Semantic Anti-Realism in Kant’s Antinomy Chapter

1 Introduction

Kant’s critical philosophy is not a systematic theory of meaning or truth, but we find an indisputable overlap between epistemology and semantics in his texts. He repetitively refers to “falsity,” “truth,” “judgements,” “empty concepts,” “sense,” and “significance” in order to explain the cognitive architecture of the human mind. However, with respect to certain aspects of Kant’s thinking, these semantic overlaps remain undertreated. By considering the semantic footings of his famous discussion of the antinomy-problem – arising through seemingly incompatible and equally justified propositions about the (empirical) world understood as an absolute totality, for example with respect to its basic causal or modal constitution – this article contributes to the ongoing debate about whether Kant was committed to semantic realism or anti-realism. That is, whether verification-transcendent judgements have a truth-value (realism) or not (anti-realism).

In contrast to the (empiricist) semantic principle that Strawson,^[1] and many others, have ascribed to Kant as the “principle of significance,”^[2] the bedrock of the article is what I call Kant’s Real Principle of Significance: an extension-based and normative principle stating that a judgement is truth-apt iff it deals “with information [Kundschaft]” which every human in principle “can get” (KrV, A 703/B 731),^[3] or: iff it is possible to specify what Kant calls a common “touchstone” (KrV, A 820–821/B 848–849)^[4] for a certain type of judgement, which refers to a method of “how we might achieve acquaintance with” its truth-value (KrV, A 498/B526–527). This principle entails a version of semantic anti-realism. This is the view, to borrow a phrase from Putnam, that truth (and falsity) is not something that “totally outruns the possibility of justification.”^[5]

One of my central claims is that Kant’s semantic anti-realism can be extracted from the antinomy chapter of KrV, in which he debunks the transcendental realist assumption that every meaningful question can be answered (either positively or negatively).^[6] As I read it, judgements of the form “the world as such is x” are truth-aptless or lack “objective validity” in virtue of being unverifiable as a matter of principle. More specifically, none of the eight propositions in the four antinomical conflicts are truth-apt because they contain a necessarily “empty [leer]” concept (KrV, A 486/B 514) and hence are “nugatory [nichtig]” (KrV, A 90/B 123; A 479–480/B 507–8), “absolutely null and void [Schlechterdings nichtig]” (KrV, A 703/B 731) or lack what Kant refers to as “sense [Sinn]” and “significance [Bedeutung]” (KrV, A 241/B 300; A 485–6/B 513–14). I will furthermore argue that his reference to some of the antinomical judgements as “false” is compatible with such an anti-realist reading. In that context, I will show that Kant operates with two distinct world-concepts: an illegitimate transcendental realist one and a legitimate transcendental idealist one.

Alongside reconstructing Kant’s solution to the antinomy-problem in terms of semantic anti-realism, my other main argument is the following: Any satisfactory anti-realist interpretation of Kant must be able to accommodate his claim that the thesis in the third antinomy about freedom can be true in a “corrected significance” (KrV, A 532/B 560) and the argument from the Critique of Practical Reason (KpV) about the “objective reality” of freedom. Although semantic anti-realist readings of Kant have been defended – usually as a way of emphasizing Kant’s restriction of truth-apt judgements to the domain of empirical discourse – as well as challenged by others,^[7] integrating such a reading with Kant’s treatment of the third antinomy on freedom and natural causality has not, to my knowledge, been carried out before. As I shall argue, Kant indeed installs certain boundaries for truth-apt thought, but those boundaries are not as restrictive as (empiricist) Kant interpreters tend to think. This semantic reading of Kant’s antinomy chapter consequently uncovers a pluralist conception of truth and objectivity that can drift between his theoretical and practical philosophy.^[8]

I begin with a brief exposition of the antinomy of pure reason (Section 2) before arguing how to read Kant as a semantic anti-realist by outlining the Real Principle of Significance (Section 3). In Section 4, I argue that Kant implicitly invokes three distinct strategies to solve the antinomy-problem and that all three are either grounded in or compatible with the anti-realist framework outlined in Section 3. I make the case in Section 5 that the Real Principle of Significance fulfils a necessary requirement for any credible anti-realist interpretation of Kant’s philosophy: to give a plausible explanation of his proposal in the third antinomy that human freedom is an objectively valid (truth-apt) concept. In the conclusion (Section 6), I summarize my arguments by sketching a couple of serious objections stated by Lucy Allais against anti-realist interpretations of Kant.^[9] I conclude that my proposed interpretation is more immune to those objections than traditional (empiricist) types of anti-realism.

2 Absolute questions: The antinomies of pure reason

Human thought tends to pose ultimate questions about how absolutely everything hangs together. Questions seeking complete descriptions of the world as such. In the Critique of Pure Reason (KrV), Kant describes this as a “natural predisposition [Naturanlage]” towards metaphysics – e.g. in a theistic or materialistic shape – which will “always remain” with human reason (KrV, B 21). In his recent study of the dialectic of pure reason in KrV, Marcus Willaschek has shown that Kant’s account entails that “metaphysical speculation arises naturally out of principles that guide us in everyday rational thought.”^[10] These are the principles of discursivity (we ask for explanations of empirical phenomena), iteration (we ask for explanations of phenomena, of their explanations, and so on), and completeness (we want our explanations to come to a satisfactory conclusion, which they only do when arriving at answers that do not raise further questions). Although it goes beyond the scope of this article, it is frequently pointed out that there is a tension between this naturalization of metaphysical thinking and Kant’s aim of his critical philosophy to “discipline” this metaphysical predisposition by determining the “boundaries” of human cognition (KrV, A 795/B 823) and evaluating the “capacity and incapacity of reason for judging” metaphysically (KrV, B 22).^[11]

Kant’s discussion of absolute questions is primarily located in the antinomy chapter of KrV. The fundamental premise in Kant’s account of the antinomies is the principle of completeness (PC):

For every existent (object, event, etc., which Kant umbrellas under the term the “conditioned” in the antinomy-chapter) in the “world of sense [Sinnenwelt]”, it is presupposed that the “whole series [ganze Reihe]”, “sum total [Inbegriff]” or “absolute totality [absolute Totalität]” of conditions (=the unconditioned) for this existent is also given. (KrV, A 408–20/B 435–48)^[12]

How should we understand this principle? A guiding ideal for our scientific activities is the following: For everything that is caused or exists as something that is not self-explanatory (e.g. certain mathematical axioms or basic laws of logic), we seek a cause, ground, or explanation for its existence in virtue of being reasoning and knowledge-seeking creatures. In other words, we are never completely satisfied with an explanation that leaves something unexplained.^[13] But PC states something more. It states that there is a possible resting point for this explanation-hunt; that we can reach a point where nothing is left unexplained. Think of the desire within some sciences (e.g. theoretical physics) to search for a theory of everything that would not give rise to further questions. What Kant calls an antinomy, which is a “contradiction in the laws … of pure reason” (KrV, A 407/B 434), arises if one assumes the validity of PC and attempts to determine the world in an absolute sense: the “absolute totality of the sum total of existing things” (KrV, A 419/B 447), “the absolutely unconditioned totality of the synthesis of appearances” (KrV, A 481/B 509), or the “absolute totality” of the series of conditions (KrV, A 409/B 436).^[14] According to Kant, this generates four so-called cosmological antinomies that are all shaped by a set of contradictory cosmo-metaphysical propositions about the world, a thesis and antithesis. The antinomies display a conflict since two mutually exclusive predicates (e.g. finity and infinity in time and space) are applied to the same presumed object (as schematized beneath).^[15]

All eight claims are cosmo-metaphysical propositions about absolutely everything (empirical), the “totality of conditions” since they propose ultimate answers about the empirical world’s quantitative (space-time), mereological, causal, and modal structure. That the propositions seemingly refer to something empirical, or at least depart from empirical cognition, is what distinguishes the cosmological idea from the ideas of the soul and God that Kant discusses in the chapter on the dialectic of pure reason.^[16] More specifically, the antinomies are generated by what Kant calls a “regressive” (KrV, A 411/B 438) or “empirical synthesis” (KrV, A 462/B 490) when we attempt to determine or explain the series of conditions (e.g. the causal series of a particular event). The antinomies arise when a theoretical leap is made in this regression from claims about particular conditions to an absolute claim about the totality of conditions. Such an absolute claim can, according to Kant, come in two different ways:

1. An unconditioned condition that completes the series (the theses, T1–T4, that all entail some sort of metaphysical finitism by positing an unconditioned absolute as a regress stopper).

2. An infinite number of conditions that form a totality (the antitheses, AT1–AT4, that all entail some sort of metaphysical infinitism by not positing a regress stopper).

As Kant frames it, the problem is that both the thesis and the antithesis in all the antinomies seem to have equally good arguments for their case. In the Prolegomena, he says that the “[…] thesis and antithesis can be established through equally evident, clear, and incontestable proofs … I will vouch for the correctness of all these proofs” (4: 340). I will not discuss the soundness of these so-called “proofs” or account for all the individual antinomies in detail (except for the third antinomy in Section 5 of the article), since I merely intend to extract Kant’s overall strategy.^[17] And this strategy does not hinge on the concrete content of A1–A4 or whether they contain incontestable proofs, since Kant’s objective is to evaluate the legitimacy of any kind of absolute proposition about the (empirical) world.

3 Kant’s Real Principle of Significance

According to Kant, the entire antinomical conflict is based on false assumptions because all the propositions are asserted without “paying attention to whether and how we might achieve acquaintance [zur Kenntnis derselben gelangen können]” with what they purportedly refer to (KrV, A 498/B 526–527), because they deal “with information [Kundschaft] which no human being can ever get” (KrV, A 703/B 731). In this section, I will argue that Kant’s (transcendental idealist) boundary-thesis of human cognition entails limitations on the kind of judgements that can be objective or truth-apt; it entails a kind of semantic anti-realism. I return to the antinomies and their relation to this anti-realist scheme in Section 4.

Kant shifts between calling truthbearers “judgements” (KrV, A 294/B 350; 24: 527) and “cognitions” (KrV, A 58/B 83). In Reflexion 2259, we are told: “truth and falsity do not lie in concepts, but in judgements [Urtheilen], namely as assertoric propositions [Sätzen]” (16: 288). That does not entail that all judgements necessarily have a truth-value, but that judgements are what can be true or false. As I will argue in what follows, Kant proposes a normativist and pluralist account of the concepts of truth and objectivity that makes them applicable across different domains of discourse (from judgements about molecules to moral affairs and music). He believes that a type of judgement is objective and truth-apt iff it contains norms (or conditions, criteria, or standards) of verification that are universally recognizable.

Before explaining how Kant presents his view, let me give some general remarks about this scheme: First, the specific norms function as specifications of the kinds of statements that are meaningful within a particular discourse. For example, the statements “It is fine to deface every child you come across” and “Holocaust is a blemish on humanity” both partake in the moral discourse, in contrast to the statement: “DNA-molecules are composed of two polynucleotide chains,” which belongs to the discourse of natural science. Second, the specific norms should be capable of separating true from false statements within a particular discourse. For example, there are evident reasons to claim that the first statement about child defacement is false, while there are evident reasons to claim that the one about holocaust is true. If you deny that you are bad at moral reasoning, which underlines the normative aspect of Kant’s theory of truth and objectivity. Naturally, explaining why that is the case requires a specification and justification of the distinct norms of verification attached to moral discourse.^[18]

This model of objectivity and truth raises suspicion towards all kinds of absolutism that assume one privileged domain of discourse. For example, the scientistic or eliminative type of naturalism, whose semantic-epistemological version asserts that only concepts and judgements that can be exhaustively explained in (or reduced to) a basic natural scientific vocabulary can be called truth-apt or objective.

It might be objected against this admittedly general sketch that Kant restricts (informative) truth and objectivity to empirical judgements or cognition. He does indeed restrict theoretical, scientific cognition to that of “empirical objects,” “appearances,” or “objects of possible experience,” as well as to what can be derived from observations of those objects: non-observable, theoretical objects, such as “magnetic matter” (electrons or magnetic waves would perhaps fare as better examples today).^[19] As is well-known, Kant stipulates a principle of caution with respect to theoretical cognition, which implies a thoroughgoing censure of assertoric^[20] (theoretical), metaphysical judgements about an immortal sole, God, and the cosmos.^[21]

But it is imperative to clarify Kant’s reasoning behind this principle of caution. As he writes, judgements about an immortal sole, cosmos, and God do not contain what he calls a “touchstone [Probierstein]” (KrV, A vii–viii). That term signifies the possibility of judging the “correctness [Richtigkeit]” (KrV, A 65/B 90; A 295–296/B 352), “genuineness [Echtheit]” (KrV, A 65/B 90), or “truth [Wahrheit]” and “content [Inhalt]” (KrV, A 60/B 84) of judgements. A touchstone should be understood as a standard or norm for verifying something’s truth, authenticity, or justified use.

This implies that a sufficient account of the concept of “truth” that goes beyond the correspondence thesis (which Kant famously “presupposes”) requires something more, namely a specification of the concept’s correct use. For short: A universally agreeable norm or “principle of application [Anwendungsgrundsatz],” as Kant himself calls it, through which we can “justify [rechtfertigen]” how the referent of a concept could be given (KrV, A 259–260/B 315). A detailed analysis of Kant’s conception of truth deserves a separate study. Let me just note that Kant takes the correspondence thesis of truth to be a trivial description of what humans mean by saying “is true,” namely that something is true iff it represents its object. But the correspondence thesis is according to Kant not a sufficient or exhaustive account of truth; it does not provide any criteria or norms for separating truth from falsity in concrete judgements, and therefore, no norms or criteria for saying under which circumstances such correspondence could come about.^[22] According to Kant, such norms or criteria are domain or context-sensitive in the sense that they make up a set of context-specific, but universally available (for all rational beings), norms for claiming that a judgement is true or false within a specific discourse. For example, there is a set of universally acceptable arithmetical rules (norms) for evaluating the truth value of arithmetical statements, and those rules are not identical to the rules (norms) connected to empirical judgements. As he writes, there is a particular “criterion” for empirical truth, which is tied to what he calls the “formal conditions of empirical truth” (KrV, A 191/B 236). And elsewhere he specifies what he calls a “mark of empirical truth” (KrV, A 451/B 479; A 651/B 679).^[23]

On that basis, we can call Kant a pluralist about truth in the following sense: There can be (and are) different criteria for verifying whether a judgement is true or false depending on the type of judgement in question. Kant rejects that there could be what he calls a “general” or “universal” criterion of truth – that is an application criterion that is valid for all types of judgements (e.g. the Cartesian criterion of clarity and distinctness or the reductive naturalist criterion of natural scientific reducibility). Such a general criterion “would,” as he writes in KrV, “be that which was valid of all cognitions without any distinction among their objects.” But, as he continues,

it is clear that since with such a criterion one abstracts from all content of cognition (relation to its object), yet truth concerns precisely this content, it would be completely impossible and absurd to ask for a mark [Merkmale] of the truth of this content of cognition, and thus it is clear that a sufficient and yet at the same time general sign [Kennzeichen] of truth cannot possibly be provided. Since above we have called the content of a cognition its matter, one must therefore say that no general sign of the truth of the matter of cognition can be demanded, because it is self-contradictory. (KrV, A 58–59/B 83)

Truth-aptness requires specific “marks” or “signs” that function as regulative norms for a certain type of judgement. That entails a set of anti-realist commitments. Following Michael Dummett’s canonization, what is at stake between semantic realism and anti-realism is whether statements can be ascribed a truth-value (be true or false). Crispin Wright has defined the semantic realist position in the following way:

To conceive that our understanding of statements in a certain discourse is fixed, as Dummett’s realist suggests, by assigning them conditions of potentially evidence-transcendent truth is to grant that, if the world co-operates, the truth or falsity of any such statement may be settled beyond our ken.^[24]

The semantic realist claims that a statement is true or false independently of whether it is possible, in principle, to verify its truth-value. That is, she assumes that there are verification-transcendent truths (and falsities). On the other hand, the semantic anti-realist challenges the global validity of the bivalence principle. She does so by claiming that non-verifiable statements – statements that cannot, in principle, be verified – have no truth-value, which implies that there is nothing of which we could not in principle know. According to the anti-realist, non-verifiable statements have no objective truth-conditions. Some philosophers argue that metaphysical judgements about God, aesthetical judgements, or judgements about the existence (or non-existence) of numbers qualify as such statements. It makes no sense, according to the semantic anti-realist, to ask truth-related questions about those kinds of things, since it is impossible specify a particular set of intersubjectively agreeable norms of verification.

One of Dummett’s many formulations of anti-realism is: “[T]he truth of a proposition consists of its being the case that someone suitably placed could have verified it.”^[25] This passage responds to an objection often confronting anti-realists, namely that statements about the past (e.g. about dinosaurs) cannot be true (or false) according to an anti-realist, since they cannot be directly verified. But as Dummett argues, the objection is inadequate since the circle of epistemic peers should not be limited to present cognizers. The only criterion for the anti-realist is that in order to be truth-apt, a statement can in principle be verified. Interestingly, Kant claims something analogous in several passages. In the following, which appears in the crucial Section Six in the antinomy chapter (entitled “Transcendental idealism as the key to solving the cosmological dialectic”), he implicitly draws an important distinction between (1) statements that can in principle be verified (empirically), and (2) statements that, until a certain historical point in time, have not been, and perhaps for practical reasons will not be, verified (empirically):

That there could be inhabitants of the moon, even though no human being has ever perceived them, must of course be admitted; but this means only that in the possible progress of experience we could encounter them; for everything is actual that stands in one context with a perception in accordance with the laws of the empirical progression. Thus they are real when they stand in an empirical connection with my real consciousness, although they are not therefore real in themselves, i.e., outside this progress of experience. (KrV, A 493/B 521)

It is not sufficient for calling a position anti-realist about a certain domain of discourse (e.g. empirical discourse) that it has not been possible until now to verify a statement within that discourse. What it takes is that such statements are in principle (necessarily) unverifiable.^[26] Paul Abela, who argues against reading Kant as an anti-realist, has suggested that Kant assumes a “recognition transcendent conception of empirical truth.” He argues that for Kant, empirical truth is not evidentially constrained: “Kant’s commitment to the reality of the past and the idea of hidden truths serves to affirm a realist, truth-condition interpretation of empirical realism.”^[27] To support this reading, he invokes the star-example,^[28] where Kant claims we can allow that distant stars “are there to be encountered in world-space even if no human being has ever perceived them or ever will perceive them” (KrV, A 496/B 524). But Kant’s commitment to so-called “hidden” empirical truths does not entail realism about truth, since “hidden” truths such as statements about distant stars or inhabitants at the moon are verifiable in principle. Kant does not believe that some empirically real things cannot be cognized. The assertion that we will never perceive distant stars is not identical to asserting that it is impossible to do so.

We can formulate the two positions like this:

Semantic realism: All judgements – including those whose conditions of verification cannot in principle be specified – have a truth-value.

Semantic anti-realism: Judgements whose conditions of verification cannot in principle be specified have no truth-value.^[29]

A group of prominent Kant-interpreters, such as Peter Strawson^[30] and Jonathan Bennett,^[31] have tried to read Kant as a verificationist about meaning in the, more or less, classical sense whereby the meaning of a sentence is the method of its verification; whereby an expression has no intension and is therefore a meaningless, nonsensical, or incomprehensible expression, if it cannot be verified through an empirical, natural scientific situation (“possible experience” in Kant’s terms). For example, Strawson ascribes to Kant a so-called “principle of significance,” which he renders as the “principle that there can be no legitimate, or even meaningful, employment of ideas or concepts which does not relate them to empirical or experiential conditions of their application.”^[32] This principle can be read in a strong (intension-based) or weak (extension-based) version depending on whether the focus is put on “legitimate” or “legitimate” and “meaningful.” The strong version entails that only empirical concepts or propositions are meaningful (understandable), whereas the weak version states that only empirical concepts or propositions are legitimate (verifiable or truth-apt). Both are inadequate reconstructions of Kant’s position. I return to the weak version in Sections 5 and 6. Regarding the strong version, Kant never suggests that concepts or statements that are not related to anything empirical are necessarily meaningless (have no intension). It is not meaningless, according to Kant, to talk about God, cosmos, or an immortal soul; we can grasp the semantic content of those concepts. He even claims that such concepts can have a “regulative” function by acting as certain heuristic background assumptions, implicitly or explicitly, in our scientific and moral practices. Furthermore, Kant recurrently suggests that a concept not possibly referring to an object can still contain something, namely a “logical function”:

For every concept there is requisite, first, the logical form of a concept (of thinking) in general, and then, second, the possibility of giving it an object to which it is to be related. Without this latter it has no sense, and is entirely empty of content, even though it may still contain the logical function for making a concept out of whatever sort of data there are. (KrV, A 239/B 298)

This implies, given what Kant elsewhere says about concepts and logical functions, that non-objective concepts can still contain “marks” (KrV, B 114; A 728/B 756), which can guarantee a coherent and meaningful, although merely non-assertoric, use of them. So, even a strictly metaphysical concept – which is “entirely empty of content” (i.e. has no possible extension) as he says in the passage cited above – can on Kant’s account have an intension.

But he does assume that a judgement must be verifiable – not necessarily empirically – to be truth-apt. When he talks about “sense [Sinn]” and “significance [Bedeutung],” he means extension or reference.^[33] As Roche^[34] and Kreis^[35] have argued, “Sinn” and “Bedeutung” should generally be read as synonyms. That is evident from a passage where Kant discusses concepts “without sense [Sinn], i.e., without significance [Bedeutung]” (KrV, A 241/B 300). If a concept or proposition has sense and significance, it does not necessarily actually refer to something, but it is about a possible referent; it has a possible extension; it does not totally outrun justification or verification.

How should we understand Kant’s concept of objectivity? His use of it (both objective “validity” and “reality”) is not always very stringent. Generally, a concept or judgement has objective validity if what it refers to is possible, whereas it has objective reality if what it refers to is real. In KrV, Kant often only mentions objective validity and reality in relation to empirically related concepts and judgements. However, we ought not to confuse that with objectivity per se. For short, objectively valid judgements are truth-apt judgements.^[36] Hence, judgements are objectively valid if one can specify their norms of verification. The mode of specification depends on the type of concept or judgement. As we are told in KpV, a judgement or concept has objective validity (or reality) iff what it refers to contains positive determinations that are accessible to all rational beings, which makes it a possible (or real) object for the faculty of (empirical) cognition or a possible (or real) determining ground or principle of the faculty of desire/will (5: 44; 5: 47–48; 5: 50; 5: 104–105). As I argue in Sections 5 and 6, this notion of objectivity conflicts with the weak (extension-based) version of Strawson’s principle of significance, which states that only empirical concepts or propositions are objective.

Based on Kant’s shifting ascription of objective validity (and sense and significance) to both concepts and judgements (or cognitions), we can reasonably assume that he proceeds from some version of the principle of (semantic) compositionality: The semantic value of a complex expression (e.g. a declarative sentence) is determined by the semantic value of the constituents of the complex expression. If a concept has no possible or actual reference (is non-objective, to use Kant’s terminology), a statement containing that concept has no reference (is non-objective or truth-aptless), whereas a statement containing an objectively valid concept is objectively valid on Kant’s account.

In A 90/B 123 in KrV, Kant writes that a concept would be “entirely empty [leer], nugatory [nichtig], and without significance [Bedeutung]” if there were no “rule” for how it could “correspond” to an object. Similarly, he states that a “concept” or “cognition” can only have “objective reality” (“be related to an object”) and hence have “significance and sense [Bedeutung und Sinn],” if its reference (object) is “able to be given in some way.” Otherwise, as he continues, the “concepts are empty, and through them one has, to be sure, thought but not in fact cognized anything through this thinking, but rather merely played with representations” (KrV, A 155–156/B 194–195). This entails that concepts or judgements that purport to expand our cognitive household and outrun any possible justification must be trashed.^[37]

Let us call this, in contrast to Strawson’s restrictive and intension-based principle, Kant’s Real Principle of Significance (RPS):

RPS: A judgement can have no significance or objective validity (truth-value) without a universally recognizable norm for verifying it.

If there is no “rule of synthesis” or “touchstone” for applying a concept – that is no norms or standards for verifying whether the concept can refer and thereby make judgements truth-apt – the concept is “entirely empty [leer], nugatory [nichtig], and without significance [Bedeutung]” (KrV, A 90/B 123). Such rules or norms, though, need not exclusively relate, to use Strawson’s wording, concepts to “empirical or experiential conditions of their application.”^[38]

4 Kant’s three solutions to the antinomy-problem as grounded in or compatible with RPS

Let us return to the antinomy-problem in light of RPS. My argument is that Kant’s solution(s) to the antinomical problem are either grounded in or compatible with the anti-realist framework outlined in the previous section. Although he does not explicitly present it that way, I believe we can reasonably extract three distinct solutions to the antinomical conflict from Kant’s texts. His discussion of the antinomies is, unquestionably, cluttered at times, so let me clarify. It is hardly recognized in the literature, but Kant states different things about the antinomies depending on which of two distinct concepts of the “world” he employs: a transcendental realist or transcendental idealist world-concept. The world-concept at play in A1–A4 is the transcendental realist’s world-concept (TRW), namely the concept of the world in an absolute sense: “the absolutely unconditioned totality of the synthesis of appearances” (KrV, A 481/B 509). When Kant talks about the “absolute,” or the world in an “absolute sense,” he refers to something that is “valid without any restriction,” as opposed to what is “restricted to conditions” and is “merely comparative, or valid in some particular respect” (KrV, A 324–326/B 380–382). This leads us to his first and most substantial solution strategy:

Solution from non-verifiability: The first solution is directly derived from the anti-realist doctrine and pertains to all four antinomies. All the propositions in A1–A4 are (in principle) non-verifiable. They are incompatible with RPS.^[39] According to Kant, it is impossible to verify any proposition involving TRW, because TRW is a concept that, by definition, is not “valid” relative to certain “conditions” or in “some particular respect” but is valid “without any restriction.” It is evoked from a God’s-eye-perspective. All proponents of T1–T4 and AT1–AT4 are what Kant calls transcendental realists who judge about a presumed object without “paying attention to whether and how we might achieve acquaintance with” it (KrV, A 526–527/B 498–499). All sorts of ultimate questions about TRW – quantitative, mereological, causal-related, and modal ones – are without sense and significance because there are no accessible norms for verifying propositions that determine something about absolute everything, since absolute everything cannot be “given in some way” (KrV, A 155/B 194). Therefore, as he says directly about TRW, this concept “must be entirely empty [ganz leer] and without significance [Bedeutung]” (KrV, A 486/B 514, modified translation).^[40] A few pages earlier, he writes:

[O]ne can answer that the question itself is nothing [nichts], because no object for the question is given … Thus here is a case where the common saying holds, that no answer is an answer, namely that question about the constitution of this something, which cannot be thought through any determinate predicate because it is posited entirely outside the sphere of objects that can be given to us, is entirely nugatory and empty [gänzlich nichtig und leer sei]. (KrV, A 479/B 507)

It does not even make sense (it is nichtig) to ask what predicates apply to TRW; every absolute, cosmological question is “nothing, because no object for the question is given.” Regarding A1, Kant writes: “the world does not exist at all (independently of the regressive series of my representations), it exists neither as an in itself infinite whole nor as an in itself finite whole” (KrV, A 505/B 533). It would not even be adequate to say that cosmo-metaphysical judgements, such as the thesis and anti-thesis in A1–A2, are false. On the other hand, they are not even judgements capable of being false.^[41] Several passages in KrV support this anti-realist reading:

1. Concepts or propositions are “absolutely null and void [schlechterdings nichtig]” if they deal “with information [Kundschaft] which no human being can ever get” (KrV, A 703/B 731).

2. “Transcendental illusion … is uncovered and its nullity [Nichtigkeit] is clearly seen into by transcendental criticism (e.g. the illusion in the proposition: “The world must have a beginning in time”)” (KrV, A 297/B 353).

3. The “result in both cases [theses and anti-theses in A1–A2] was something quite empty of sense (nonsense) [Sinnleeres (Nonsens)]” (KrV, A 485/B 513).

4. “If two mutually opposed judgments presuppose an inadmissible condition, then despite their conflict (which is, however, not a real contradiction) both of them collapse, because the condition collapses under which alone either of them would be valid” (KrV, A 503/B 531).

Solution from logical impossibility: If I am correct that all the propositions in A1–A4 about TRW are truth-aptless, why does Kant go on to claim that both the thesis and antithesis in A1–A2 are “false”? Does that not go against the anti-realist proposal? I believe we can extract two reasons solution 2 and solution 3 in the text for calling them false, which are both distinct from the reason he gives for calling them “nugatory.” The first reason is that TRW in A1–A2 is self-contradictory in virtue of being framed as both empirical and non-empirical.^[42] That goes against Abela’s interpretation that there is “nothing analytically contradictory in the mere concept of the world.”^[43] For some reason, Kant is not very clear on this point in the KrV, but underlines it several times in the Prolegomena and in his Price Essay on the Progress of Metaphysics (published in 1804). In KrV he does claim, though, that in the first two antinomies each proposition “searches for the unconditioned among conditioned things” (KrV, A 621/B 649) and calls TRW, at least in A1–A2, an “impossible concept”: “the affirmative as well as the negative part, taken in by transcendental illusion, have as their ground an impossible concept of the object, and then, the rule holds that non entis nulla sunt predicate” (KrV, A792–793/B 820–821).^[44] Coupling that passage with the following lengthier passage from the Prolegomena, in which he proposes that with a self-contradictory and hence logically impossible concept two contradictory propositions can be false simultaneously, there should be sufficient textual evidence against Abela’s interpretation:

Of two mutually contradictory propositions, both cannot be false save when the concept underlying them both is itself contradictory; e.g., the two propositions: a square circle is round, and: a square circle is not round, are both false. For, as regards the first, it is false that the aforementioned circle is round, since it is square; but it is also false that it is not round, i.e., has corners, since it is a circle. The logical mark of the impossibility of a concept consists, then, in this: that under the presupposition of this concept, two contradictory propositions would be false simultaneously; and since between these two no third proposition can be thought, through this concept nothing at all is thought … Now the world, considered as an absolute whole, is thought as a noumenon, and yet by its beginning or infinite duration as phenomenon. If I now assert this intellectual totality of the world, or if I ascribe limits to it qua noumenon, both statements are false. For with the absolute totality of conditions in a sensory world, i.e., in time, I contradict myself, whether I may fancy it given to me in a possible intuition as infinite, or as having limits. (4: 341)^[45]

TRW is “pseudo-empirical”: it purports to be empirical, but it also purports to be something non-empirical (“intellectual”). The transcendental realist thus commits the following category mistake: In TRW, the “world” is both thought of as intelligible (abstractly) and sensible (as something concrete and given). Any proposition containing the self-contradictory concept of a “sensible world in itself” is necessarily false just like any proposition containing the concept of a “square circle” is.

We should compare this with Kant’s reflections in the Amphiboly-chapter in KrV on so-called empty objects without concepts (nihil negativum) or “non-entities [Undinge]”: “The object of a concept that contradicts itself is nothing because the concept is nothing, the impossible, like a rectilinear figure with two sides” (KrV, A 290–292/B 347–349). Such a “non-entity,” which “cancels itself out” as logically impossible, is different from what Kant calls an “empty concept without object” (ens rationis) – a “thought-entity [Gedankending] – which is not self-contradictory, but an “empty concept” that is not “counted among the possibilities” and hence is non-objective. Such a concept has no real possibility, and that is the basis of the Solution from non-verifiability.^[46]

On my reconstruction of Kant’s view, TRW is impossible in both senses: It is a non-verifiable thought-entity and a logically impossible thought. Hence, calling the propositions “false” in virtue of being logically impossible does not promote a semantic realist interpretation. His view is not that judgements whose norms of verification cannot in principle be specified have a truth-value; that there are verification-transcendent truths or falsities. The norm of verification in this case is simply that the judgement itself is logically impossible. Hence, the Solution from logical impossibility is compatible with RPS.

Solution from the distinction between two world-concepts: Kant also gives a second reason for calling the propositions “false,” which however only applies to A1 and A2. From the standpoint of transcendental realism, we are dealing in each antinomy with a straightforward contradiction between opposing predicates applied to one and the same alleged object, but it turns out that judgements containing TRW are both “nugatory” (because verification-transcendent) and necessarily “false” (because containing a self-contradictory concept). It often goes unnoticed that Kant introduces a legitimate concept of the world in a “corrected significance.” That is, in a transcendental idealist significance, where the world is defined as follows: “The All [All], in an empirical signification, is always only comparative” (KrV, A 483/B 511, modified translation). In this sense, the world refers to the indefinite sum of what we can cognize (what is verifiable) in the regressive series of conditions.^[47] Let us call this the transcendental idealist’s world-concept (TIW).

If one reads the antinomical propositions as being about TRW, then they are contradictories (and we would think one of them should be true) since contradictory predicates are applied to one and the same concept. However, none of them are true, because the concept is “nugatory” (truth-aptless) and self-contradictory (making both false). But when invoking TIW, the propositions are not contradictories, but instead function like contraries (adopted from the square of opposition in Aristotle’s’ logic), which implies that they can be, and both indeed are, false. On the classical square of opposition, two propositions are contrary when they cannot both be true but can both, as opposed to contradictories, be false (e.g. “all planets are gas giants” and “no planets are gas giants”). He does not say that the propositions in the antinomies are contraries in this sense, but that they function similarly.^[48] As he writes, when one set of propositions are “contradictory opposites, then one assumes that the world … is a thing in itself” (KrV, A 503–505/B 531–533). But if we “take away this presupposition” – because it invokes a verification-transcendent and self-contradictory concept – then “the contradictory conflict of the two assertions is transformed into a merely dialectical conflict” (ibid.). In such a dialectical conflict, both propositions can be false. Why? It is true that either the proposition that X (e.g. TIW) is infinite or its negation (not-infinite) must be true, since those two propositions exhibit what Kant calls an “analytical opposition.” However, if two predicates are mutually exclusive, but X (e.g. TIW) does not satisfy any of them, then both can be false. They exhibit a “dialectical opposition.” Kant provides the following example to illustrate the point:

If someone said that every body either smells good or smells not good, then there is a third possibility, namely that a body has no smell (aroma) at all, and thus both conflicting propositions can be false. If I say the body is either good-smelling or not good-smelling (vel suaveolens vel non suaveolens), then both judgments are contradictorily opposed, and only the first is false, but its contradictory opposite, namely that some bodies are not good-smelling, includes also those bodies that have no smell at all. (KrV, A 503/B 531)

Staying with the analogy, we could say that TIW has “no smell (aroma) at all.” That is, TIW is neither finite or infinite. The only legitimate (non-empty) concept of the “world” is TIW, the “comparative all,” but to apply any of the predicates (completely infinite, completely finite, completely divisible, completely indivisible) in A1–A2 to TIW would be false predication. Hence, Kant’s Solution from the distinction between two world-concepts is compatible with RPS.

5 The significance of freedom

Whereas Kant concludes that the propositions in A1–A2 are “nugatory” (in virtue of being verification-transcendent) and “false” (in virtue of containing a self-contradictory concept and, when assuming TIW, predicating falsely), he argues that the propositions in A3–A4 (the so-called dynamical antinomies) – all can be true, although only in revised versions (KrV, A 532/B 560; A 562/B 590; 4: 343). Since any satisfactory anti-realist interpretation of Kant must be compatible with his claim that the thesis in the third antinomy about the objective reality of freedom can be true, this section examines that claim.

The third antinomy is a dispute about the causal structure of the universe. The underlying trigger of the dispute is this: If we want a complete answer to why a particular event happens, we must answer all the questions about the causes of this event. It requires a total causal history. This entails the following question that stimulates the third antinomy: Is there a first (uncaused) cause of the universe or is there an infinite chain of (physical) causes? Kant formulates the thesis and anti-thesis in the following way:

T3: “Causality in accordance with laws of nature is not the only one from which all the appearances of the world can be derived. It is also necessary to assume another causality through freedom in order to explain them.” (KrV, A 444/B 472)

AT3: “There is no freedom, but everything in the world happens solely in accordance with laws of nature.” (KrV, A 445/B 473)

However, Kant’s formulations of T3 and AT3 are somewhat misleading.^[49] T3 is explicitly about “causality” and asserts a first, uncaused cause underlying all physical events (“from which all the appearances of the world can be derived”). AT3, on the other hand, does not mention “causality,” but rejects uncaused causes (freedom) and asserts that every physical event (only) happens in accordance with physical (natural causal) laws. Nevertheless, we can reasonably conclude that AT3, cosmologically speaking, is implicitly committed to the opposite claim of T3, namely that there is no first, uncaused cause underlying all physical events, and that the causal chain of physical events must therefore be infinitely long. Therefore, we can reformulate T3 and AT3 in the following way:

T3*: There is a first, uncaused (unconditioned) cause of the world.

AT3*: The world is causally made up of an infinite (total) chain of natural causes.

The “world” here refers to TRW, since both T3* and AT3* pose ultimate descriptions of the empirical world’s causal constitution. Such descriptions are verification-transcendent since we are not justified in thwarting the empirical verification norm of confirming one condition (cause) at a time in the chain of conditions (causes) and refer to the (causal) chain in absolute terms – either in finitist or infinitist form. We have not ultimately answered why one event, X, happens if we have not answered all the questions about X’s causes. But in Kant’s view, it is impossible to answer all the questions about X’s causes and arrive at a satisfactory answer that does not raise further questions. We can only affirm that the regressive chain of causal explanations according to the natural principle of causality goes on “in indefinitum” – not “in infinitum” (KrV, A 510–511–538–539). Hence, just like all the propositions in A1–A2, AT3* does not satisfy RPS. The same is the case for T3*. It is just as impossible to verify the ultimate answer to the cosmo-causal question that invokes a first, unconditioned cause of the world. Such a question, and hence any answer to it, is “absolutely null and void” (KrV, A 703/B 731).

Looking at the three solution strategies presented in Section 4, the first (the propositions are truth-aptless in virtue of verification-transcendence) and partly the second (the propositions are necessarily false in virtue of positing a self-contradictory concept) apply to the third antinomy. The third solution strategy involving predication of TIW can also be said to be operative, but in the opposite way, since the predicates in A3* (“freedom” and “natural causality”) can, according to Kant, be assigned to TIW.

Why does the second strategy partly apply to the third antinomy? For the most part, Kant only ascribes self-contradictoriness to the concept of the world in the first and second antinomy. However, there are good reasons to claim that AT3* does assume a self-contradictory concept of the “world,” whereas T3* does not. T3* does not purport to describe the world in purely empirical or non-empirical (intelligible terms), but states that all the events of the physical world can be derived from a first, uncaused cause. That is not logically impossible (although it is verification-transcendent). On the other hand, AT3* does in fact consider the world as both purely sensible and intelligible – thereby assuming a self-contradictory concept – since it purports to give a purely empirical description of the causal structure of the universe and at the same time abstracts from such a description intelligibly by invoking an absolute (non-empirical) description of it by asserting an infinite (absolute) causal chain.

Why does Kant anyway claim that the thesis and the anti-thesis in A3 “can both be true [beide wahr sein können]” (KrV, A 532/B 560; A 562/B 590; and 4: 343)? It is clearly not in what we could call their global, cosmological, transcendental realist sense. Although the third antinomy arises in the cosmo-metaphysical context, Kant’s intention behind claiming that both T3* and AT3* can be true in what he refers to as a “corrected significance” (KrV, A 532/B 560) is to make a reasonable case for the idea that human beings can both be a part of the natural causal order and free beings.^[50] Whenever Kant claims that T3* can be true, it is in what we could call the local or non-cosmological significance that refers to human agency. We can refer to this as T3**.

Regarding AT3*, Kant stresses that the “correctness” of the natural principle of causality “will suffer no violation” (KrV, A 535–537/B 563–567). That is, as he argues throughout the Transcendental Analytic of KrV, every physical event must have a physical cause. That is how AT3* should be understood in a “corrected significance,” namely in the local sense whereby every empirical (physical) event must have a preceding (physical) cause, which is itself always caused by another (physical) cause, and so on (in indefinitum). We can refer to this as A3**.

The question then remains, as he asks, “whether, despite this, in regard in regard to the very same effect that is determined by nature, freedom might not also take place, or is this entirely excluded through that inviolable rule?” (ibid.), or whether “freedom and natural necessity in one and the same action contradict each other” (KrV, A 557/B 585)? Kant’s strategy in the third antinomy in KrV is primarily negative. He does not intend to “establish the reality [Wirklichkeit] of freedom” (KrV, A 558/B 586), but merely to leave a conceptual space for the concept of freedom, so as to leave room for further positive treatment of it in KpV.^[51] How does he carve out this conceptual space in KrV? He invokes a domain descriptive argument by claiming that every event in the domain of empirical (physical) objects must happen according to natural causality. But this does not exclude that a non-empirical causality (freedom) can have effects in a non-empirical (non-physical) domain. This conceivability argument merely states that it is not logically impossible (because of the domain specificity) to allow for a condition (cause) of an empirical event that is not itself empirical. This means that one can imagine (1) an empirical event or object that is caused by another empirical event or object, and (2) an empirical event or object that is caused by a non-empirical cause – a free will. In short: To say that “X caused Y” does not imply that X and Y are necessarily of the same type.^[52] However, carving out a conceptual space for freedom does not make T3** truth-apt or objective; logical possibility does not entail real possibility.

Freedom is what Kant calls an empirically “empty concept,” since it, per definition, cannot have empirical “significance [Bedeutung]” or “reference [Beziehung]” (KrV, A 55–63/B 79–87; A 146/B 185; A 241/B 300; A 245/B 303) and is “without truth and reference to an object” (KrV, A 489/B 517). Trivially, as he writes in the Prolegomena, ascribing freedom to human actions would be impossible if freedom were understood as objective in the same way as empirical concepts, since “the same would then be confirmed and rejected of one and the same object in the same sense” (4: 343), namely that it both is and is not empirical. But this does not mean that freedom is without sense and significance per se. As I argued in Section 3, Kant defends a pluralistic (although constrained) conception of truth and objectivity. According to the Real Principle of Significance, a concept or judgement has significance (objectivity) if there is a recognizable norm for verifying it. In relation to freedom, that requires a specification of such a norm; an “application criteria” of the concept that is non-identical to, for example, the ones for mathematical or empirical judgements.

Although Kant’s strategy in KrV is primarily negative, it does contain an important passage that lays the ground for his defence of the truth, not merely the truth-aptness, of T3**. In the B-Preface to KrV, he claims that we must be open to thinking some objects (i.e. human actions) in a “twofold meaning [zweierlei Bedeutung].” If the two propositions – that the “will is free” and that “it is simultaneously subject to natural necessity” – take the human subject “in just the same meaning,” an evident contradiction arises. But, as he writes, if the “critique has not erred in teaching” that some objects

should be taken in a twofold meaning [zweierlei Bedeutung], namely as appearance or as thing in itself … then just the same will is thought of in the appearance (in visible actions) as necessarily subject to the law of nature and to this extent not free, while yet on the other hand it is thought of as belonging to a thing in itself as not subject to that law, and hence free, without any contradiction hereby occurring. (KrV, B XXVII–XXVIII)

It goes beyond the scope of this article to discuss in detail Kant’s distinction between appearances and things (considered) in themselves. However, I take there to be sound textual and philosophical reasons not to read Kant’s distinction as an ontological distinction between two different ontological realms or objects.^[53] Rather, I believe it should be read as a distinction between different ways of considering or thinking about certain objects (namely actions).^[54] One of the clearest pieces of evidence for this reading is found in Kant’s Opus postumum: “The difference between the concepts of a thing in itself and that of an appearance is not objective, but merely subjective. The thing in itself (ens per se) is not a different object, but another consideration [Beziehung] (respectus) in the imagination of the same object” (22: 26).^[55]

However, this reflexive operation of considering or thinking about certain objects in a twofold meaning does not grant a concept like freedom objective validity. In his Price Essay, Kant moves beyond the reflexive claim and writes that judgements can be true or objective from both standpoints, even though they seem mutually exclusive. Just like two subcontrary judgements in classical (Aristotelian) logic can both be true if the subject (the human action, say) is “taken in a different significance [Bedeutung]” or “sense [Sinn]” (20: 291, modified translation).^[56] In the passage on the third antinomy, he writes:

Third, the thesis and antithesis may also contain less than is needed for logical opposition, and may both be true – like two judgments in logic that are opposed to one another merely by difference in their subjects (judicia subcontraria) – as is actually the case in the antinomy of the dynamical principles; if, that is, the subject of the opposing judgments is taken in a different significance [Bedeutung] in each; for example, the concept of cause, as causa phenomenon in the thesis: All causality of phenomena in the world of sense is subject to the mechanism of Nature, seems to stand in contradiction to the antithesis: Some causality of these phenomena is not subject to this law; but such contradiction is not necessarily to be met with there, since in the antithesis the subject can be taken in a different sense [Sinne] from that in the thesis – the same subject, that is, can be conceived [gedacht] as causa noumenon, and then both propositions may be true, and the same subject, qua thing-in-itself, be free from determination by natural necessity, which qua appearance, with respect to the same action, is not free. (20: 291–292, modified translation)^[57]

Whenever Kant mentions that T3 and AT3 “may be true,” it is in the local or non-cosmological, corrected significance that describes specific actions (“the same subject” or “the same action”): T3** and AT3**. It is important to note that “sense” and “significance” here still mean possible extension or reference – not intension. Otherwise, one could in principle think and ascribe objective validity to an object in an unrestricted number of senses (intensions). Instead, Kant argues that we are justified in regarding human actions as (also) being free if we can specify a set of norms or application criteria for the concept of freedom. Whether and how that is possible is what KpV revolves around, which I will sketch briefly in what follows. I do not aim to defend every element of Kant’s argument, but to outline its structure in order to extract the kind of anti-realism I believe he endorses based on his view that the third antinomy is truth-apt.

Kant provides what we could call a transcendental argument for the truth of T3** by proceeding from our self-understanding as moral agents (what he refers to as a “Factum der Vernunft”):

1. Human beings are imputable (moral) agents.

2. A necessary condition for this is that human beings can act from causality of freedom (that T3** is true).

3. Hence, human beings can act from causality of freedom.

We can understand the “fact of reason” as the, according to Kant, irrefutable claim, authenticated through common moral judgement, that whenever human beings deliberate about what to do and why to do it, they can recognize an unconditional and obligatory reason to act (a moral reason) that is unmistakably distinct from agent-relative (happiness-based) reasons. According to Kant, this “proofs” the objective reality of the concept of freedom and hence the truth of T3**.

In a famous footnote in the Preface to KpV, moral consciousness is described as “the condition under which we can first become aware of freedom.” The moral law is the ratio cognoscendi of freedom. Later, he explains this by turning to what he refers to as (moral) experience: “But experience also confirms this order of concepts in us” (5: 30). He then provides the famous gallows example, where a person is demanded to give false testimony against an innocent person or be executed and realizes that the moral action to not give false testimony is possible for him (ought implies can): “He judges, therefore, that he can do something because he is aware that he ought to do it and cognizes freedom within him, which, without the moral law, would have remained unbeknownst to him” (5: 30). Kant thereby derives the objective, practical reality of freedom from the objective, the practical reality of the moral law. As he writes, the “reality” of freedom “is proved by an apodictic law of practical reason [ML] […]” (5: 3). Moreover, he states that the moral law gives “content” (5: 6), “significance” (5: 50), and “objective and, though only practical, undoubted reality” (5: 49) to the concept of freedom, and that the fact of reason “establishes” and “furnishes reality” (5: 6) to it.

As already noted in Section 3, Kant argues that a judgement (or concept) has objective validity (or reality) iff what it refers to contains positive determinations that are accessible for all rational beings, which makes it a possible (or real) object for the faculty of (empirical) cognition or a possible (or real) determining ground or principle of the faculty of desire/will (5: 44; 5: 47–48; 5: 50; 5: 104–105). Hence, a judgement can be called objective if it is possible to specify its truth criteria, or what Kant also calls “sources of cognition,” that is the necessary conditions for ascribing a type of judgements cognitive significance or objective validity: To ascribe a concept or judgement “objective validity,” something “more” is required, but, this “more,” however, “need not be sought in theoretical sources of cognition [Erkenntnisquellen]; it may also lie in practical ones” (KrV, B XXVI).

If the moral law is a possible (or real) determining ground of the will, which Kant believes is an evident fact of the practical lives of human beings, it has objective validity (or reality). And it is an analytic truth, Kant assumes, that an “ought”-representation comes with the possibility of acting upon it (ought implies can), since the moment an action is, in principle, impossible for me to carry out, it can no longer be required of me. That satisfies the defined criteria for objective reality since the presence of an unconditional ought in our lives makes the moral law (and freedom) positively “specified [angegeben]” (5:56) or “determined from a practical perspective” (5: 105) as a “practical concept” with “practical use” (5: 6; 5: 135) and a “real application [wirkliche Anwendung]” because it is “exhibited [darstellen] in concreto” in the course of practical deliberation (5: 56).^[58] We are warranted in asserting the truth of T3** from this practical “standpoint” (4: 450; 4: 458) or “perspective” (5: 105). T3** turns out not only to be truth-apt (objectively valid) but true (objectively real). Any satisfactory semantic anti-realist reading of Kant’s transcendental idealism ought to be compatible with that assumption. What I have rendered as Kant’s Real Principle of Significance, in contrast to what I called the weak version of Strawson’s empiricist principle of significance (only empirical concepts or judgements are objectively valid), is compatible with that assumption.

6 Conclusion: Was Kant really a semantic anti-realist?

To conclude, I will present and reply to a set of criticisms of reading Kant as a semantic anti-realist with the purpose of summarizing the perspectives of an anti-realist approach as well as the ways in which my interpretation differs from other interpretations.

In an article comparing a set of anti-realist views with Kant’s, Lucy Allais acknowledges “an element of the Dummettian position that is helpful for understanding Kant’s position,” a position she describes as the view that “claims which transcend possible experience cannot be true of the world of experience.”^[59] But she concludes that there are “profound differences between the Kantian position” and the semantic (Dummettian) “versions of anti-realism”^[60] and that we consequently should be careful of labelling Kant as a semantic anti-realist. Her main objections are the following:

Objection (1): Anti-realism is committed to a rejection of the principle of bivalence (PB): Every meaningful statement is either true or false. However, Kant does not reject bivalence, although it could appear so in the antinomy chapter when he denies that the world is both infinite and finite in size in A2. He does not reject bivalence, because he concludes that both the thesis and anti-thesis in A1–A2 are “false.”

Objection (2): his transcendental idealism “is not a theory of meaning or truth,”^[61] and to “to this extent”, she says, “it is clearly not like Dummettian anti-realism.”^[62]

Objection (3): Kant does “believe in verification-transcendent reality.”^[63] Hence, to ascribe him the anti-realist view that judgements lacking verifiable truth-conditions can have no truth-value is wrong.

Objection (4): Kant finds global anti-realism to be “arrogant,” as she puts it.^[64] That is, he finds it arrogant to claim that there is nothing which we cannot know by extending empirical verifiability to everything that exists.

Reply to (1): The feasibility of this objection comes down to two things: (1) How we cash out “meaningful statements” in PB, and (2) Kant’s reasons for calling all propositions in A1–A2 “false.” If “meaningful” equals “non-verifiable,” an anti-realist need not be committed to rejecting PB: All meaningful statements are either true or false. But if a rejection of PB involves rejecting that all intensional, comprehensible statements have a truth-value, then Kant rejects PB, since some statements might be comprehensible, but are “absolutely null and void,” hence lack significance and truth-value.^[65] Regarding (2): As I argued in Section 4, Kant gives two reasons for calling the propositions in A1–A2 “false”: the concept of the world in A1–A2 is logically impossible (hence judgements containing it are necessarily false) and on the legitimate concept of the world (TIW), the predicates in A1–A2 (completely infinite, completely finite, completely divisible, completely indivisible) are not satisfied. Those two reasons for calling the propositions “false” are compatible with the outlined anti-realist framework in this article.

Reply to (2): To answer this objection requires some strong methodological elaborations on how to interpret past philosophers which exceeds the ambitions of this article. However, the short answer is this: I agree that Kant’s critical project should not first and foremost be understood as a semantic project. That does not mean, though, that (1) a semantic reading cannot be a valuable way of understanding Kant’s position or (2) that a semantic reading has no textual support. There are indisputable overlaps in Kant’s texts between epistemology and semantics that have been undertreated with respect to certain aspects of Kant’s thinking. In particular, a semantic reading of the antinomy chapter is not unfounded, since he repetitively refers to “falsity,” “truth,” “judgements,” “empty concepts,” “sense,” and “significance” in order to explain what kind of concepts that have and lack cognitive value. Such a reading can hopefully shed some light on parts of Kant’s oeuvre that are still heavily disputed; in this case, his solution(s) to the antinomy-problem and the related claim that freedom is an objectively valid concept.

Reply to (3): There is no consensus within Kant studies on his view of transcendent objects and their ontological status (Allais refers to “verification-transcendent reality”). Following Allison’s non-metaphysical interpretation of Kant’s transcendental idealism, I believe there are good reasons to claim that Kant is not necessarily committed to verification-transcendent reality. As I stressed in Section 5, there is sound textual evidence against reading the distinction between appearances and things in themselves as an ontological distinction between two different ontological realms of objects. Instead, we could read Kant’s “dualism” as a normative dualism of descriptions. In contrast to Allais, I therefore find it very doubtful to claim that Kant is “a transcendental realist with respect to things as they are in themselves.”^[66] A transcendental realist is someone who judges about a presumed object without “paying attention to whether and how we might achieve acquaintance with” it (KrV, A 526–527/B 498–499). Kant is not a transcendental realist. Furthermore, Allais claims that, according to Kant, “if an object exists, then we can, in principle, have experience of it. In other words, Kant rejects experience-transcendence for the appearances of things.”^[67] Regarding the latter sentence, it is trivially true that Kant rejects experience-transcendence for the appearance of things since empirical objects can only be verified through experience. But regarding the first sentence, that is only correct on a very restrictive conception of objects or objectivity, namely what I have called Strawson’s weak version of the principle of significance. But to say that for an object to exist, we must in principle be able to experience it, does not square with Kant’s broader use of objectivity. If my interpretation is correct, Kant’s anti-realist commitment is that verification-transcendent concepts or judgements lack objective validity (truth-valuedness), but that what can be verified is not restricted to empirical (experiential) concepts or judgements.

Reply to (4): Allais writes:

Just as most realists do, Kant thinks that the claim that there is nothing of which we cannot have knowledge, in general, is inexplicable and arrogant; on the contrary, he says that transcendent ideas serve, in part, to restrain the understanding’s arrogant claims, namely, that (since it can state a priori the conditions of the possibility of all things it can cognise) it has thereby circumscribed the area within which all things are possible’ (CJ168, also B167). Kant does not agree, in general, with the claim that we can have experience of everything that exists.^[68]

The idea seems to be that Kant finds it “arrogant” to claim that empirical concepts or judgements are the only ones that are “possible” or that only what can be experienced “exists,” to use Allais’ wording. My proposed framework agrees with that. Here, we could adjoin the passage in KrV where Kant writes that the “dogmatic empiricist,” who believes that only empirical concepts are objective and “boldly denies whatever lies beyond the sphere of its intuitive cognitions,” makes a blameable “mistake of immodesty” (KrV, A 471/B 499). Allais’ objection to anti-realism relies on a certain version that is committed to the weak version of Strawson’s principle of significance: Only empirical concepts have significance or objective validity. But that is simply not Kant’s view. It does not correspond with what Kant himself says about the objective validity of freedom in the third antinomy and the objective reality of freedom in KpV. In contrast, what I have called Kant’s Real Principle of Significance can fulfil a necessary requirement for any sound anti-realist interpretation of his general position: to give a plausible explanation of his view that human freedom is an objectively valid and real concept. He does not find it “arrogant” to claim that there is nothing of which we cannot in principle “have knowledge,” as Allais writes. The underlying pluralism of RPS simply points out that there can be (and in fact there is) more than one type of objective judgement.