, we should not read the Critique as really intended to refute skepticism about causation in ordinary knowledge and natural science, let alone skepticism about other fundamental concepts, but should emphasize Kant’s original intention to refine Hume’s skeptical method for his own critique of metaphysics.

This approach to Kant’s relation to Hume is misleading in two regards. First, it is true that Kant recognized that Hume did not mean to argue against our ordinary practices of making causal judgments in science and everyday life—this is why he says that Thomas Reid and other Scottish commonsense philosophers “missed the point of [Hume’s] problem . . . proving with great vehemence and, more often than not, with great insolence exactly what it had never entered his mind to doubt” (Prolegomena, Preface, 4: 258). It is also true that Kant held that “No critique of reason in its empirical use was needed, since its principles were subjected to a continuous examination on the touchstone of experience; [and] it was likewise unnecessary in mathematics, whose concepts must immediately be exhibited in concreto in pure intuition, through which anything unfounded and arbitrary instantly becomes obvious,” and that a “discipline” of reason is instead necessary “to constrain its propensity to expansion beyond the narrow boundaries of possible experience” “where neither empirical nor pure intuition keeps reason in a visible track” (Pure Reason,A 710–11/B 239–40). Kant also, especially in the Prolegomena and the Preface to the second edition of the Critique of Pure Reason (B x–xiv), appealed to the secure progress of mathematics and science as a model for what needs to be done in metaphysics. Nevertheless, Kant also thought that in the absence of a precise way of demarcating the domain of ordinary experience and normal science where our ordinary practices of causal inference are reliable from the disputable domain of metaphysics where they are not, Hume had no way of preventing his skeptical doubts about the metaphysical use of causal inference from undermining our ordinary use of causal inference. Thus, in Kant’s view, Hume was inevitably led into skepticism about the concept and principle of causation in ordinary life and natural science even though he had no intention of being skeptical in those domains. Second, it seems misleading to separate Kant’s positive project of grounding the first principles of human thought, including mathematical and scientific thought, from his negative project of eliminating metaphysical dispute by confining those principles to possible experience. To be sure, Kant sometimes stressed the negative rather than the positive side of his project, especially after the first edition of the “Transcendental Deduction of the Pure Concepts of the Understanding,” the part of the book that he said in the Preface to the first edition of the Critique was of the utmost importance “for getting to the bottom of that faculty we call the understanding, and at the same time for the determination of the rules and boundaries of its use” (A xvi, emphasis added), met with rejection from its first readers.14 This is particularly so in the Preface to the Metaphysical Foundations of Natural Science of 1786, where Kant says that “if we can prove that the categories which reason must use in all its cognition can have no other use at all, except solely in relation to objects of possible experience . . . then, although the answer to the question how the categories make such experience possible is important enough for completing the deduction where possible, with respect to the principle end of the system, namely, the determination of the limits of pure reason, it is in no way compulsory, but merely meritorious.”15 Yet this seems to have been somewhat of an overstatement on Kant’s part: it seems clear that in his original conception of the Critique the positive task of establishing the validity of the a priori categories of the understanding and the synthetic a priori principles of judgment (including the concept and principle of causation) within possible experience was every bit as important to him as proving that these concepts and principles cannot yield cognition beyond the limits of possible experience, and that from the outset he considered the former just as much a critical rebuttal of Hume as he considered the latter a critical refinement of Hume. This is evident not just from the nearly equal sizes of the constructive “Transcendental Aesthetic” and “Transcendental Analytic” and the destructive “Transcendental Dialectic” (in the first edition, 273 and 348 pages respectively), but also from Kant’s programmatic statements in the first edition of the Critique. In the Critique’s most extended discussion of Hume, in the “Doctrine of Method” where he is explaining the significance of his position in the history of philosophy, Kant first stresses the inadequacy of Hume’s account of causation and then stresses that because of the further inadequacy of Hume’s boundary between possible experience and what lies beyond it he has no way of preventing his doubts about the use of causation—and other a priori concepts that he should have considered, such as persistence—in metaphysics from casting doubt upon their use everywhere. Kant says that “the most ingenious of all skeptics” raised his doubts about causation because he recognized, although he did not name, the problem of synthetic a priori cognition, but did not know how to solve it:

Hume perhaps had it in mind, although he never fully developed it, that in judgments of a certain kind we go beyond our concept of the object. I have called this sort of judgment synthetic. There is no difficulty about how, by means of experience, I can go beyond the concepts that I possess thus far. Experience itself is a synthesis of perceptions that augments my concept which I have by means of one perception by the addition of others. But we also believe ourselves to be able to go beyond our concepts a priori and to amplify our cognition. We attempt to do this either through pure understanding, with regard to that which can at least be an object of experience, or even through pure reason, with regard to such properties of things, or even with regard to the existence of such objects, that can never come forth in experience. Our skeptic did not distinguish these two kinds of judgments, as he should have, and for that reason held this augmentation of concepts out of themselves and the parthenogenesis, so to speak, of our understanding (together with reason), without impregnation by experience, to be impossible; thus he held all of its supposedly a priori principles to be merely imagined, and found that they are nothing but a custom arising from experience and its laws, thus are merely empirical, i.e., intrinsically contingent rules, to which we ascribe a supposed necessity and universality . . . he made a principle of affinity, which has its seat in the understanding and asserts necessary connection, into a rule of association, which is found merely in the imitative imagination and which can present only contingent combinations, not objective ones at all. (A 764– 66/B 792–94)

Kant follows this passage with the paragraph I have already quoted in which he says that Hume has thereby created a “general distrust” of the understanding and “dispute[d] all its capacity to extend itself a priori,” not just its capacity to extend itself to objects beyond the limits of experience (A 767/B 795). I see no way to interpret this extended discussion of Hume, present in its entirety in the first edition, except as stating Kant’s belief that Hume’s inadequate foundation for the concept and principle of causation and, by implication, other key categories of thought, could only lead to skepticism about all of the uses of those concepts and principles, and thus that he faced the dual task of rebutting the skepticism to which Hume was led, whether he liked it or not, within the realm of experience, thus the realms of ordinary life and normal science, as well as that of preserving Hume’s skepticism about the metaphysical or transcendent use of these concepts and principles by determining the boundary between possible experience and what lies beyond in a way that Hume had not. As Kant understood his situation, he had to rebut what he saw as the general skepticism implied by Hume’s approach before he could refine Hume’s skepticism about traditional metaphysics. In the terms I will suggest in chapter 1, Kant shared with Hume the project of eliminating Pyrrhonian skepticism arising from conflicting metaphysical dogmas about what lies beyond the bounds of experience, but he did not think that this project could be successful unless what he saw as Hume’s own skepticism—Humean skepticism—about the use of first principles even within the realm of possible experience was first rebutted.

Kant also made it clear in the Prolegomena that he intended his defense of the principle of causation in the second Analogy of Experience, as already presented in the first edition of the Critique just two years earlier, to be aimed specifically at Hume. In the Prolegomena, Kant does not restate the details of the Analogies of Experience, saying rather that, “For the most part the reader must be attentive to the method of the proof of the principles that appear under the name of the Analogies of Experience” (Prolegomena,§26, 4: 309). He then says:

Here is now the place to remove the ground of the Humean doubt. He rightly asserted that we in no way have insight into the possibility of causality, i.e., of the relation of the existence of one thing to the existence of something else, through which the former is necessarily posted, through reason. I add to this that we have just as little insight into the concept of subsistence, i.e., of the necessity that a subject that cannot itself be the predicate of any other thing should underlie the existence of things, indeed that we cannot even form the concept of the possibility of such a thing (although we can point out examples of its use in experience), and that this same incomprehensibility also affects the community of things, in that we have no insight how from the state of one thing an inference to the state of entirely different things outside of it and vice versa can be drawn, and how substances, each of which has its own, separate existence, can depend upon one another, indeed necessarily. Nevertheless I am far from holding these concepts to be merely borrowed from experience and the necessity that is represented in them to be invented and a mere illusion created for us by long experience; rather I have sufficiently shown that they and the principles from them stand firm a priori prior to all experience and have their indubitable objective correctness, although to be sure only with regard to experience. (§27, 4: 310–11)

This is an explicit commentary on the first edition of the Critique of Pure Reason. Kant thus asserts in 1783 that his constructive theory of causality and the other fundamental concepts of experience had already been aimed against Hume in 1781, although that does not mean that it did not have other targets and sources as well. It would seem very strange not to take this statement written so soon after the original edition of the Critique as a sincere statement of Kant’s original intentions in that work. So in spite of the fact that Kant did not mention Hume’s name in the immediate proximity of the Analogies of Experience, he tells us what is in fact a very short time later, with no indication that he had undergone any major change of heart, that they were directly aimed against Hume.16 It seems more reasonable to suppose that in the Critique he had thought this so obvious that it did not need to be mentioned than that he was revising his intentions for the Critique so soon after it had been published.

Another revisionist who downplays the centrality of refuting Hume among Kant’s motivations in the Critique of Pure Reason is Eric Watkins. The focus of Watkins’s concern is indicated by the title of his 2005 book Kant and the Metaphysics of Causality.17 Watkins argues against seeing Kant’s treatment of causality as intended to refute Hume’s position on causation for both external and internal reasons. The external reasons are that there was already a well-developed debate about the nature and reality of causality within the German rationalist tradition, going back to Leibniz’s provocative position that causal relations are merely the appearance of parallel successions of states in different monads whose histories are determined solely by their internal principles, with which Kant had been arguing from very early in his career, and moreover that those in Germany who took Hume seriously, such as Johann Georg Sulzer and Johann Nicolaus Tetens, had not taken Hume’s arguments about causality seriously, so there was no reason why Kant should have either. But it hardly follows from these facts that Kant could not have realized that Hume raised more serious concerns about causality than had been raised by Leibniz’s fanciful monadology, or that he, even alone among his countrymen, could have realized that Hume’s concerns required a far more powerful and general solution than Sulzer or Tetens had offered. That Kant did exactly that is why, after all, he is remembered as a far greater philosopher than Sulzer or Tetens.

Watkins has three internal reasons for denying that Kant was out to refute Hume. First, he holds that for Kant causality is a relation between enduring objects with active and passive powers rather than between mere events, as he takes Hume to have held. Second, he holds that Kant thought of the change from one state of an object to another that is the effect of the agency of a cause as continuous rather than as an instantaneous succession, as he thinks Hume had held. And finally, he holds that Kant was not out to refute Hume because he did not construct an argument for a conclusion that Hume rejected from premises that Hume accepted, so that he instead aimed only to replace Hume’s treatment of causation with an altogether different approach.18 But Watkins’s claims are problematic. First, Hume often speaks of causes and effects as objects rather than events, especially in the Enquiry concerning Human Understanding, Kant’s source for Hume’s views. There, for example, Hume represents the fundamental problem about causation as explaining how to get from the proposition “I have found that such an object has always been attended with such an effect” to the proposition “that other objects, which are, in appearance, similar, will be atcatended with similar effects”(Enquiry I, Section 4, Part 2,p. 30). Conversely, since for Kant the role of causation is to allow us to determine “the position of the appearance . . . in time,” that is, to determine that one state of affairs has, for example, succeeded rather than preceded another, the invocation of a cause must tell us “that in what precedes, the condition is to be encountered under which the occurrence always (i.e., necessarily) follows” (Pure Reason,A 201/B 246). Simply appealing to a certain kind of agent as a cause will not do this; only appealing to the state of an agent at a certain time, or more precisely to a particular change in the agent, that is, to an agent’s having come to be in a certain state at a certain time, will—and that is to appeal to an event, the event of an object’s having come to be in a certain state at a certain time, as the cause. Upon analysis, Hume and Kant both have very much the same conception of a cause, that of an object’s being in a certain state at a certain time, which is taken to be the condition of another object’s being in a certain state at a certain time. Second, Watkins’s claim that Hume and Kant had different and incommensurable conceptions of causation because Hume conceives of an effect as an immediate succession of one state of affairs upon another while Kant conceives of causation as continuous change is misleading. Kant did recognize that the changes we explain causally are often continuous rather than punctiform, and indeed that the effect often appears to be simultaneous with the cause rather than successive to it; but he was so wedded to the Hu-mean model of causation as a necessary succession of one state of an object upon another state of that object triggered by the intervention of a second object at a determinate moment in time, that he went out of his way to explain that even where the effect seems to be simultaneous with the cause there has to be a vanishingly small temporal gap between them so that the effect really is successive on the cause (A 202–3/B 248). He would not have argued this had it not been his larger intention to argue that the very same sort of causation that Hume thought was merely contingent was in fact necessary.

Finally, Watkins’s position depends on what is, to say the least, a debatably narrow conception of refutation. As the long quotation from the “Doctrine of Method” has shown, Kant certainly thought of his own treatment of causation as a critical response to Hume, and as chapters 2 and 3 will argue, this response takes the form of explaining that Hume could not explain a cognitive ability that he himself took for granted on the basis of his own view that causal concepts and beliefs are not in any way a priori but are acquired entirely in the course of experience, and never attain genuine necessity. Specifically, Kant will argue that the possibility of our determining of the objective order of states of affairs, or even, ultimately, of our own experiences, a cognitive ability that Hume took for granted as much as anyone else, could not be explained or justified if the reality of causation were doubted or if our concepts of it were derived only from an antecedent experience of successive states of affairs. On my account, Kant and Hume do share a premise, namely that we are capable of determining the temporal order of states of affairs, including at least our own experiences, and the form of Kant’s argument is to show that this premise cannot be held consistently with the rest of Hume’s claims, but only on the basis of Kant’s other claims. This would seem to be entitled to the title of “refutation,” unless refutation is to be construed so narrowly that the only thing that counts as one is showing that an opponent holds all the same premises that the refuter holds but has somehow mistaken what follows from those premises. That, I think, is not consistent with the ordinary usage of the term. But however one labels Kant’s argument about causation, Watkins is surely right to stress that it is by no means aimed just at Hume, yet just as surely wrong to claim that it is not aimed at rebutting Hume at all. Kant’s own account of the goal of the Second Analogy in the Prolegomena should put that beyond doubt.


A comprehensive treatment of Kant’s philosophy as a response to Hume would require a detailed investigation of all the sources for Kant’s acquaintance with Hume’s works as well as a thorough discussion of both philosophers on epistemology, the critique of metaphysics, practical philosophy, aesthetics, teleology, and philosophy of religion, the latter including their criticisms of the a priori ontological and cosmological arguments and the empirical argument from design but also of Kant’s defense of an “ethicotheology” (Judgment,§86) against Hume’s rejection of all philosophical theology whatsoever.19 The chapters of this volume, originally written over a period of six years for a variety of different occasions, have been revised to go together (and to take account of some recent work by myself and others), but will not be that comprehensive or systematic. But I do hope that they demonstrate the fruitfulness of thinking of many aspects of Kant’s philosophy, not just his treatment of causation, as a response to Hume, although again not as a response to Hume alone.

Chapter 1, “Common Sense and the Varieties of Skepticism,” originally written for a conference on skepticism at the Instituto de Investigaciones Filosóficas of the National Autonomous University of Mexico in 2001, argues first for the complexity of Kant’s conception of skepticism, in which Pyrrhonian or dialectical skepticism, Humean skepticism about first principles, and Cartesian skepticism about external objects are distinguished, and then argues that refuting the first two forms of skepticism was central to Kant’s aims in both his theoretical and practical philosophy, while refuting Cartesian skepticism about external objects was only an afterthought and subsidiary theme in his theoretical philosophy.20

Chapter 2, “Causation,” written for a special issue of Philosophical Topics on early modern philosophy in 2003, distinguishes the three different questions about causation that Hume raised in the Treatise, namely, about the origin of our concept of necessary connection, about the basis for our belief in particular causal laws, and about the basis for our belief in the general principle that every event has a cause; it then examines Kant’s strategy for answering these different questions, but concludes that although Kant had a plausible account of the origin of our idea of causation itself and a persuasive account of the indispensability of causal belief for the most basic forms of our empirical knowledge, he did not directly reply to Hume’s worries about the rational foundation for our claim to know particular causal laws on the basis of induction, even though these worries are particularly prominent in the Enquiry concerning Human Understanding, the work of Hume’s which, unlike the Treatise, Kant did know firsthand.

Chapter 3, “Cause, Object, and Self” written in 2004 for a festschrift for Vere Chappell, takes up the theme of Kant’s generalization of Hume’s problem about causation and of his own response to it, and shows how the general strategy Kant employs in his treatment of causation also provides an approach to Hume’s concerns about our knowledge of both a continuing self distinct from its fleeting impressions of its states and continuing external objects distinct from our fleeting impressions of their states even though Kant remained unaware that in the Treatise Hume had raised questions about self and object similar to his questions about causation.

Chapter 4, “Reason, Desire, and Action,” written at the invitation of Rachel Cohon and Lorne Falkenstein as a plenary address for the Thirty-Third International Hume Congress in 2006, addresses relations between the moral philosophies of Hume and Kant. However, it does not focus on the obvious contrasts between these two approaches to moral philosophy, namely those between Hume’s insistence that reason cannot set ends but merely discovers means to ends set by sentiment, and Kant’s insistence that pure practical reason is the source of the end in itself that underlies all moral imperatives. Rather, this chapter brings out affinities between the moral psychologies of Hume and Kant, specifically in the details of their models of moral motivation and in their conceptions of the psychological results of moral conduct. The point of this approach is to show that Kant has a more complex model of the motivation of human action than is usually supposed, and to intimate that there is much in this model that remains worthy of consideration even if his transcendental idealist insistence that we are always free to act in accordance with the moral dictates of pure practical reason in spite of all empirical circumstances cannot be accepted.

Finally, chapter 5, “Systematicity, Taste, and Purpose,” written for the Cambridge Companion to Kant and Modern Philosophy published in 2006, shows how Kant’s treatments of the systematicity of science, of judgments of taste, and of teleology in the Critique of the Power of Judgment can all usefully be read as responses to Hume’s treatments of natural law, taste, and the argument from design.

Return to Book Description

File created: 11/28/2007

Questions and comments to:
Princeton University Press

New Book E-mails
New In Print
PUP Blog
Princeton APPS
Sample Chapters
Princeton Legacy Library
Exam/Desk Copy
Recent Awards
Princeton Shorts
Freshman Reading
PUP Europe
About Us
Contact Us
PUP Home

Bookmark and Share