For a Dialectic-First Approach to Kant’s Critique of Pure Reason

1 Unpopularity of, and problem for, a Dialectic-first approach

Say we want to take a direct look at the broadest features of the Dialectic, without the influence of assumptions from other sections in the book, or anywhere else. If we do so, we would find not just the well-known ambition of rejecting rationalist special metaphysics. We would find two strands of thought: One more positive and one more negative, or one more constructive and one more destructive.^[10]

The claims of the positive strand are clustered around the idea that the faculty of reason itself takes an ineliminable interest in unconditioned grounds – and that principles of reason, concerning the unconditioned, play an irreplaceable guiding role in all theoretical inquiry (e.g., A307/B365). One place in which these claims seem very positive is this: Kant mentions a “logical maxim” requiring the pursuit, in the logical use of reason, of cognitions that are the conditions of any conditioned cognitions. He continues on in a manner that, at face value, suggests seeing guidance by this prescription as requiring that “one assumes” (annimmt) a “supreme principle” of reason:

SP: “…when the conditioned is given, then so is the whole series of conditions subordinated one to the other, which is itself unconditioned, also given (i.e., contained in the object and its connection)” (A307–8/B364).

I have argued elsewhere that SP is a form of the rationalists’ principle of sufficient reason (PSR); I explain the connection in this way:

for anything that is not a sufficient reason for itself, or for anything conditioned, there must be a complete series of conditions that provides for it a sufficient reason.^[11]

Meanwhile, the claims of the negative strand are clustered around the idea that this guiding interest of reason leaves us subject to “a natural and unavoidable illusion” (A298/B354). Deception by the illusion involves mistaking the status of reason’s principles about the unconditioned (A297/B353). The Dialectic is concerned with critique in the sense of “critique of dialectical illusion” (A63/B86). The point of this critique is not to remove the illusion – which is, again, “unavoidable” – but to learn to avoid being deceived by it:

The transcendental dialectic will … content itself with uncovering the illusion in transcendental judgments, while at the same time protecting us from being deceived by it. (A297/B354)

And the Dialectic makes the case that the arguments of rationalist special metaphysics, where these conclude with the existence of some form of unconditioned ground – as in, for example, a cosmological argument for the existence of God – are paradigm cases of the kind of deception to be critiqued. But other ambitious metaphysical arguments are also forms of deception as well, including arguments concluding with a denial of the existence of all unconditioned grounds, including God and an unconditioned free will (e.g., A471/B499).

Again, a Dialectic-first approach would hold that the proper place to begin, in understanding the project of the Critique, is with reason’s irremovable interest in the unconditioned, the unavoidable illusion, and its critique. And, again, I think that none of the many established, familiar, and agenda-setting approaches to the book are really like that. For example, it has become popular to see Kant’s project as more about metaphysical aims – or, sometimes, about the defense of transcendental idealism, where this is nowadays supposed to include metaphysical aims: Developing a metaphysics of properties, arguing that there are things in themselves in the sense that things have some (otherwise unknowable for us) absolutely intrinsic properties (or non-relational, in various debated senses of these terms).^[12] And it is also now familiar and popular to approach the Critique via the idea of a fundamental focus on our capacities and how they fit together or stem from a common root.^[13] Both approaches are different than Strawson’s. But they are no more oriented around the idea of an unavoidable illusion of reason. They may have something to say about the Dialectic – just as Strawson does – but what they say is not that its ideas about the unconditioned, illusion, and critique of illusion are the place to begin to understand what unifies the project of the Critique.

Now, one problem, if we want to face the Dialectic directly, without interpretive guidance from assumptions about the project rooted elsewhere, would be to explain how the positive and negative strands are even consistent: One seems positive about, for example, the kind of principle powering rationalist metaphysics; the other negative about the same. But I think recent work already suggests optimism here. For example, Proops defends consistency by arguing that the very transition to the non-methodological formulation of the supreme principle is supposed to be a case of deception by reason (Fiery, 1.3). But Willaschek argues that we can and should see Kant endorsing the transition to SP, and the need for us to assume it, even with “given” understood in an ontological sense;^[14] consistency is preserved in interpreting the requirement that “one assumes” SP as weak, a kind of hypothetical commitment (Kant, 115). It may be difficult to decide who is more on the right track, or whether their debate opens up some even better third way; but for approaches to consistency, I think we have more an embarrassment of riches than any lack of leads.^[15]

However – and this is where I aim to focus in the present article – it is one thing to be optimistic that two lines of thought can be consistent, and quite another to see how they promise to fit together into a philosophical argument promising as grounds for a broad, unified project in theoretical philosophy. Proponents of the non-Dialectic-first approaches see elsewhere in the Critique arguments promising in this sense. Just for example, Strawson sees arguments earlier in the book for a principle of significance. And everyone can at once see how such arguments (if good) could be meant to define and ground an extremely broad project: They would justify application of the principle to reform all the parts of philosophy, prominently metaphysics.

But pessimism about finding something similar for a Dialectic-first reading can seem to block any viable path, even before we begin moving. To see why, consider first the positive strand of the Dialectic: It is central here to claim an unavoidable interest of reason in the unconditioned, and an unavoidable need for such an interest to guide all theoretical inquiry. This unavoidability has clearly visible implications on the negative side: It would be needed to support the claim that an illusion of reason is “natural and unavoidable” or “cannot be avoided at all” (A298/B354). And to support the claim that each cosmological question at stake in the Antinomy is “one that every human reason must necessarily come up against” (A422/B449), so that “it is not feasible … for reason to withdraw and look upon the quarrel with indifference” (A464/B492).

Unfortunately, where the world of Kant-interpretation looks to these unavoidability claims, the tendency is to see them as more explained by Kant’s biography and times, in a sense supposed to contrast with support by philosophical argument. Allison takes Walsh’s view here as typical:

The illusion of which he speaks was perhaps ‘natural and inevitable’ to a thinker with Kant’s background in rationalist metaphysics, but would be less dangerous for, say, a scientifically-minded positivist.^[16]

Something like that seems to me to be the default today.^[17] In this environment, work on Kant can certainly try to defend a more sympathetic line on the unavoidability claims,^[18] but I don’t think it can presume that there is likely promising argument in the Critique on this point and take things from there. Compare the way in which the interpretive landscape would allow work to more or less presume that there is promising argument in the Deduction and depart from that presumption in various directions.

Further, I think we should cede that, given any now familiar conception of how to put a philosophical project together, the Critique would be surprisingly light on arguments for the positive unavoidability claims. Perhaps – as I will further argue – a better understanding of the project of the book would reveal that the amount and kind of argument on this score fits the project well and is unproblematic. But prior to such an argument, there is an initial problem facing those who would look to the positive side of the Dialectic for arguments promising enough to ground the project of the Critique. The same applies to the idea what is central in the Critique is the project of balancing the Dialectic’s positive and negative lines of thought concerning reason and similar.

What about the negative side of the Dialectic? Certainly, recent interpreters are more likely to see its conclusions about the unviability of rationalist special metaphysics as supported by promising argument. But if we ask what has seemed so promising, the answer tends to come back to the application of epistemological ideas from earlier sections of the book about forms or conditions, which in turn establish an epistemic limit – which limit rationalist special metaphysics seeks to overstep. But this is not Dialectic-first; on this account, what moves the critical argument central to the book are epistemological arguments earlier in the book. This is one of many ways of imagining the Dialectic as merely, to borrow Meer’s apt complaint, “ein Zusatzt zur Transzendentalen Analytik.”^[19]

True, we could also look to the Dialectic for some more distinctive critical argument, independent of earlier epistemological results. I will argue that the Antinomy, for example, is meant to provide such an argument. But the Antinomy, and other contending critical arguments more specific to the Dialectic, presume something else: Claims from the positive side of the Dialectic. Again, if reason really takes an unavoidable interest in the unconditioned, then perhaps it makes sense to think that each Antinomy question is unavoidable, and that an illusion is unavoidable which requires critique. But since the positive side of the Dialectic has the problem above, of seeming to lack sufficient defense, any really Dialectic-first line of critical argument threatens to turn out superfluous to, and weaker than, the kind of critical argument that seems to promise to support critique and ground a broader project.^[20]

And so, the general problem is this: Wherever we might hope to find arguments really based in the Dialectic, or Dialectic-first arguments, there can seem to be little hope of promising argument, able to support a broad project in theoretical philosophy.

To answer, I take up the idea that it is central to the Critique to develop a critical argument that ambitious forms of metaphysics, prominently including rationalism, are unacceptably dogmatic – an idea that can seem to maximally encourage worries that any line of thinking in the Dialectic that is not merely secondary would be merely superfluous to stronger argument grounded elsewhere. This should make things maximally tough for my purposes of putting the Dialectic first. But I argue that even this idea forces us back toward prioritizing the Dialectic: §2 shows that, given a problem about begging the question, Kant’s ambition to argue critically against rationalist metaphysical systems cannot rest content with consideration of epistemic or semantic conditions and limits; that aim requires an independent basis in the arguments of the Dialectic. §3 argues that, once we see why the Dialectic is required, we see why it is no problem for the Dialectic to say the relatively little it does say in defense of its positive claims about reason, the unconditioned, and unavoidability; and if we desire more, it is easy enough to develop what Kant does say into more complete arguments. §4 turns to explain how a distinctive kind of argument in the Dialectic – in particular, the Antinomy – does promise to deliver what is required of the Dialectic to the end of non-question-begging criticism of ambitious forms of metaphysics. I conclude, then (§5), that the problem of project-defining arguments does not blook the path to a Dialectic-first approach to the Critique; consideration of the problem rather brings out attractions that should pull us into more exploration of that kind of approach.

2 Critique and epistemic limits need support of the Dialectic

One worry above was that any argument founded internally to the Dialectic itself, and critical of rationalist metaphysics, would be superfluous to and weaker than an argument from consideration of epistemic conditions and limits from earlier. In this section, I argue that this sense of superfluity is mistaken. Those epistemological considerations are not sufficient for arguments against ambitious forms of metaphysics, in the sense sought in the Critique. The critical ambition will require the Dialectic to provide something different, and for that something different to play a central role in critique.

To begin with, the Critique surely aims, at least in part, to convince philosophers that rationalist metaphysics is unacceptably dogmatic. The philosophers Kant seeks to convince are not already proponents of Kantian critical philosophy: There are not many of those around, prior to publication of the Critique! The audience is made up largely of proponents of the kinds of the philosophy that, according to Kant, are unacceptably dogmatic. Rationalist metaphysics in particular is popular in Kant’s context. The Critique seeks to change this. Not, of course, by means of violence or threats, but by means of argument.

The implications of this can be difficult to keep in mind, given that there are not that many committed rationalist metaphysicians around anymore. But as soon as we think of actually convincing such philosophers with argument, some points should be easy to see. First, Kant cannot advance his aims by just announcing his critical position: Rationalist metaphysics is unacceptably dogmatic. This is not argument; it is just stating the desired conclusion. And to advance that as argument is to fail, begging the question.

But begging the question can take more subtle forms. Imagine a philosopher argues that there are abstract objects, perhaps numbers in particular. An opponent is a nihilist and replies as follows:

1. Nothing exists.

2. Thus, numbers do not exist.

But the defender of the numbers has given an argument for existence, and they take it to provide excellent reason to infer that at least something exists, against (1). The worry may be easier to see if we imagine an historical context in which abstract objects are popular. But the point itself is not really about historical context. To put the above forward as an argument engaging with someone defending numbers, something that should in principle provide reasons for that audience to change their mind – the above is still too much like begging the question.

For the same reasons, to attribute this kind of approach to Kant is also potentially question-begging (QB):

QB1: Begin with one part of the overall package of claims that is the critical philosophy.

QB2: Show that this one part supports the other parts of the package, extending all the way to the parts which conclude that rationalist metaphysics is unacceptably dogmatic.

Say the first part of the package is something rationalist metaphysicians reject, taking their own arguments or position to provide reason for rejection. There would then be no sufficient argument here against rationalist metaphysics. This may be easier to see when we manage to remember the popularity of rationalist metaphysics in Kant’s context. But the point is not just about context; if this line is offered as sufficient critical argument engaging rationalist metaphysics, it would be too much like begging the question.

Now, return to the idea that consideration of epistemic conditions – as in Kant’s treatment of the understanding and sensibility, concepts and intuitions, in the Analytic and Aesthetic – would support an epistemic limit, which would then be sufficient to conclude that rationalist metaphysics is dogmatic. I think this will turn out all too close to QB, above, to serve the critical purposes of the Critique.

Granted, it is easy to see how resources in the Critique fit together into such an “argument.” But this is not to say either that it is sufficient for argument or that Kant would take it to be sufficient – more on that below. In such an argument, the first premise would be something like what is often called the “discursivity thesis”: Concepts of the understanding can contribute to genuine cognitions only to the extent that a kind of matter for cognition can be provided by intuitions from an entirely distinct, different source, sensibility. Analytic judgments, then, cannot amplify our cognition (e.g., A7-8/B12), and synthetic judgments “are possible only by the relating of a given concept to an intuition” (C 11:39). Since we began by saying we have access to intuition only where this is provided by sensibility – our “discursive” understanding would not be able to get beyond any limits or “bounds of sensibility”:

No concept can have its objective reality be secured, save insofar as it can be presented in a corresponding intuition (which for us is always sensory), so that beyond the bounds of sensibility and thus of possible experience, there can be no cognition whatever, that is, no concepts of which one is sure that they are not empty. (UE 8:188–9)

Further, it is central to the Aesthetic that our sensibility has a priori forms, space, and time. Now, Kant understands the ambitious forms of metaphysics, targets of critique in the Dialectic, as those which claim to derive conclusions about unconditioned grounds. To apply the epistemic limit above, we simply need to add his further view that “corresponding intuition” can never be provided by our sensibility for the unconditioned. For example, “in sensibility, i.e. in space and time, every condition to which we can attain in the exposition of given appearances is in turn conditioned” (A508/B536). One then might – which is not to say that Kant does – put this line of thought forward as sufficient on its own to support argument for the conclusion that rationalist metaphysics is unacceptably dogmatic.

But, again, if we try to take this as a sufficient case against rationalism, I think we would end up with something too much like begging the question, as above. I will give two arguments for this, hoping that at least one will speak to any given reader.

The first is the discursivity argument: What I labeled “first premise,” above, requires that concepts and intuitions are entirely different or from entirely distinct roots; the rationalists are supposed to have “intellectualized” intuitions, while the empiricists “sensitivized” concepts, instead of “seeking two entirely different sources of representation in the understanding and the sensibility” (A271/B327). The problem would be that it is then clear, even in Kant’s own statement here, that a central point of rationalism is to reject this view. And, meanwhile, epistemological argument in the Aesthetic and the Analytic seems more organized by this basic claim, than it seems to offer direct argument in support of it in the face of that particular challenge.^[21] So, it is hard to see how this provides the rationalist audience with a step-by-step argument that should convince them to turn to critical philosophy.

Now, to anticipate my conclusion below, my position is not that the Critique fails as a whole to support its claims about discursivity or the epistemic limits that follow. Rather, only that the epistemological consideration of conditions and limits in the Aesthetic and Analytic don’t directly support either. I argue below that the Antinomy of the Dialectic is needed to (§3), and does (§4), support the epistemic limits; I return in §5 to Dialectic support for discursivity as well.^[22]

But I return to present concerns: Perhaps some are not convinced by the case concerning discursivity. So, I advance as well a second argument, from metaphilosophy: Consider a rationalist metaphysician at the height of rationalist metaphysics – call them RM. Should an Aesthetic/Analytic argument be sufficient to give reason to RM for a change of mind? I don’t think so. RM has made their case that, say, their cosmological argument provides adequate justification to conclude that there is an unconditioned ground of the universe. They can naturally infer from this that, if we have epistemic limits, they are found farther out than such conclusions of rationalist special metaphysics. It would be different if a critical argument included a direct response to RM’s cosmological argument, engaging on its domain – namely, consideration of worldly conditions and the unconditioned, or this kind of metaphysics. But our present critical argument does none of that. It just proceeds to consider something else entirely, namely, epistemology (broadly speaking) and epistemic conditions.

What is at issue here is metaphilosophical: Which subdomain of philosophy takes priority? The epistemology-based argument above presumes that consideration of epistemic conditions and limits take a kind of authority throughout philosophy. Compare, for example, what seems obvious to Strawson, namely, that consideration of the limits of meaningful concept use is prior, and dictates what else in philosophy will and will not be legitimate. This may be more about what we could call “semantics” than a narrow sense of “epistemology,” but I link them together under a broad sense of an “epistemology-first” metaphilosophical commitment.^[23] RM remains free to hold that consideration of the metaphysics of worldly conditions and the unconditioned can rationally shape in turn our position on such broadly epistemological concerns.^[24] I do not here express any doubts about the possibility of arguing for the metaphilosophical view that such epistemological considerations should have authority over that kind of metaphysics. The point is that the issue at stake is whether that sort of consideration has the authority needed to resolve the issue; in defending the affirmative answer, we cannot just appeal to those considerations we are trying to show are authoritative over others – that would essentially assume what is to be established.

So far in this section, I have argued for the philosophical insufficiency of the epistemological considerations from the Aesthetic and Analytic for even just the critical aims of the Critique. I suppose some might question the interpretive relevance of this if they think it is clear that Kant himself takes those epistemological considerations as sufficient for that purpose. Of course, I cannot here consider all of the passages that bear on the relations between the sections of the Critique. But no one should think it is clear that Kant sees a critical case against rationalist metaphysicians that is independent of the Dialectic and sufficient on its own; there are plenty of suggestions of a position consistent with the above.

Consider in particular the B-Preface. Here, Kant holds that what is of first importance for metaphysics is to enter “the secure path of science” (Bxv). The central requirement this carries – parallel to the scientific revolution – is that of finding a central role for experiment. And the experiment is not an analysis of our faculties of sensibility and/or understanding. Rather, the “experiment” uncovers the “unavoidable conflict of reason with itself” (Bxviii-Bxix): The reference to reason refers us to the Dialectic, and the reference to a conflict of reason with itself to the Antinomy within the Dialectic.^[25]

And the B-Preface continues to develop the point. Kant distinguishes the epistemological line of argument through the Aesthetic and Analytic, on the one hand, from the Dialectic, on the other. From the former, “there emerges a very strange result,” and in particular, an epistemic limit: “we can never get beyond the boundaries of possible experience.” Kant immediately adds:

But herein lies just the experiment providing a checkup on the truth of the result of that first assessment of our rational cognition a priori, namely that such cognition reaches appearances only[…] (Bxix–Bxx)

This is consistent with the position I defended above: as far as the epistemological argument tells us in itself, the conclusion would be strange, leaving us unsettled. But the experiment discharges the concern, and it stems again from considerations about reason, the unconditioned, and contradiction (Bxx).

There would be no reason to restrict such a textual case to the B-Preface. In the last chapter of the Analytic, Kant makes clear that we cannot be “certain” that we have no epistemic “hope” beyond “the land of pure understanding” until we actually explore beyond its limits, demonstrating that what lies beyond is “the true seat of illusion.” We cannot do so “before we venture out on this sea, to search through all its breadth and become certain of whether there is anything to hope for in it” (A236/B295-6). And the account in the Prolegomena is similar, describing the Antinomy as:

[t]he single possible case in which reason would reveal (against its will) its secret dialectic (which it falsely passes off as dogmatics[…]

a decisive experiment (Versuch), which must necessarily disclose to us a fault that lies hidden in the presuppositions of reason[…] (P 340–1)

The Antinomy is what can “force a self-examination” of reason (P 340).^[26]

3 Status of positive claims about reason and the unconditioned

The problem of begging the question, then, points us to philosophical and textual reasons to think that the Dialectic is and is supposed to be required to support critical argument that rationalist metaphysics is dogmatic. But my approach immediately suggests challenge: Why would an essential role for the Dialectic, and the Antinomy in particular, make any difference specifically with respect to the problem of begging the question? Once we see the answer to this question, we can proceed to the worry that the positive side of the Dialectic is philosophically weak because insufficiently supported: We can see that the argumentative context imposes no need for more argument than the Dialectic provides for its positive claims about reason and the unconditioned; and that, where more argument is required, it is easy to see how Kant’s comments can extend to provide it.

Let us first look in more detail at the B-Preface idea about an experiment. The experiment draws on a premise:

P: “that which necessarily drives us to go beyond the boundaries of experience and all appearances is the unconditioned, which reason necessarily and with every right demands in things in themselves for everything that is conditioned, thereby demanding the series of conditions as something completed.” (Bxx)

There are plenty of interpretive issues about this that would have to be addressed in work focusing just here; for our purposes, Kant’s formulation will be enough.

We also find an assumption – which Kant takes to be natural – to be tested by experiment. For our purposes, it is best to take the assumption to just be the denial of the epistemic limit Kant seeks to establish, namely, the “strange result” that “we can never get beyond the boundaries of possible experience” (Bxix). So, the assumption (A) is:

A: we have no epistemic limit to within the bounds of possible experience.

Given the premise, the assumption is going to generate contradictions, showing that we must reject the assumption and accept that we have an epistemic limit, the bounds of sensibility. “Thus,” Kant later writes, “the antinomy of pure reason leads inevitably back to that limiting of our knowledge[…].”^[27]

Now, Kant in different places considers larger and smaller versions of the natural assumption. In the Antinomy, it is clear that the full natural assumption to be considered is “transcendental realism.” For the Antinomy is clear that its contradictions force the rejection of this and adoption of the whole package of “transcendental idealism.”^[28] In the interest of space, I have so far focused on the epistemic limit that is just one part of that package, and so on the natural assumption that we do not have such an epistemic limit, highlighted in the B-Preface. So, I mean to highlight part of what follows from the contradiction, namely, we have an epistemic limit to appearances. Or, we have an “unavoidable ignorance in respect of the things in themselves” (Bxxix).

In any case, what is needed from (P) and (A) together is to establish the kind of premise that drives Antinomy arguments for contradictory conclusions. And Kant is later pretty clear about what this is:

The entire antinomy of pure reason rests on this dialectical argument: If the conditioned is given, then the whole series of all conditions for it is also given; now objects of the senses are given as conditioned; consequently, etc. (A497/B525)^[29]

The major premise here is reason’s principle, SP, or perhaps it would be better to say that the major premise helps itself to SP in a sense that would support metaphysical conclusions, as in the Antinomy arguments. Kant says that this “major premise seems so natural and evident” (A497/B525), and that is simply the application of his view about a natural temptation to endorse metaphysical use of that principle. And SP, understood in this way, is again the rationalists’ PSR.

Coming back to the sketch of the experiment in the B-Preface, then, the idea of the setup is this: Kant endorses in (P) some sense in which reason legitimately demands completeness of conditions or the unconditioned. And he cedes that it is natural to assume (A) that we do not have the sharp epistemic limit he is defending. In the absence of some reason for that epistemic limit, the way would be clear to infer from what reason legitimately demands we think, to how reality in itself must be. It must be in accord with (SP), understood in this way as a metaphysical principle.

However, from (A) – the lack of a limit – it follows that we can apply the principle demanding the unconditioned to objects of experience; and this, the Antinomy promises to show, generates contradictions:

[…]we find that on the assumption that our cognition from experience conforms to the objects as things in themselves, the unconditioned cannot be thought at all without contradiction[…]

And so, the assumption (A) must be rejected, yielding the position Kant wants the experiment to confirm:

our representation of things as they are given to us does not conform to these things as they are in themselves but rather that these objects as appearances conform to our way of representing. (Bxx)

As soon as we instead assume this, “then the contradiction disappears” (Bxx). And so, “this would show that what we initially assumed only as an experiment” – in particular, Kant has emphasized the epistemic limit that initially was strange and in need of independent confirmation – “is well grounded” (Bxx–xxi).

It is easy to see why this form of argument makes a difference with respect to the problem of begging the question against rationalist metaphysics: Everything that is supposed to set the experiment in motion is entirely congenial to rationalist metaphysicians. Certainly, they would want to think that, (P), reason itself demands the kinds of unconditioned grounds that rationalists want to defend, and that we all make use of and are guided by this demand in everything at all like theoretical inquiry. Compare Leibniz on the PSR: “[h]as not everybody made use of this principle upon a thousand occasions?.”^[30] Further, Kant holds that the assumption (A) is entirely natural, so that the burden of proof would fall to anyone rejecting it. Further, Kant holds that, if the natural (A) holds up, then a form of rationalist PSR is justified for use in metaphysical reasoning about reality. So, overall, the burden of proof is on anyone rejecting such use of the PSR. This is entirely congenial to rationalist metaphysics.

The point so far, then, is this: The form of argument Kant promises in the Antinomy of the Dialectic is one driven by premises that rationalist metaphysicians have reason to accept. That is, problems arise internal to the form of philosophical project that rationalists themselves privilege – consideration of the metaphysics of conditions and the unconditioned – and those problems internal to the favored rationalist domain force a revision.

Having explained in this way why the Antinomy promises to do better with respect to the problem of begging the question against rationalists, I return to this concern from above: Any Dialectic-first argument that is critical of rationalist metaphysics would presume the claims of the positive side of the Dialectic, about reason, the unconditioned, and unavoidability. But those positive claims don’t seem to get a lot of support by argument in the Critique, and interpreters have tended to read them as more prejudice of the day than conclusions of promising arguments. Allison, again, takes as typical Walsh’s worry that what Kant thinks is unavoidable would seem obviously optional and unwise for a “scientifically-minded positivist.”^[31]

But understanding how the argumentative context requires the Dialectic to play the leading role should already show that this worry should not stand in the way of further exploration of Dialectic-first approaches. What would seem to promise to impress Walsh more would be if we could begin from the commitments of an anti-metaphysical positivist and then build support, in those terms, for Kant’s claims about the unavoidability of reason’s demand for the unconditioned, which then might support a claim about an unavoidable illusion. But this could not be more out of place with respect to the demands of argument, or the (in the ordinary sense of the word) dialectical or argumentative context. The challenge is to mount a non-question-begging argument against rationalist metaphysics. In that argumentative context, beginning from positivist principles – designed to rule the kind of metaphysics pursued by rationalists as meaningless – would be a paradigmatic case of begging the question.

To be sure, if the arguments of the Dialectic are to be project-defining, then we might still think it makes sense to ask whether the positive claims about reason and the unconditioned can be more supported by argument. And perhaps, it is true that one reason why Kant does not give more argument is that the historical context does not demand it: There were not in his time many anti-metaphysical positivists around, and no such positivists are working as gatekeepers, or the analogue at that time of go-to referees at prestigious journals. But this too is no reason to pre-emptively turn away from the Dialectic-first path. For the Dialectic does have something to say for its positive claims. And it is easy to point out the direction in which we can begin filling out more argument, based on what Kant does say. In the rest of this section, I will sketch the way I have developed the argument, specifically in relation to Walsh’s worry.^[32]

More specifically, the anti-metaphysical positivist response, suggested by Walsh, would be that theoretical inquiry can be perfectly well guided without need any thought of supposed unconditioned grounds. Kant claims that guidance is unavoidably required by a faculty, reason, that demands conditions and ultimately the unconditioned. I begin with the first guidance claim (G1), just about conditions (holding off on the unconditioned at first):

G1. Theoretical inquiry requires at least the guiding goal that we seek, and so think of, underlying conditions.

I take a “condition” to be anything to which one can legitimately appeal in answering an explanatory why-question.^[33] What might hide the need for such guidance would be the assumption that theoretical inquiry can be adequately guided simply by the aim of avoiding false beliefs. But this is not enough: That goal could be served by an attempt to hold as few beliefs as possible, and that attempt is not theoretical inquiry.

Might inquiry only require the aim or interest of discovering truth? No, because truths can be trivial and not worth pursuit in theoretical inquiry. I like to make the point by saying that, after discovering one truth, p, the easiest path would be to draw an unending number of inferences: p or q, for any q. But that is too trivial to be a path of theoretical inquiry. This kind of concern, about trivial truth, is not just an idea from a bygone period of the dominance of rationalist metaphysics; it still plays a large role today in arguments that truth is not the only epistemic value.^[34] A similar point applies to the aim of avoiding contradictions. As Kant puts it, avoiding contradiction:

[…]constitutes no part of its interest but is instead the condition of having reason at all; only its extension, not mere consistency with itself, is reckoned as its interest. (KpV 5:120)

The case for this positive claim about an interest of reason in conditions is, then, that theoretical inquiry requires some guiding aim or interest that points away from the merely trivial directions that would be allowed by thinner aims of which Walsh’s positivist would approve, including avoiding contradiction and falsity and piling up truths. And Kant’s provides this: Reason demands that we seek the conditions in terms of which we can explain.

Here is an example from Kant, in the Appendix to the Dialectic account of the role of reason in natural science. In observing the shapes of the orbits of the planets, “we find variations,” rather than one uniform shape. But scientific inquiry does not just end by noting this orbit shape of Mars, and this of Venus, etc. “Reason… seeks the unity of these cognitions.” It does so by demanding that we find a unified account in the “properties and powers of things.” It requires the assumption that there is in this case underlying simple kind (Gattung) obeying a simple law of force, and that we seek unified explanation of multiple phenomena in these terms:

under the guidance of those principles we come to a unity of genera [Gattungen] in the forms of these paths, but thereby also further to unity in the cause of all the laws of this motion (gravitation); from there we extend our conquests, seeking to explain all variations and apparent deviations from those rules on the basis of the same principle. (A662–63/B690–91)

Reason requires us to seek something new and non-trivial: The natural kinds and the ways in which they govern or condition natural phenomena.^[35]

Of course, Walsh’s “scientifically-minded positivist” will endorse the goal of seeking explanations for natural phenomena, but provide a more deflated conception of what this consists in. In particular, they might seek an understanding of explanation that limits it to epistemic issues about inferences that fit a pattern. Here, I will just point out that this positivist approach faces well-known difficulties; for example, it has trouble with asymmetries of explanation. It is not as if only philosophers working in a context where rationalist metaphysics is dominant would find those worries persuasive; it played a significant role in the context of the dominance of positivism.^[36]

I proceed now from G1, on conditions, to a parallel claim about the unconditioned:

G2: Theoretical inquiry requires the guiding goal that we seek the unconditioned.

Kant’s way of arguing from conditions to the unconditioned in the case of the logical use of reason is this: If reason demands, we find a rule that is the condition of some cognition, and that rule is further conditioned, then reason must require reapplication of the same demand, and so on; the aim that unifies the specific demands in these cases is the aim of discovering the unconditioned. Kant says:

since this rule is once again exposed to this same attempt of reason, and the condition of its condition thereby has to be sought… we see very well that the proper principle of reason in general (in its logical use) is to find the unconditioned for conditioned cognitions of the understanding, with which its unity will be completed. (A307/B364)

This is about the logical use of reason and rules about cognitions. But the point about reason’s interest in worldly conditions – underlying natural kinds and laws, for example – is parallel: If reason demands that I find something in terms of which to explain something conditioned, and that something is further conditioned, then the demand reapplies; the aim that unifies the specific demands is a general demand that I seek underconditioned grounds.

Certainly, an article focused just on constructing this defense from Kantian elements would need to say more. But in this section, my aim has been a more preliminary engagement with worries about the need for and status of the Dialectic’s positive claims about reason and the unconditioned. What seemed worrisome at the start, I have argued, turns out to be no reason for any doubt about viability to tell against further exploration of Dialectic-first approaches to the Critique.

4 Antinomy as promising non-question-begging critique

One might worry that the arguments above do not resolve the problem of project-defining argument, so much as shifting its weight. Say for the moment, the above is correct: The Antinomy is needed to generate a problem ending the prospects for rationalist metaphysics, just from premises of which it is both true that:

1. Rationalism gives reason to favor them, and

2. Kant would either accept them or at least think them entirely natural, enjoying the burden of proof, and such that any rational thinker will always be tempted to endorse – so that avoidance of endorsement requires the critique of illusion.

But this is now a tall order for the Antinomy! How could premises favored by rationalism be fatal to it? Perhaps unstated skepticism on that score is what often moves interpreters to think that the Antinomy arguments require some other form of support – and, in particular, support from the very epistemological commitments that they are (as I argued above) supposed to support. If that is needed, then there won’t be much hope to find in the Antinomy promising arguments to define a broader project.

But I think we have seen enough to see why interpreters would so often estimate so poorly the chances for the Antinomy. As discussed above, interpreters often neglect the positive strand of thought in the Dialectic. Or, where they note it, they see it as critically weak in philosophical argument, for reasons like Walsh’s. For this reason, interpretive charity can seem to counsel seeking to understand Kant’s project in ways independent of that weakness, and those positive claims about reason and the unconditioned.

Note in this connection that the critical Kant in general, and the Dialectic in particular, is indeed very dismissive of some metaphysical lines of thought as unnatural and easily avoidable. This is the case with the ontological argument, at least in the sense of this passage from the Transcendental Ideal:

It was entirely unnatural, and a mere novelty of scholastic wit, to want to take an idea contrived quite arbitrarily and extract from it the existence of the corresponding object itself. (A603/B631)

No premise from that argument could play a role in the Antinomy: The premises must be at least such that Kant thinks them natural and enjoying the burden of proof. Anyone like Walsh’s positivist, however, will be just as skeptical of rationalist metaphysics generally. They will think that being “scientifically-minded” is enough to get one over and done with the whole affair. The closer an interpreter is to that mindset, the narrower the scope they will see for acceptable premises, promising premises in the Antinomy arguments: There won’t be much of anything they would find natural that rationalists would also approve, available to support the Thesis and Antithesis arguments. Such interpreters will more likely find Kant’s own positions on epistemological limits more congenial and may then suspect that nothing in the Antinomy arguments would be promising without such support. This would mean begging the question, as far as what Kant promises for the Antinomy – namely, supporting those very epistemic limits. But then again, it may not seem very costly for such an interpreter to see the Antinomy as failing in this, if they anyway take such Dialectic-first argument for limits as superfluous to and weaker than arguments for limits that they think they see in the Aesthetic and Analytic.

But if we just clear away these kinds of assumptions – and hopefully the arguments above about the need and prospects for the Dialectic can help – and look anew at the prominent features of the Dialectic and Kant’s idea of reason, then it is entirely clear that Kant’s position is not that all of rationalist metaphysics is unnatural. The B-Preface very prominently advertises that those who think themselves merely indifferent to the concerns of reason – on Kant’s account, that means the concern with conditions and the unconditioned – are self-deceived:

[…]it is pointless to affect indifference with respect to such inquiries, to whose object human nature cannot be indifferent… so-called indifferentists, to the extent that they think anything at all, always unavoidably fall back into metaphysical assertions, which they yet professed so much to despise. (Ax)

And the whole idea of an unavoidable illusion, in the Dialectic, is that we are naturally deceived into accepting … the key tenants of rationalist special metaphysics, in particular.

And this opens the way for the simple position that the Antinomy arguments are supported by just the kind of premise Kant identifies: The supreme principle of reason demanding the unconditioned, specifically understood in such a way that would support metaphysical conclusions – as a kind of PSR. We don’t need to see Kant as accepting such a principle; we need to only recognize that he clearly regards it as natural, in the sense demanded in an Antinomy argument. Again:

The entire antinomy of pure reason rests on this dialectical argument: If the conditioned is given, then the whole series of all conditions for it is also given; now objects of the senses are given as conditioned; consequently, etc. (A497/B525)

Maybe different readers would find on further investigation and interpretation that the arguments, read in this way, work – or not. But I do not think that prospects are so dim, in advance of that consideration, as to make us already pessimistic about the viability of Dialectic-first approaches.

Here is how I apply this idea to the Second Antinomy, for example, on parts and composition of spatio-temporal objects. There is of course no space here for complete interpretation and defense; the point is just to make the case that there is no good reason right from the start to count out success relative to Kant’s specific promise of supporting (without presupposing) the Kantian epistemic limits that would then rule rationalist metaphysics dogmatic.^[37]

The Antithesis argument concludes that there can be no final, simple, or indivisible parts of spatio-temporal composites. Kant begins by making an assumption, for the sake of reductio: There are simple parts of composites. The composites of interest in the Antinomy are spatio-temporal. Say there is such a spatio-temporal composite. Then, a conditioning-relation seems to hold: Why does the whole fill this region of space? Because it is made of these parts. But when the resulting regress reaches supposedly simple parts, the previous conditioning relationship will require that the simples fill a region of space – so that they can add up, as it were, to something filling the whole region. But then, we encounter Kant’s point about the re-application of reason’s demand for conditions, so important to his case that reason’s interest is the unconditioned: Our simples fill a certain space, and reason would demand some further explanation of why, in terms of some further conditioning relation connecting the whole to smaller parts. And so, the argument seeks to conclude that such parts would have to fill space, and so be divisible, resulting in contradiction: “the simple would be a substantial composite” (A435/B463). And the attempt to think simple parts in space as a completion of conditions, or as unconditioned – as demanded by the principle of reason – would contradict itself, and we must conclude instead that the regress to conditions is exhausted by, and find its only completion in an infinite descent of composition all the way down.^[38]

While there is no space here for complete analysis of the Antithesis, Bowman (“Demand”) has undertaken the project with a focus on showing the weight to rest on SP. They consider, for example, Guyer’s claim that the Antithesis argument assumes and must assume the kind of epistemological commitment that it is meant to support; Guyer thinks Kant argues that “because simple parts could not be perceived in space they could not exist” (Kant, 410). That would be question-begging, in the sense above. But it is Kant’s Remark on the Antinomy which refers back to the Aesthetic (A441/B469). Bowman points out that if we simply take the Remark as a remark, a meta-commentary, and not part of the text of the Antithesis argument, then there is little to connect that argument itself to the epistemological assumptions.^[39] If a form of SP or PSR can support the weight, then, there would be little reason to import such epistemological assumptions.

Now, consider the Thesis of the Second Antinomy. Here, we assume what the Antithesis defended: The series of conditions in the sense of parts is exhausted by an infinite regress to ever smaller parts. Here, we should think in terms of the model of a rationalist cosmological argument, which would seek to bundle together an infinite series of contingent causes, arguing that a bundle of contingent beings is still something contingent; if anything contingent and existing requires a cause of its existence, the series cannot be exhausted by an infinite series of contingent causes. The parallel claim would be that if we take some composition relations, and add more composition relations, we only ever have more composition relations. We then only need to add the conditioning relation between relations and the relata needed to stand in them. Even an infinite regress of composition relations would still need relata. In Kant’s terms:

[A]ssume that composite substances do not consist of simple parts: then, if all composition is removed in thought … nothing at all would be left over […] (A434/B462)

Quite generally, relations require relata, then we can “think away” the relations and must have something “left over”: The relata. The way is open to, along this path, argue that reason’s demand for completeness in a series of conditions demands in this case simple parts.

Again, for reasons of space, I have focused here on the Antinomy as support for Kant’s position on epistemic limits, not on the broader issues of support for the whole of transcendental idealism. Looking at it in this light, we can return to the broad strokes of the experiment as sketched in the B-Preface. There is a premise, supposed to be common ground with the main, if not the only, target – rationalist metaphysics, along these lines: Reason demands, with every right, the unconditioned. There is a natural assumption. If we focus on the specific issue of limits, the assumption is that we are not unavoidably ignorant of anything but appearances. Given the premise, as long as the natural assumption stands, what reason requires us to think is supposed to be as good a justification as anything else we could have for drawing conclusions about reality. And so, we would have a form of reason’s supreme principle, understood as supporting metaphysical conclusions about worldly conditions and the unconditioned, like the rationalists’ PSR. I have argued that we should not count out the possibility that such a principle would be strong enough to justify contradictory conclusions, as in the Second Antinomy. And so, I conclude here that there is no reason, in advance of further interpretation and investigation, for seeing the prospects for the Antinomy supporting Kant’s epistemic limits – as he promises – as so dim as to conclude that the path to a Dialectic-first reading is not viable at all.

5 Down the Dialectic-first path

In what sense is the path we have been traveling in a Dialectic-first direction? The first point to make here is this: It is a familiar idea that a critical argument, which includes rationalist metaphysics within its target, is at least one central aim of the Critique. Further, I have argued that such a critical aim cannot rest on the discussions of epistemic conditions and limits in sections prior to the Dialectic, lest it merely beg the question; Kant sees the Dialectic, and the ways in which the account of reason there supports the Antinomy in particular, as needed to ground such a critical argument (§2). There is no reason to rule in advance that status of Kant’s positive claims about reason and the unconditioned would undermine such an argument (§3). Nor to rule in advance that using the Dialectic’s arguments of the Antinomy to support epistemic limits would have to beg the question (§4). So, precisely, if the critical argument, whose targets include rationalism, is central to the aims of the critique, then we can already see that something central or prior in importance is being done in the broader endeavor by the Dialectic, specifically insofar as it must argue in a way independent of earlier sections and not merely an addendum to them.

One might object that this line of argument so far does not reach the promised destination, as follows: It would only follow from this (the story will be) that there is an independent line of argument in the Dialectic, which has at least one important strength relative to arguments elsewhere in the book (strength relative to the problem of begging the question against rationalism). For all that tells us, one might well conclude that there are also independent arguments in the Aesthetic/Analytic which have their own strengths relative to the Dialectic. And so, we might get a kind of independence of different strands, but not yet any priority of the Dialectic with the project of the book.

In fact, I suspect that many would not only see such an independent argument about epistemic conditions, but also that its strengths ultimately eclipse those of any independent argument in the Dialectic. In particular, an independent argument in the Dialectic can seem rather narrow in scope: At first, it might seem that the argument of the Antinomy targets only rationalist metaphysics. For the objection to have prospects, it would have to include a fairly wide sense of rationalist metaphysics: It would include arguments like those of the Theses, which employ reason’s demand to support the existence of unconditioned grounds separate from the whole of a series of conditions. But it would have to include as well arguments like those of the Antitheses, which deny the existence of such separate grounds – or, to put the point differently, take a series of conditions itself to satisfy the demand of reason for the unconditioned.^[40] Still, that might still seem a fairly narrow kind of reasoning, namely, only the kind that makes metaphysical use of reason’s demand for the unconditioned. Imagine we had in hand and were comparing:

1. An argument (from the Dialectic) showing that reasoning making metaphysical use of reason’s demand for the unconditioned is unacceptably dogmatic.

2. An argument (from the Aesthetic and Analytic) showing that any reasoning attempting to draw conclusions beyond the bounds of sensibility is unacceptably dogmatic.

Thinking in this way, we can see why Willaschek, for example, characterizes Analytic/epistemic limit–based critique as “much more general and potentially more damaging.”^[41] One might even think that these strengths of non-Dialectic-based critique, relative to the Dialectic, still leave it much more important and central to the project of the Critique.

But there is just here an opening for a Dialectic-first approach. It would argue as follows: The argument about epistemic limits, earlier in the Critique, still have the premises noted above: It still assumes the metaphilosophical position that consideration of epistemic conditions has the authority to stand in judgment of work on the domain of the metaphysics of conditions and the unconditioned, rather than the reverse. And it still assumes that our understanding is merely “discursive,” so that our cognition requires for our concepts some corresponding intuition, which for us is always from the entirely distinct root, sensibility.

The opening for a Dialectic-first approach to the Critique is to argue that Dialectic-based critique is in fact needed to support those very premises crucial to the analytic. With respect to the authority of epistemological concerns, note that the Antinomy threatens to show that reason contradicts itself through and through, supporting what Kant calls “skeptical hopelessness” that goes so far as the “euthanasia of pure reason” (A407/B433–34). That is, the threat is that no critique of illusion, and via this avoidance of deception, will remove the contradiction, which would infect all rational inquiry and thought. And so part of the point of the Antinomy would be to demonstrate that we cannot avoid recognizing the authority of epistemological concerns. Just pursuing metaphysics, on the domain favored by rationalist metaphysicians, naturally generates a large skeptical difficulty. To be sure, it is not much like what we might have expected, especially if we take problems about the objectivity of knowledge as central to the critique, such that they would be addressed by deducing the objective validity of categories. Nor are anything like Cartesian skepticism about correspondence of our representations with something outside them, where we might imagine that common sense is at least some antidote. The idea is rather that common sense – what comes naturally to us – is in agreement with the metaphysical reasoning that generates contradictions and critique.^[42] And there would then be no chance of metaphysicians clinging to their reasoning on their favored domain, as if this could operate without epistemological intrusion generating consistent conclusions, from which we could draw further inferences about what is within our epistemic reach.

With respect to Dialectic providing the support for discursivity in particular, it is important to distinguish this from support for some limit more generically.^[43] I think there are two kinds of things to say for a Dialectic-first approach here. One involves dropping my narrower focus on epistemic-limit and widen out to consider how the Antinomy grounds the indirect argument for the larger package of Kant’s transcendental idealism. The more complete context of a citation mentioned above connects this broader package with the epistemic limit; concerning reason’s demand for the unconditioned, Kant states his experiment and its results in this way:

If we now find in practice that it cannot be applied without contradiction to objects in space and time, then there is no possible escape from this contradiction, unless we assume that objects in space and time, as objects of possible experience, are not to be regarded as things-in-themselves, but merely as appearances, whose form depends on the subjective constitution of our mode of intuiting them.

Thus the antinomy of pure reason leads inevitably back to that limiting of our knowledge… (Fortschritte, 290f.)

To conclude that we have an epistemic limit to “appearances, whose form depends on the subjective constitution of our mode of intuiting them” requires a principled, sharp distinction of concept from intuition, understanding from sensibility.^[44]

And here is another way to look at the Dialectic support for discursivity: Say we see from the Antinomy that we must accept some epistemic limit that cuts us off before we get into Antinomy contradictions. But just in itself this wouldn’t tell us exactly where the limit is. It would seem ad hoc to conclude that we have an epistemic limit that just happens to cut off all and only precisely the area that, if not cut off, would otherwise generate Antinomy contradictions. What would be needed would rather be a principled theory of epistemic limits, which might have the attractions of a systematic explanation of where and why our cognition reaches some limit point. Or perhaps it is better to say that the Antinomy would prove that we are limited in some respect, opening up a gap between our finite cognition and some ideal or perfect intellect. And then the best, principled explanation of this gap is that our understanding is merely discursive, so that extension of our knowledge requires not only concepts from the understanding but also an entirely distinct source of intuitions from a receptive faculty of sensibility. This would allow the conclusion not just that we have some epistemic limit, but that we are limited within the “bounds of sensibility.”^[45]

On this reading, then, the broader argument for a limit in the earlier sections of the book would need support from the Dialectic, and the extended critical argument that is supported at base by the Antinomy in the Dialectic would itself provide the needed support for the broader limit. So, it would not be correct to think that the epistemological considerations of the Aesthetic and Analytic are any independent argument, which might then also have comparative advantage in being broader in scope. Rather, it would really be the account of reason in the Dialectic, and especially the Antinomy, that gives direction and force to the broader critical project of the book – even if this is not something that a reader could understand before getting very far into the book. Now, we are really thinking in a Dialectic-first manner, even if the priority at issue is not temporal but philosophical.

At this point, we can also begin to point to the path that would broaden out from my focus here on the aim of the Critique for critical argument against rationalist metaphysics. The more we emphasize that aim as central, the more the worry tends to grow that any Dialectic-first arguments are superfluous to and weaker than the core arguments of the book. So, I take it as good news if exploration of this focus ends up rather removing barriers to, and bringing out further attractions of, a Dialectic-first approach to the book. Further exploration of the path would then best shift toward understanding the organizing focus of the book in a manner more balanced between the constructive and destructive strands of the Dialectic.

But all that lies beyond my scope in this particular article. Here, my aim has been, first, to defend the importance of a problem that goes some way to explaining the relative unpopularity of the idea that the Dialectic is primary to the organizing aims of the Critique as a whole. This is the problem of project-defining arguments. I tried to develop the problem far enough that we could see – should the problem remain unsolved – sufficient reason to doubt the viability of a Dialectic-first reading, and so sufficient reason to direct our interpretive energies elsewhere. And then I have tried to resolve the problem. The problem is no reason to doubt, in advance of further exploration, the viability of Dialectic-first approaches. On the contrary, further consideration of the problem brings out more and more philosophical and interpretive attractions of the idea that the Critique of Reason is organized by the idea of a critique of reason, understood specifically in terms of that section of the book that is most about reason, and its critique: The Transcendental Dialectic.