Reason, Its Real Use, and the Status of Its Ideas and Principles: Response to Caimi, Gava, and Lewin

1 Introduction

For a long time, the Transcendental Aesthetic and the Analytic were regarded as the central parts of Kant’s Critique of Pure Reason, in contrast to the Transcendental Dialectic and the Doctrine of Method. In particular, the Transcendental Dialectic was viewed mainly as the negative and destructive part of the first Critique, primarily containing the rejection of traditional (pre-critical) metaphysics. The papers in this topical issue of Open Philosophy contribute to a recent dual trend in Kant scholarship which consists in (a) emphasizing the importance of the Transcendental Dialectic within the Critique of Pure Reason and (b) foregrounding the positive and constructive aspects of the Transcendental Dialectic. Some of my own contributions to Kant scholarship have fed into this tendency, particularly my book Kant on the Sources of Metaphysics: The Dialectic of Pure Reason (Cambridge University Press, 2018). Therefore, I am very grateful to the editors of this topical issue, Michael Lewin and Rudolf Meer, for the opportunity to respond to papers in this issue by Mario Caimi, Gabriele Gava, and Michael Lewin, who criticize some of the views I put forward in my book. I am also grateful to the three authors for seriously discussing my views, thereby prompting me to clarify and develop them further. In what follows, rather than commenting on all of the points on which our views differ, I will concentrate on those parts of their discussions where they explicitly criticize my interpretation.

2 Response to Caimi

Mario Caimi’s paper discusses the “real use of reason.”^[1] In the Transcendental Analytic, Kant distinguishes between the logical use and the real use of the understanding, with the twelve logical forms of judgment belonging to the former, for instance, and the twelve categories belonging to the latter. In the Transcendental Dialectic, Kant distinguishes in a parallel way between the logical and the real use of reason, with the three forms of syllogistic inference (categorical, hypothetical, and disjunctive) belonging to the former and the transcendental principles of reason and the ideas of pure reason belonging to the latter: “As in the case of the understanding, there is in the case of reason a merely formal, i.e., logical use, where reason abstracts from all content of cognition, but there is also a real use, since reason itself contains the origin of certain concepts and principles, which it derives neither from the senses nor from the understanding.”^[2] A little later, Kant distinguishes between these two uses of reason by asking: “Can we isolate reason, and is it then a genuine source of concepts and judgments that arise solely from it and thereby refer it to objects; or is reason only a merely subordinate faculty that gives to given cognitions a certain form, called ‘logical’ form?”^[3]

Caimi contrasts two interpretations of the real use of reason: “most interpreters,” including Zöller, Klimmek, and myself, “explain the real use of reason in connection with objects. … According to Willaschek, real use of reason is characterized by its reference to things.”^[4] By contrast, on Caimi’s interpretation

the real use of reason does not reach in any case beyond the generation of ideas and principles. It has no reference to objects. Reference to objects is a further step which is not performed by the real use of reason, but by the product of that use: namely by the ideas. That is why ideas require a deduction in order to establish the legitimacy of their reference to objects.^[5]

That the real use of reason, according to Caimi, does not include a reference to objects may be surprising in light of the fact that Kant explicitly explains the real use of reason, as we just saw, as a potential “source of concepts and judgments that arise solely from it and thereby refer it to objects.”^[6] But Caimi explains: “It is to be noted that this characterization of the real use of reason contains two parts: according to the first, reason is the source of concepts and judgments; according to the second part, reason makes use of those concepts and judgments generated by reason itself to refer to objects.”^[7] He continues: “Our task in the present article consists in referring to the first part of the mentioned passage KrV A305/B362, as well as to the definition of real use in KrV A299/B355, where there is no mention of any claim of validity for real objects.”^[8]

I agree with Caimi that “validity for real objects” is not a part of Kant’s characterization of the real use of reason. Kant merely speaks of its being “related to objects,” which does not mean that this use, as such, is valid for those objects. But if Kant does mention a relation to objects in one explanation of the real use of reason but not in another, why should one ignore or deny that part of Kant’s characterization and merely concentrate on the other part (the source of concepts and judgments)? Moreover, there is another formulation that clearly shows that the real use of reason includes a relation to objects. When Kant explains how the logical use of reason in inferences can give us “guidance” as to its real use, he says that pure reason, although it does not have “immediate reference” to objects, nevertheless “also aims at objects.”^{[9] [10]} Thus, there is a strong textual basis for the claim that the real use of reason includes some relation to objects.

By contrast, Caimi insists that the real use only consists in the production of ideas, not in their application to objects, which is a separate step.

Thus, it is possible to conceive of the production of pure concepts [of the understanding] independently from their application to objects. Application requires a specific deduction. The same exigence appears when considering not just the understanding but reason and its real use: reason in its real use produces concepts (ideas). Objective validity of those concepts (their reference to objects) must be demonstrated through an additional operation, different from the mere production of them.^[11]

But Caimi’s claim that any application of pure concepts requires a deduction needs to be qualified. According to Kant, there is an application of ideas to objects (which means: using them with the intention of representing objects) without a deduction, namely in pre-critical metaphysics, which Kant criticizes in the main parts of the Transcendental Dialectic. For instance, the rational psychologist uses the idea of the soul as a representation of a class of objects, namely immaterial thinking substances. Thus, Caimi’s claim that application requires a deduction is wrong unless re-interpreted as “legitimate application requires a deduction” or perhaps “successful application that leads to cognition requires a deduction.”

What this suggests is that when Caimi criticizes those interpreters who include reference to objects in Kant’s definition of the real use of reason, he assumes that what they mean is that prior to any deduction, there is a legitimate or successful use that reason can make of its concepts by applying them to objects. However, speaking only for myself, this is not my view. On my reading, in its real use, pure reason produces concepts and principles that, unlike in the case of the logical use of reason, do not abstract from their content (from what is being represented) and hence at least purport to represent objects. That is what I mean when I write in my book: “the real use of reason … ‘aims at objects.’^[12] That is, it goes beyond the logical relations between judgments and aims at cognizing objects (in the widest sense of the term). Thus, ‘real’ (from Latin res, thing) here means ‘object-related’ or ‘concerning not just representations but things’.”^[13]

Another way of expressing this point is as follows: if someone makes a real use of their reason, they will typically assume without further reflection that reason’s ideas and principles do represent real objects (that they have objective validity). It is only after reading the Critique of Pure Reason that they can become aware that this assumption was naïve and that, without a deduction, the ideas of reason may not have objective validity. (As it turns out, they do have objective validity when used regulatively, but only indirect and indeterminate objective validity.)

By contrast, Caimi claims that “according to these widely spread interpretations,” which include reference to objects in the conception of the real use of reason, “the real use of reason would be nothing but an illegitimate use of reason.”^[14] Caimi’s picture seems to be as follows: either one includes reference to objects in the very concept of the real use of reason, which means that the real use of reason is illegitimate (because it lacks a deduction), or one does not include reference to objects in one’s conception of the real use, which means that, once a deduction of the concepts of reason is given, the use is legitimate.

But this alternative is incomplete, since there is a third possibility: the real use of reason includes reference to objects; such use can be either illegitimate, namely when the ideas are used constitutively and without a deduction, or legitimate, when they are used regulatively, in which case a deduction of their (indirect and indeterminate) objective validity is possible.

Once this possibility is on the table, I think we can see that Caimi’s interpretation of the real use of reason misses one half of Kant’s conception, namely the illegitimate real use of reason in pre-critical metaphysics that Kant critiques in the Transcendental Dialectic. When, in my book, I present the real use of reason as being related to objects, this is what I mean: prior to the Critique of Pure Reason, metaphysicians made real use of their reason by assuming, falsely, that the ideas of God, the soul, and the world refer unproblematically to objects. The reason for this assumption is that these ideas appear to function just like any other concepts: they contain a combination of conceptual marks in them, which seems to guarantee that they represent all objects with that combination of features. Just as the concept of a human being refers to all objects that are both animals and rational (assuming that “animal” and “rational” are the two marks contained in the concept of a human being), so the concept of a soul is taken to refer to all objects that are substances and immaterial (assuming that these are the marks contained in the concept of the soul). Perhaps even the naïve metaphysician admits that it is unclear whether we can know that there are souls; if there are souls, however, then our concept of a soul refers to them or represents them. Thus, the naïve metaphysician assumes that the concept of the soul (just like the other ideas of reason) has objective validity.

As I understand Kant, this assumption, or at least the tendency to make it, is built into the very concept of the real use of reason. It thus explains why, without a critique of pure reason, the fallacies of traditional metaphysics are unavoidable. It is only once we distinguish between appearances and things in themselves and acknowledge that our concepts have objective validity only with respect to the former that we can become aware that the real use of reason is illegitimate unless properly restricted to being merely regulative.

In conclusion, while I agree with much of what Caimi very knowledgably explains about the real use of reason in Kant, his insistence that this use does not involve reference to objects seems to me to rest on a misunderstanding. The real use of reason as such involves at least a relation to putative objects; that is, representations within the real use of reason are naturally taken to represent real objects. This natural assumption is often unfounded, according to Kant, particularly in the context of pre-critical metaphysics. But in order to understand the natural tendency of reason toward this kind of metaphysics, Kant rightly insists that reason has a real use, that is, a use taken to pertain to objects. Caimi is right that this use of the representations of reason (principles and ideas) is legitimate only once a deduction of their objective validity has been given and that this validity is limited to their regulative, not their constitutive, use. But this seems to me to presuppose, rather than contradict, the claim that the real use of reason as such involves a relation to (real or putative) objects.

3 Response to Gabriele Gava

In his contribution to this volume, Gabriele Gava investigates the attitude that, according to Kant, we take toward ideas and principles of pure reason when using them regulatively.^[15] The ideas and principles in question include the ideas of God, the soul, and the world, the so-called Supreme Principle and the principles of homogeneity, specification, and continuity, which Kant discusses in the Appendix to the Transcendental Dialectic. In particular, Gava asks whether we take these ideas and principles to be true representations, false representations, or neither. (By speaking of “true/false representations,” I mean to cover both ideas and principles.^[16]) Some interpreters (such as Wartenberg and O’Shea) have argued that when using ideas and principles regulatively, we consider them to be true descriptions of nature. In my book, I called this kind of reading “objectivist” and contrasted it with a “fictionalist” reading.^[17] According to the latter, defended for instance by Michelle Grier and Henry Allison, although we can know that, taken as representations of reality, the ideas and principles of reason are false (since nothing in nature can ever fully correspond to them), we can nevertheless use them in an as-if-attitude as heuristic devices.

In my book, I argue against both objectivist and fictionalist readings and for a third, alternative reading. According to this reading, to use a representation regulatively is to use it as a highly general hypothesis of which we do not know whether it is true or false. By generating more specific hypotheses, and by empirically verifying these more specific hypotheses, we gradually confirm the truth of the underlying general hypothesis without ever fully verifying it, since full verification would go beyond all possible experience. For instance, in using the principle of continuity regulatively, we assume the hypothesis that between any two species, there is a mediating species. This allows us to generate more specific hypotheses, such as the one that there is, or has been, a mediating species between humans and chimpanzees. Once we find such a species, this gives further credence to our overarching hypothesis, namely the principle of continuity, which, however, will never be fully verified.

Gava agrees that there is something wrong with the objectivist reading of “regulative use.” He argues against it by insisting, as I did in my book, that a principle’s being transcendental is not the same as its being used constitutively and that a principle’s being logical is not the same as its being used regulatively. The objectivist reading, by contrast, tends to conflate the transcendental/logical and the constitutive/regulative distinctions, since it assumes that transcendental principles of reason as such must be constitutive of nature. As Gava points out, if we keep these distinctions separate, as we should, the objectivist reading loses much of its initial plausibility. I fully agree.

But Gava does not agree with my claim that the fictionalist reading is also mistaken. He denies that all ideas and principles are used regulatively by serving as very general research hypotheses of which we do not know whether they are true or false. Rather, according to Gava, this holds only for some ideas and principles, while for others, the fictionalist reading is correct.

Specifically, Gava names two examples of principles of pure reason that, according to Kant, we must use regulatively but know to be false, namely the Supreme Principle (“when the conditioned is given, then so is the whole series of conditions subordinated one to the other, which is itself unconditioned, also given”^[18]) and the principle of continuity (for every two species, there is an intermediate species^[19]). In addition, Gava points out that in the context of the mathematical antinomies, the idea of the world, considered as the totality of appearances, plays a regulative role, even though we can know that no such totality exists. While my “hypothetical” reading may be correct for other ideas and principles, according to Gava it does not work for these cases, which require a fictionalist reading. Gava therefore concludes that the attempt to find a uniform account of the regulative use of ideas and principles is misguided. Rather, we must acknowledge that there are two irreducibly distinct types of ideas and principles and two distinct ways of using them regulatively.

Moreover, Gava claims that there is a “tension” in my own interpretation, since I seem to acknowledge, at least in the case of the Supreme Principle, that it can be used regulatively, even though we can know that it is false with respect to the world of appearances.^[20] As he quotes from my book: “The only conditioned objects of which we can have cognition, according to Kant, are appearances, not things in themselves. But the constitutive Supreme Principle that we need for the inference to the unconditioned holds only for things in themselves (noumena in the positive sense). Hence, the inference is invalid.”^[21] Thus, while I am officially committed to the hypothetical reading of regulative use in general, in the case of the Supreme Principle I seem to be implicitly assuming a fictionalist reading of its regulative use.^[22] Gava suggests that I avoid an outright contradiction only by thinking of the Supreme Principle, when applied to things in themselves and when applied to appearances, not as two employments of the same principle but as two different versions of that principle,^[23] that is, strictly speaking, as two different principles.

Gava makes a very strong case for his non-uniform reading of regulative use, and I think he is right to point out a tension in my account of the regulative use of principles and ideas. In particular, Gava is right that if the Supreme Principle as such was used regulatively with respect to the empirical world (as I misleadingly suggest in my book), then this would be incompatible with a hypothetical reading of regulative use, since clearly, for Kant, there is no totality of conditions given in the empirical world. But while I do not make this clear enough in my book, I think that there is a way to defend a hypothetical reading of the regulative use of the principles of reason. Such a reading is not possible for the ideas of reason; however, and I never intended it to apply to them.

In the Antinomy chapter, Kant makes it clear that the regulative use of what he there calls “the major premise of the cosmological syllogism,”^[24] “the cosmological principle of totality,” or simply “the principle of reason”^[25] does not require the assumption that there is a complete series of conditions in nature, but merely that in nature there is a condition for everything conditioned, even if we have not yet found that condition:

Now this proposition, which says only that in the empirical regress I can always attain only to a condition that must itself in turn be regarded as empirically conditioned, contains the rule in terminis that however far I may have come in the ascending series, I must always inquire after a higher member of the series, whether or not this member may come to be known to me through experience.^[26]

Similarly, Kant says that the regulative use of the

cosmological principle of totality … retains its genuine validity only in a corrected significance: not indeed as an axiom for thinking the totality in the object as real, but as a problem for the understanding, thus for the subject in initiating and continuing, in accordance with the completeness of the idea, the regress in the series of conditions for a given conditioned.^[27]

As I read these passages, the correction in the “corrected significance” of the principle is twofold: first, in that the principle is valid not as a constitutive “axiom” but as a regulative “problem” or task; and second, in that we do not have to “think” a complete series of conditions as real (as in the Supreme Principle) but must only assume that, for everything conditioned, a regress to its conditions is empirically possible. Kant continues:

Thus the principle of reason is only a rule, prescribing a regress in the series of conditions for given appearances, in which regress it is never allowed to stop with an absolutely unconditioned. … it is a principle of the greatest possible continuation and extension of experience, in accordance with which no empirical boundary would hold as an absolute boundary; thus it is a principle of reason which, as a rule, postulates what should be effected by us in the regress, but does not anticipate what is given in itself in the object prior to any regress.^[28]

Although these passages can perhaps be read in different ways, at least one plausible reading is as follows: the regulative use of the “principle of reason” consists in assuming hypothetically that we will actually find a condition for everything conditioned. Thus, we entertain the following descriptive principle concerning nature: if there is something conditioned, then there exists a condition for it that is empirically accessible. For instance, if there is a change, then it is not only the case that there exists a cause (which is guaranteed by the Second Analogy or principle of causality), but there is a cause that we can empirically discover. This principle can be used constitutively, that is, as a descriptive truth about nature, in which case it is unwarranted, because we have no guarantee that we will find the condition for everything conditioned. Or it can be used regulatively, that is, as a heuristic hypothesis that we do not know to be either true or false. In this case, we can use it to generate more specific hypotheses about specific empirical objects and their conditions, which we can then test empirically. As Kant puts it elsewhere, the regulative use of the cosmological principle consists in the requirement to pursue the regress of conditions and to assume (that is, on my reading: to assume hypothetically) “that there could not fail to be conditions given [that is: given to us in experience] through this regress.”^[29] Read in this way, the regulative use of the Supreme Principle (“in a corrected significance”) does fit my hypothetical reading of regulative use after all.

Note that in the passage Gava cites from my book (Willaschek 2018, 162, see above), I do not say that the Supreme Principle is false when applied to appearances but must nevertheless be used regulatively; rather, I say that the Supreme Principle can be known to be true only for things in themselves (according to Kant), so that from the fact that there are conditioned appearances, it does not follow that the complete series of their conditions exists. In the book, I may not have made it clear enough that the Supreme Principle can be used regulatively only “in a corrected significance.” But I do not think that anything I said was incompatible with the hypothetical reading of the regulative use of that principle.

The same strategy of a “corrected significance” can also be applied to the principle of continuity. Using this principle regulatively does not require us to assume that nature contains an infinity of species, but only that for any two species, no matter how similar they are, we will be able to find a mediating species. That is a hypothesis which we do not know to be either true or false but which can guide empirical research toward finding mediating species.

The situation is different for the ideas of reason because in the Appendix Kant makes clear that the ideas of God, the soul, and the world go beyond anything that can be met with in experience.^[30] Nevertheless, we can use these ideas as foci imaginarii by representing the world as if it were created by God^[31] and as if it contained the totality of conditions for anything conditioned^[32] and by representing the soul as if it were a simple substance.^[33] As Gava rightly points out, in the case of the idea of the world as containing the totality of conditions, this means representing nature in a way we can know to be false. I agree that here, a fictionalist reading of regulative use is correct. But note that in my book, I develop the hypothetical reading only with respect to the principles of reason, such as the Supreme Principle of the principles of homogeneity, continuity, and specification. When, in Chapter 8, I finally turn to the regulative use of the ideas of reason in the second part of the Appendix, I do not claim that the hypothetical reading holds for them as well. Admittedly, I also do not claim that the hypothetical reading does not hold for ideas. Thus, I am very grateful to Gabriele Gava for making me aware of this unclarity in my book and for giving me the opportunity to amend it.

Ideas of reason are conceptual representations of objects that, if they exist, go beyond all possible experience. As concepts, ideas may imply propositions and judgments, as Gava indicates. But they cannot serve as hypotheses, since they are not themselves propositions or judgments. Thus, the hypothetical reading of regulative use does not apply to them, but only to principles of reason.

The distinction between ideas and principles of reason and their respective regulative use is obscured by the fact that Kant himself repeatedly calls the ideas of reason “principles.”^[34] But note that here the English word “principle” is a translation of the German “Prinzip,” not the German “Grundsatz” (as in “oberster Grundsatz der reinen Vernunft,” “supreme principle of pure reason”^[35]). While the term “Prinzip” can refer to objects, events, representations, and propositions,^[36] “Grundsatz” always refers to something with a propositional form, something that can serve as the major premise in an inference – or as a hypothesis. As indicated above, I take the hypothetical reading of regulative use to hold only for the regulative use of “Grundsätze,” not ideas.

Thus, I agree with Gava that no uniform interpretation of regulative use is possible. But I draw the line differently. While Gava thinks that a fictionalist reading is necessary for those ideas and principles that in one way or another present nature as containing a totality of conditions or an infinity of species, but not for the other ideas and principles, I think the hypothetical reading is correct for all propositional principles (“Grundsätze”) that can be used as hypotheses, but not for the ideas of reason, since these do not have propositional form and thus cannot be used as hypotheses. For them, a fictionalist reading seems correct, not only in the case of the idea of the world, but also in the case of the other two ideas: God and the soul.

Against this, one might argue that a fictionalist reading would require that we know that these ideas are “false” representations, while according to Kant we cannot know whether or not God exists and whether or not we have a substantial soul, which implies that we do not know that these ideas are “false” (misrepresent reality). As Kant explains in the Appendix, however, the regulative use of reason is always directed at nature as its object.^[37] When we use the ideas of God and the soul constitutively, we treat them as representations of something that can be found in nature or experience (e.g., evidence of divine design or a substantial soul in human beings). Thus understood, they are “false” representations according to Kant, which is why a fictionalist reading applies even to the ideas of God and the soul.

All in all, while I thank Gava for forcing me to clarify my reading of regulative use and to make clear that the hypothetical reading is not meant to apply to ideas of reason, but only to propositional principles, I think that a uniform hypothetical reading of the latter can be defended against Gava’s perceptive criticism.

4 Response to Lewin

Michael Lewin explores Kant’s conception of reason, narrowly conceived (which he abbreviates as “Rn”).^[38] While Kant sometimes uses the term “reason” in a wide sense to refer to the entire higher cognitive faculty (including the understanding and the power of judgment), in the Transcendental Dialectic Kant is interested in reason specifically as a faculty distinguished from understanding and judgment. While the primary products of the understanding are concepts and those of the power of judgment are judgments, reason (narrowly conceived) produces syllogistic inferences. At least, that is its logical use. To include its possible “real use” (see Section 2), Kant goes on to describe reason, narrowly conceived, as the “faculty of principles.”^[39] According to Lewin, however, Kant’s official explanation of reason as the faculty of principles does not capture what is essential to it. Instead, Lewin offers a ninefold characterization centered around the claim that reason is the faculty of ideas.

It is in the conclusion to his paper, and only in passing, that Lewin objects to my claim that reason, according to Kant, is a discursive faculty.^[40] By contrast, Lewin claims that reason also has an “intuitive use,” namely in mathematics, where, according to Kant, we construct concepts in pure intuition. Lewin asks: “is there a higher concept of reason that is subdivided into an intuitive and discursive faculty … or should Kant have found a different word to denote the former?”^[41]

Lewin thus raises an important question, particularly in the context of understanding Kant’s conception of reason, narrowly conceived (Rn). As I have argued elsewhere (and briefly recapitulate in my book), Kant works with two different, non-equivalent conceptions of reason in the Critique of Pure Reason.^[42] On one conception, prevalent in the Introduction to the first Critique and in the Architectonic, reason is the faculty of a priori cognition,^[43] which therefore includes mathematical cognition. On the other conception, which Kant uses in the Introduction to the Transcendental Dialectic, reason is the faculty of “cognition from principles,” which Kant further describes as “that cognition in which I cognize the particular in the universal through concepts”^[44] and which he distinguishes from mathematical cognition: “I cannot therefore say that in general and in itself I cognize this [mathematical] proposition about straight lines from principles, but only that I cognize it in pure intuition.”^[45] In other words, the first conception of reason is built on the a priori/a posteriori distinction, the other on the distinction between intuition and concepts, or, more generally, between (intuitive) sensibility and (discursive) understanding. These two conceptions of reason do not coincide, because there are a priori cognitions that are not discursive, namely mathematical cognitions.^[46]

We can thus see that Lewin’s question arises from an ambiguity in Kant’s conception of reason. Lewin is right to insist that in mathematics there are rational cognitions that are intuitive, not discursive. There is indeed a “higher concept of reason that is subdivided into an intuitive and discursive faculty,”^[47] namely reason as the faculty of a priori cognition. When I say in my book that reason is a purely discursive faculty, however, I do not want to deny that there are rational cognitions in mathematics that are intuitive. Rather, I explicitly restrict that claim to the conception of reason as Kant discusses it in the Introduction to the Transcendental Dialectic,^[48] where Kant thinks of reason more specifically as the faculty of a priori cognition “from concepts” or “from principles.” But it is that conception of reason that is the topic of Lewin’s paper: reason as distinguished from understanding and judgment, that is, reason in the narrow sense.

In sum, we can thus see that with respect to this narrow conception of reason, it is entirely correct to claim that reason is purely discursive (that is, cognition from concepts). At the same time, Lewin is right that there is a conception of reason that includes intuitive cognition in mathematics. That is not reason in the narrow sense (Rn), however, as distinguished from other sources of a priori cognition such as the understanding, but reason more generally as the faculty of a priori cognition, which Kant addresses in the Introduction to the first Critique and in the Architectonic.

In closing, I would like to again thank Mario Caimi, Gabriele Gava, and Michael Lewin for their careful consideration of the interpretation of the Transcendental Dialectic that I develop in my book. I have profited from all three papers, as well as from the other papers in this topical volume.