ROBERT HANNA | Cognition, Content, and the A Priori. A Study in the Philosophy of Mind and Knowledge | Oxford University Press 2015


By Dennis Schulting

Renowned philosopher and Kant scholar Robert Hanna’s most recent book Cognition, Content, and the A Priori. A Study in the Philosophy of Mind and Knowledge is probably his most ambitious to date. It’s also an exhilarating read: an exemplarily lucid, fast-paced philosophical thriller replete with pithy nuggets of philosophical substance and in addition, as the allspice added to the main ingredient, jesting but no less scathing in-house attacks on the elementary dogmas set in stone and religiously patrolled by the vanguard of analytical philosophy. In his critique published earlier last January, David Landy focussed on Hanna’s topic of nonconceptualism, a main theme of the book. I have dealt with that subject in the past and also again in my own forthcoming book (Schulting 2017), and I think I have said enough about it. I personally believe Kantian nonconceptualism is now dead and buried, and the less said about it, the better. I say this with not a little irony, only very recently having edited myself an entire volume on the topic. I firmly believe though that whatever continues to be written about it can only be a repetition of past moves: it seems to me that every conceivable worthwhile position has been made sufficiently clear in the existing literature (see e.g. the most recent here).

For this notice, I therefore set out to concentrate on Chapters 4 and 5, which consider the far more interesting more strictly logical side of the subjects dealt with in Hanna’s book (most of my comments concern Chapter 4). I came away with the strong feeling—I always suspected it—that (a) the likes of Quine and especially Kripke and their latter-day acolytes can without regret be cast out from the canon, regardless of the misleading sophistication of their and their disciples’ logico-metaphysical puzzles, and, more intriguingly even, that (b) analytic philosophy as we know it is in fact not analytic philosophy properly speaking, or at least, cannot account for itself—which is kind of odd, to say the least, for an approach to philosophy that prides itself on its rigour. Hanna’s book should be read by any philosopher, in particular, self-declared analytic philosophers, worth their salt, especially as there are some substantial things about a core aspect of philosophy, certainly analytic philosophy, at stake that Hanna discusses in this important, somewhat unorthodox book, and which I would like to highlight here in terms of the Kantian context of the book and invite Hanna to cast some more light on.

As said, back in January my co-critic David Landy discussed Hanna’s well-known nonconceptualist construal of Kant’s philosophy of perception, which Hanna rehearses in great detail in Chapter 2 of the book, and I myself have dedicated quite a few pages to critically discussing his views in this regard, beginning in my 2010 article in Dutch (Schulting 2010),[1] in my German language article in Kant-Studien (Schulting 2015), and most recently in my forthcoming second monograph on the Deduction (Schulting 2017, ch. 5). In Chapter 3, Hanna presents his theory of what he calls “radically naïve realism”. Further, after a chapter on the Benacerraf dilemmas, in the remaining chapters of the book Hanna dedicates quite some space to delineating a new, Kantian theory of intuition. Interesting and important though they may be, I have nothing to say here about these topics. Nevertheless, in the context of the subject of my critique here, I do need to briefly address the topic of nonconceptualism in my little réquisitoire towards the end of the paper (Section III).


In Chapter 4, Hanna argues for a return of the ‘analytic-synthetic’ distinction in logic, more in particular for a reappraisal of the synthetic a priori, dismissed out of hand by analytic philosophers, almost by default—a rejection that is perhaps even definitional of analytic philosophy, which points exactly to the problem that Hanna highlights, namely, that by dismissing this distinction, and a fortiori the notion of the synthetic a priori, analytic philosophy can’t even explain its own status as analytic philosophy, for it has no satisfactory means of explaining analyticity.[2] By the same token, in dismissing, or at least downplaying the importance of, this distinction as well as the idea of the synthetic a priori, it can’t explain truths other than strictly-logical or merely-conceptual truths, that is, it can’t explain truths about non-conceptual facts in the world.

As Hanna notes, “Quine rejected the analytic-synthetic distinction […] on the grounds that it could not be reductively explained in other terms, and proposed its elimination” (p. 147). By contrast, Hanna wants to argue that “the analytic-synthetic distinction itself can be adequately explained, in a contemporary Kantian way, in terms of intentional idioms and intentionality, or more precisely, in terms of mental content and human rationality” (p. 148). He argues convincingly—and I shall not attempt to repeat his meticulously laid out arguments for it here—that the possibility of analytic philosophy itself is hard to explain if the analytic-synthetic distinction “were either unintelligible or indefensible” (p. 151). What’s more, “the very idea of a semantic content would go down” and “then, like so many dominoes, the very ideas of belief, cognition, thought, understanding, justification, knowledge, intentionality, and human rationality […] would all go down, too, since all these notions inherently involve and basically presuppose the notion of semantic content” (p. 151). This is quite a far-reaching and devastating outcome if what Hanna claims were true, so much so that one wonders how Quine, not to speak of all those who came after him, could ever have been so rash as to dismiss the analytic-synthetic distinction, or, more in particular, the synthetic a priori.

By way of clarification, Hanna writes that

by the notion of “a robust analytic-synthetic distinction” I mean a version of the analytic-synthetic distinction that explanatorily includes and fully preserves an essential difference between (i) analytic truths, which are inherently necessary and a priori, and (ii) synthetic truths, with the possibility also being explicitly left open of explanatorily including and fully preserving another essential difference between (iia) synthetic necessary and a priori truths, and (iib) synthetic contingent and a posteriori truths. (p. 152)

The truths under (iia) and (iib) are, respectively, “non-logically, essentially non-conceptually, or strongly metaphysically necessary synthetic a priori truths”, and “contingent synthetic a posteriori truths” in virtue of nonconceptual content. Hanna thinks there is “nothing less than [a] categorically sharp contrast” between a priori truth in virtue of conceptual content and a posteriori truth “as represented by autonomous essentially non-conceptual content”, that is, content whose “truth is never in virtue of conceptual content” (p. 153–4). Hanna also criticises the view that “there is one and only one basic kind of necessity, and thus only one basic kind of necessary truth (=modal monism)” (p. 159). I return to this so-called “categorically sharp contrast” between (analytic or synthetic) a priori truths and a posteriori truths in my critique towards the end of this note.

Further, Hanna says that according to the orthodoxy of analytic philosophy

any statements that are discovered to be a priori and necessary “must be” analytic or conceptual necessities, even if they do not fit any classical profile of analytically or conceptually true statements, and even if in fact they also satisfy the classical criteria of synthetic apriority. But such statements are “analytic” or “conceptual” truths only in a misnomer-based, Pickwickian, or so-called sense, simply because they deviate importantly from all the classical conceptions of analyticity and conceptual truth, and because they also satisfy the classical criteria for synthetic a priority. Strictly speaking, then, they should be called “synthetic a priori statements”, although it would perhaps be even more accurate to call them “schmanalytic” statements. (p. 163)

This is the enigma of contemporary orthodox analytic philosophy: either a priori and necessary truths are analytic truths or truths in virtue of essentially conceptual content, or if they are not, then truths are either not a priori or not necessary (think contingent a priori statements or Kripke’s necessary a posteriori statements; see further below). “Schmanalytic” statements are then statements that are not strictly speaking or purely analytic statements, but they are nonetheless deemed to be analytic in some sense since they can’t be synthetic a posteriori—while they obviously aren’t analytic in the standard sense (analytic purely in virtue of logical or conceptual content). They are therefore in fact “so-called analytical”, as Hanna puts it (p. 163). This is an explanatory embarrassment that directly results from the rejection of synthetic a priori truths: “schmanalytic” statements are not purely analytic, and not synthetic a posteriori, but a middle term like the synthetic a priori is not acceptable, hence they must be analytic in some alternative sense.

This quandary is made clear by Quine’s contradictory position on logic and his revisability principle. Hanna points out that Quine makes a strict distinction between “sheer logic” (the logic of analytic truth strictly speaking) and (the logic of) all other truths (pp. 167–8), but it becomes clear that this leads to contradictions in Quine’s account. That is to say, Quine’s Universal Revisability Principle is in tension with his Sheer Logic Principle (p. 166). Given Quine’s Universal Revisability Principle and his Sheer Logic Principle, Hanna says, “no statements are unrevisable and yet some statements are unrevisable, and the law of non-contradiction is both revisable and unrevisable”. This inconsistency is what Hanna calls “Quine’s Predicament”, which, as Hanna infers with impish glee, “goes like a dagger into the very heart of Quine’s overall critique of the analytic-synthetic distinction”. At the risk of seeming to lay it on rather thick, Hanna concludes that, “in effect, Quine’s Predicament is Quine’s committing cognitive suicide by logical self-stultification” (all quotations from p. 166). All in all, Quine’s Predicament is “philosophically dire” (p. 171). With such an indictment of a famous philosopher’s legacy, one asks oneself why Quine was and still is[3] ever so considered a major deity in the analytical pantheon.

Hanna has caught Quine out, for if there is, as Quine admits, a clear distinction between “analytically true statements of elementary logic” and “all other truths” (p. 168), and “radical indeterminacy does not hold for words that express the classical truth-functional logical constants” (p. 176),  then it seems there is “an intelligible and defensible analytic-synthetic distinction after all” (p. 168). Hence, there are no good grounds for dismissing out of hand the analytic-synthetic distinction.

Nevertheless, without jumping to the conclusion that since Quine is wrong about the analytic-synthetic distinction, Kantians are right to hold that there is such a distinction, Hanna acknowledges the possibility that rejecting Quine’s critique might still leave open the possibility of

rejecting the existence of an intelligible or defensible analytic-synthetic distinction, if it could be shown that analyticity, apriority, and necessity can be detached from one another. […] Kripke and early Putnam offer widely influential arguments for the detachability of the necessary and the a priori, in both directions, from the existence of necessary a posteriori statements and contingent a priori statements. (p. 177)

But after thorough analysis of the available putative solutions to the question what analyticity is—I can’t match the sophistication and analytical (!) precision with which this is carried out in the book—Hanna comes to the conclusion that the explanations of Russell, Katz, Boghossian, Juhl & Loomis et al are all “merely theories of schmanalyticity, not theories of analyticity” (p. 182). According to Hanna,

it is very unclear whether appealing to stipulation in order to explain analyticity* [as do Juhl & Loomis], in the end, is any more explanatory than simply appealing to intentionality in order to explain analyticity* […]. [A]nalyticity* is still schmanalyticity, not analyticity. So my most general worry about the post-Quinean accounts of analyticity is that Russell, Katz, Boghossian, and Juhl and Loomis, for all their philosophical ingenuity, insight, and rigor, have simply changed the subject. (pp. 181–2)

Startlingly—since this looks like it may sound the death knell of the entire Kripkean and post-Kripkean oriented or influenced modal metaphysics industry—Hanna propounds that

all the arguments offered for the existence of necessary a posteriori statements, contingent a priori statements, and analytic contingent statements are unsound, but also that there are really no such things as the necessary a posteriori, the contingent a priori, and the analytic contingent. All three of these pseudo-concepts must be eliminated. (p. 182)

I will freely and fully admit that this contemporary Kantian eliminativist project in particular is a very strenuous task, given the canonical—indeed, almost biblical—status of the fictional conjoined philosopher Kripke-Putnam’s writings in recent and contemporary Analytic philosophy, and especially Analytic metaphysics. But if they’re wrong, they’re wrong, and somebody needs to point out the Emperor is actually wearing no clothes. (p. 183)

This much is clear from reading Hanna’s account: contemporary theories about analyticity (or “analyticity*” or “so-called analyticity”) are just so many vain attempts to avoid having to admit defeat to the Kantian, who has a perfectly valid solution to the problem of explaining the difference between analyticity and other a priori and/or necessary truths, namely the synthetic a priori.

But the operative question is: what then does the notion of the synthetic a priori accomplish in terms of explaining analyticity that all those valiant efforts by so many lights in logic and metaphysics couldn’t? Here I found Hanna less helpful. Hanna rightly relates Kant’s analytic-synthetic distinction to his content dualism, which holds that there are “two essentially distinct but complementary kinds of intentional content or mental content”, namely conceptual content and intuitional content, whereby “analyticity is grounded on conceptual content and syntheticity is grounded on intuitional content” (p. 198). But to present the analytic-synthetic distinction as neatly mapping onto the concept-intuition distinction is too easy, I think, and probably incorrect.

What is meant by “conceptual content”? Sure, “analyticity” is grounded on conceptual content if by “conceptual content” is meant the relation between two or more concepts, whereby (1) that relation is ultimately one of subordination, in virtue of analytic unity, in a categorical judgement as the basic form of any logical relation (in Kant’s logic) and (2) analytical relations between concepts can be explained by conceptual content only, for any actual reference to objects, via intuitional content, is otiose.

But does that imply that syntheticity is not based, at least partly, on conceptual content in any sense at all? Clearly, for Kant at least, syntheticity should in some sense also be related to analyticity—not in the sense that any synthetic judgement is in the last instance explainable by conceptual analysis alone, which would get us right back to square one, namely, the Leibnizian position that Kant criticised, but in the sense that synthetic judgements are judgements with conceptual content, which is constituted by the way in which concepts are always related by virtue of analytic unity, as Kant puts it (A79/B104–5; I discuss this passage below), as much as they are also grounded on intuitional content.

There is another sense in which the analytic-synthetic distinction cannot be seen as simply mapping, one-to-one, onto the concept-intuition distinction: the synthetic a priori, which Hanna rightly associates with the analytic-synthetic distinction, is however not equivalent to that distinction. I think Hanna would concur with this, but it wasn’t clear to me how he sees them related. They cannot be equivalent theses or notions, for the synthetic a priori also in some way, to put it in Kantian terms, grounds the possibility of analyticity, as one term in the distinction relation, something that Hanna himself appears to suggest in the way in which the synthetic a priori is the substitute for whatever is not analyticity strictly speaking (“so-called analyticity”, “analyticity*”), and also thereby first enables the demarcation from sheer logic, analyticity strictly speaking. To put this differently, the synthetic a priori is the enabling condition for differentiating analyticity from syntheticity and vice versa, that is, for making the distinction between the two, and hence, it in some sense explains analyticity.

Of course, the synthetic a priori is a necessary condition only, obviously not a sufficient condition of analytic statements or truths, which if it were would collapse the distinction between the the synthetic a priori and the analytic again. It would also conflate two distinguishable levels: The synthetic a priori operates at the transcendental level; analytic statements (or synthetic a posteriori statements, for that matter) do not. The synthetic a priori thus enables analyticity, by grounding the semantic content of analytic statements, their own aboutness, in the same way that statements about the world are enabled by transcendental truth (synthetic a priori truth). All this doesn’t mean that the rules of general or formal logic (“sheer logic”) or the criteria of analytic truths, truths in virtue of logical or conceptual content alone, are not sui generis. Here, there is a perfect symmetry between synthetic (a posteriori) and analytic statements: the synthetic a priori doesn’t provide the sufficient condition for both their semantic content, but only a necessary condition, namely the possibility of having semantic content in the first place—I expand on this in the next section.


How is the logical content of an analytic truth, e.g. a is not not-a, explained? Or is an analytic truth an unexplainable basic fact? In a sense it is; there is no determining reason or ground for the truth of the principle that a is not not-a. And this holds in general for all logical truths: all logical truths are explainable from basic analytic truths, in particular the principle of non-contradiction. There is no further basic fact from which these analytic truths can all be deductively derived, and they can’t be explained other than by means of pure analysis, reductively down to the principle of non-contradiction.

However, it is striking that Kant suggests, in an oft-noted-but-never-discussed passage in a footnote in the B-Deduction, that there is a way that logic as a whole is grounded in a more fundamental ground. He writes:

And thus the synthetic unity of apperception is the highest point to which one must affix all use of the understanding, even the whole of logic and, after it, transcendental philosophy (B134n.; boldface mine)

Thought, or more precisely, the principle that constrains thought, namely the original synthetic unity of apperception, is the a priori ground of logical and conceptual truths as well as non-logical truths, which are about objects in the world, or anything that is not logically or merely conceptually analysable.[4]

That this is so can be demonstrated with the help of the very first lines of §16 of the B-Deduction, where Kant advances the well-known principle of apperception, namely “the I think, which must be able to accompany all my representations”. Kant adds that if this were not a necessary possibility, “something would be represented in me that could not be thought at all, which is as much as to say that the representation would either be impossible or else at least would be nothing for me” (B132). The phrase “…which is as much as to say that the representation would […] be impossible” (IMP) is sometimes taken to refer to a contradictory representation or idea or proposition, i.e. a representation expressive of a typically logically contradictory proposition a is not-a. For example, Dietmar Heidemann (2012) reads IMP this way. In the case of a contradictory idea or statement, Heidemann argues, the “I think” is “unable to accompany representations contentswise [sic]”. He continues:

Here representations are ‘impossible’ if they are contradictory or illogical like the idea of a ‘round circle’ [I suppose he means the idea of ‘squaring the circle’, D.S.], the mathematical equation ‘5+7=13’, or the thought that the reader of this article does not exist.

Heidemann says that

I might be able to somehow mentally generate such contradictory ideas yet I am not able to make them intelligible, i.e. to accompany them with the ‘I think’, since they are logically ‘impossible’. (Heidemann 2012:51–2)

This doesn’t make sense. How can I “mentally generate” contradictory ideas yet not “make them intelligible” to myself by accompanying them with an ‘I’-thought, that is, while not thinking them? It seems to me that if I can’t make thoughts intelligible to myself, I’m not able to think them in the first place. For Kant, the whole point is to make it clear that thinking a thought means to accompany that thought by an ‘I’, that is, to make that thought intelligible, transparent to that ‘I’. (The modal construction of the proposition shouldn’t confuse one into believing that it only refers to a potentiality; it rather signals the modal structure of a conditional: I think a representation [a thought] if and only if I accompany it.)

So it depends on what Heidemann means by “mentally generating”. If I could “mentally generate” a contradictory proposition such as CON:

Fire and non-fire are identical

but not make the contradictory nature of CON intelligible to myself by effectively thinking it, since I do not occurrently accompany the representations ‘fire’, ‘non-fire’ and ‘identical’ and combine them, through synthesis, by virtue of the copula ‘are’, how could I ever know that the proposition is contradictory? It seems that Heidemann’s belief here is itself a contradiction of sorts, i.e. a transcendental-logical impossibility: I cannot generate a contradiction and understand it as a contradiction, while not thinking the representations that putatively amount to the contradiction. Well, of course I could accidentally let out two or more representations that just happen to be contradictory, by calling them out or silently imagining them: “fire!”, “non-fire!”, “identical!”—though I would also then have to combine them accidentally by involuntarily letting out a copula (“are” in this case), otherwise my calling them out is just a random concatenation of representations, not even an accidental contradiction. Clearly, this is a tall order.

But I could never do so with a logical contradiction of which I know that it is a logical contradiction. A logical contradiction, if I think or “mentally generate” it, is not an impossible thought, as Heidemann suggests, it is rather a necessarily possible thought which has a semantic content that expresses a conceptual contradiction between two representations (or between two statements). The only impossible thoughts are thoughts that are not my (actual or possible) thoughts, or more radically, thoughts that do not belong to any possible thinker. Impossible thoughts therefore cannot exist, either “for me”, as Kant says (B132), or at all. If contradictions were indeed literally impossible thoughts, nobody could think contradictions.

Therefore, I believe Kant is here, at the start of the B-Deduction, not alluding to logical contradictions or impossibilities, but rather to a metaphysical impossibility which concerns the identity of the self and the unity of all of her own possible representations—but let’s bracket that issue here (I say something more on this later below).

At any rate, and here I differ fundamentally from Heidemann and others who espouse similar views, logically contradictory thoughts, ideas or propositions have phenomenally positive semantic content—they are phenomenally present to one, when they are thought, so must be thoughts that are accompanied by the ‘I think’ for any case of the selfsame ‘I’ having such thoughts. When the principle of apperception governing our thoughts is concerned, we are operating at the transcendental level of the analysis of the faculty of the understanding, of thought itself. Logic strictly speaking, “sheer logic”, does not operate on that level. Logical thoughts (propositions, statements, truths, principles or what have you) have semantic content just as much as non-sheerly-logical, conceptual thoughts about objects. When I think, utter or more precisely think about a contradictory proposition such as the proposition “Fire and non-fire are identical” (CON) my thought has positive semantic content, namely the logical content that says that this proposition is contradictory and violates the principle of non-contradiction. My thought of CON is itself not contradictory and therefore impossible.[5] Rather my thought of CON itself is necessarily possible if I were indeed to utter, contemplate, or reflect on CON, it being the positive content of my occurrent thinking. Thought itself is a transcendental condition of having a logical idea (or any idea, any semantic content), uttering a sheerly logical proposition expressing (the violation of) a logical principle such as PNC, as much as it is a necessary condition of any objectively valid contentual idea, proposition, judgement or statement.

I should note at this point that Kant himself does appear to think that I cannot even think contradictions, and to this extent Heidemann’s reading of IMP would be on interpretatively stronger ground than mine. In his essay against Eberhard, Kant writes that “whatever conflicts with [the] principle [of non-contradiction] is obviously nothing (not even a thought)” (ÜE, AA 8:195; emphasis added). This would seem to straightforwardly contradict my account given above. In the Critique, Kant is more cautious, when he says that “no cognition [Erkenntnis] can be opposed to [the principle of contradiction] without annihilating itself” (A151/B191). Notice that Kant is talking about Erkenntnis here, not about thought as such (recall the distinction between denken and erkennen that Kant makes at B146). But what Kant seems to be saying, at least in the Eberhard passage, is that it is impossible to flout the principle of non-contradiction as a principle of general logic if it is the case that I actually think, thus disallowing contradictory thoughts.[6] This would make it prima facie difficult, on Kant’s account, to assess why, if someone S were to flout the principle of non-contradiction in making a contradictory statement pp is contradictory. One could however make the case that such an assessment is in fact a second-order thought or statement which points out or expresses the contradictory nature of p; but q would of course itself thereby not be contradictory, and hence ‘illogical’ (cf. Tolley 2006:391, 403n43). Nevertheless, the more interesting case is where S herself must be seen to be able to apprehend the link between herself uttering the first-order statement p and then correcting herself by stating, by means of q, that p amounts to a contradiction. S should be able to connect or take, by virtue of the principle of apperception, both q and p as statements or thoughts uttered by herself; and this would be difficult to maintain if p were, as Kant indeed says in the Eberhard passage, not “even a thought” since it is contradictory, for how could know that p was indeed a statement uttered by herself if not by accompanying with an ‘I think’ the representations that make up p as her own and thus unmistakably connecting them to her selfsame representations that make up q? It seems, then, that in the Eberhard passage Kant flouts his own principle of apperception as the universal principle of discursive thought, as “the highest point to which one must affix all use of the understanding, even the whole of logic” (B134n.).

Whatever the case may be in regard to the question whether Kant allows contradictory thoughts, by no means does transcendental logic explain the principles of general or formal logic simpliciter, that is, answer the question of e.g. why we abide by the principle of non-contradiction or why the principle of the excluded middle says that either 𝜙 or ¬𝜙; transcendental logic merely stipulates that there are certain metaphysically or quasi-metaphysically necessary conditions for the possibility of what Kant calls general or formal logic (Quine’s “sheer logic”) (see B79; B170), which makes general logic in that sense conditioned on Kant’s transcendental logic, namely in the sense that general logic is not even possible without transcendental logic grounding it. Transcendental logic does not explain why the principle of non-contradiction is an absolute law of logic, though it can be demonstrated from what constitutes human discursive thought that the principle applies to any thought one thinks, without, again, thereby explaining why this is so. This can be illustrated by looking at how the law of non-contradiction is conveyed in the very principle of transcendental apperception, as the principle of any discursive understanding (which includes making statements of analytic or conceptual truth), itself:

The principle of transcendental apperception, the ‘I think’ proposition, states that I must be able to accompany all my representations, which implies—and I must pass over here many details that explain why this is so[7]—that representations that are not my representations strictly speaking are not accompanied by my ‘I think’, for any instantiation of the ‘I think’. In other words, representations can only be called my representations if I so accompany them, by thinking them, for any instantiation of the ‘I think’. All other representations, which are not accompanied by my ‘I think’, if the ‘I think’ is instantiated, are eo ipso not my representations sensu stricto. That means that a representation cannot be mine if it does not belong to the set of “all my representations”, which a representation is only when it is accompanied by my ‘I think’ jointly with all my other occurrent representations (representations that I take together as one in whatever complex thought I entertain). It is intrinsically contradictory to have a representation which is claimed to be mine, while not thinkingly accompanying that representation, jointly with all my other occurrent representations, thus claiming it to belong to me. This by no means reductively explains what the principle of non-contradiction is, nor, as I made it clear above, are contradictory statements ipso facto impossible thoughts, but at least it shows that thought itself fundamentally expresses the bindingness of the principle of non-contradiction, and in the sense that the transcendental logic of our discursive thought is a necessary condition of all thought, transcendental logic grounds even the principles of “sheer logic”, such as the principle of non-contradiction.

Recently, Clinton Tolley (2012) has argued, in contrast to the orthodox interpretation that with respect to the unrestricted scope of general logic transcendental logic is a special logic with a restrictive scope, that transcendental logic is not “domain-subordinative”, as traditionally most commentators suggest or argue, but rather “domain-coincident” with general or formal (traditional) logic. Tolley’s views in this regard tie in with my view, represented in outline above, that transcendental logic grounds even the principles of “sheer logic” (Kant’s general or “traditional” logic, as Tolley refers to it). In the conclusion to an article that must be regarded as effecting an important shift in the understanding of the relation between Kant’s general or “traditional” logic and his transcendental logic, Tolley writes:

[T]ranscendental logic has been shown to deal with principles that govern all kinds of thinking and judging, no matter what sort of object is being thought about. This is because transcendental logic specifies a condition without which thinking would have absolutely no content whatsoever, because there simply is no other kind of content that is possible for thinking. To think at all is to cognize, to consciously represent an object, through concepts; to think at all is to think about an object. Yet because the generic concept of an object of thought just is the subject-matter of transcendental logic, transcendental logic, no less than traditional logic, provides a conditio sine qua non for any instance of thinking and understanding. Both logics, therefore, are equally and unrestrictedly general in their scope, which implies that their domains must be viewed instead as perfectly coincident. The contrast between the logics is not to be understood in terms of the difference between kinds, or the difference between genus and species, but rather in terms of the difference between aspects of thinking or judgment that are at issue—namely, the difference between the form and the content of understanding. (2012:441; my underlining)


I have said very little that is explicitly critical of Hanna’s account. That’s because I wholeheartedly agree with almost all that he actually says (and he does say a lot that is going to throw a monkey wrench in the prevailing debates in metaphysics, logic, epistemology, Kantianism, and so forth). However—and perhaps this is beyond the remit of the book, which is not devoted to strict Kant interpretation—it was not always clear to me in what positive manner transcendental logic, or the logic of the synthetic a priori, can be said, in one way or the other, to ground or constrain analyticity, or properly mark out the distinction between analytic or conceptual truths and synthetic a priori as well as synthetic a posteriori truths, truths in virtue of nonconceptual content. What is clear to me is that Hanna has sufficiently shown, in glorious analytic detail, that all efforts heretofore to explain the notion of analyticity and the difference(s) between strictly- or sheerly-logical truths and truths that are not strictly- or sheerly-logical truths have come to nought, and that all so-called solutions are not explanations of analyticity or analyticity* and whatever is non-necessary and non-a-priori truth, but of “schmanalyticity”, i.e. no explanation at all. So much for analytic philosophy as the paradigm of rigorous philosophy!

But perhaps in his reply Hanna can divulge some more about what he thinks that Kant brings to the table precisely, whether he’d agree with my account above or has a different take on this—beyond the fact that we must assume the notion of the synthetic a priori so as to explain analyticity, and how this relates to a “sharp” distinction between the purely analytic and the synthetic. Sometimes, as I tried to convey earlier, Hanna’s reasoning struck me as emphasising the distinction between analytic or conceptual truths and truths about non-logical matters, facts in the world (non-conceptual truths?), too much. Clearly, they are non-reducible kinds of truth, but given the paramount importance of the synthetic a priori as the quintessential middle term that is sorely missing in analytic philosophy, one would assume that, as I believe, the distinction between the analytic and the synthetic, necessary though it is, can’t be absolute, but is relative, just because both rely on the synthetic a priori as their enabling ground (without thereby dissolving or conflating in any way the distinction between transcendental and general or formal logic; that hard border in fact first enables us to understand the synthetic a priori as the middle term between the analytic and the synthetic).

This last point also relates to Hanna’s emphasis on the absolute distinction between conceptual truths and truths in virtue of nonconceptual content, which concerns his essentialist Kantian nonconceptualism. As he says, there is “nothing less than [a] categorically sharp contrast” between a priori truth in virtue of conceptual content and a posteriori truth “as represented by autonomous essentially non-conceptual content”, that is, content whose “truth is never in virtue of conceptual content” (p. 153–4). However, in properly Kantian terms, it strikes me as a matter of course, and as the whole point of Kant’s Critical metaphysics, that any truth about strictly-non-conceptual matters, facts in the world, must at least in some sense be internally related, in virtue of the synthetic a priori, to whatever conceptual truth about those facts is stated in propositional form, for how else could we express true statements about and have true knowledge of these facts? In other words, strictly-conceptual truths and strictly-non-conceptual truths are grounded in a truth they share: the transcendental truth which grounds both alike. This is most emphatically expressed by the famous Leitfaden passage at A79/B104–5, where Kant writes:

The same function that gives unity to the different representations in a judgment also gives unity to the mere synthesis of different representations in an intuition, which, expressed generally, is called the pure concept of the understanding. The same understanding, therefore, and indeed by means of the very same actions through which it brings the logical form of a judgment into concepts by means of the analytical unity, also brings a transcendental content into its representations by means of the synthetic unity of the manifold in intuition in general, on account of which they are called pure concepts of the understanding that pertain to objects a priori. (A79/B104–5)

Kant does not say or suggest here that there could not be intuitions or sensible content that are or is not subsumed or at least subsumable under concepts (either a priori or empirical concepts). Conceptualists who take this passage to refute Kantian nonconceptualism are simply wrong (see Schulting 2017, ch. 5; see also Allais 2016 and Golob 2016). But he does say or at least strongly suggest that given any particular judgement p, the empirical, sensible or intuitional content of p is combined by the same function that combines the concepts in p. This function, which is the necessary original, hence a priori, synthetic unity of apperception, both unites concepts analytically and provides p with transcendental content synthetically, namely it constitutes its very aboutness, the “fact” that the judgement is about objects or events in the world. This means that the “categorically sharp contrast” between a priori truth in virtue of conceptual content and a posteriori truth in virtue of nonconceptual content is not stipulated by Kant in the Leitfaden, but rather the opposite, at least insofar as synthetic a priori truth is concerned. The very same function that unites concepts insofar as their basic analytical subordination relation is concerned also establishes synthetic a priori truth in terms of the application of the categories to any empirical content in a judgement p (that is, the manifold of representations in an intuition that is the x underlying the subject term of a categorical judgement), and hence, because synthetic a priori truth (transcendental truth) is the necessary condition of synthetic a posteriori truth (material or empirical truth), it also thereby establishes synthetic a posteriori truth, given sensory input of course.[8]

One might want to argue that here in the Leitfaden Kant says nothing about analytic truths, and technically speaking that’s right. But recall Kant’s remark in that earlier quoted footnote to the B-Deduction, in which Kant makes the sideways observation that original apperception is a condition even on the whole of logic. As I have said above, this could be interpreted in such a way that analytic statements, which are prima facie governed merely by the rules of what Kant calls general or formal logic (B79; B170), are also constrained by the necessary rules of transcendental logic, in that they are as much thoughts with semantic content as judgements about objects are, even though the analysis of their purely conceptual content does not require reference to the functions of thought as metaphysical categories, let alone to actual objects. To put this differently, analytic truths are not in any way reducible to synthetic a priori truths (which would result in an inverse kind of ‘schmanalyticity’), but clearly the distinction between the analytic and the synthetic cannot be a “categorically sharp” one, as Hanna believes, lest one dismiss this idea of transcendental logic being a condition on logic itself—but then, if he were to do so, Hanna would contradict his own attempts to reintroduce or reappraise the analytic-synthetic distinction, and a fortiori, the synthetic a priori, as a means of explaining the very possibility of analyticity or the principles of sheer logic, and its distinction from syntheticity.

Acknowledgements: I would like to thank Clinton Tolley for his feedback as well as my co-editors for their helpful comments on an earlier draft of this essay.


[1] An abridged English version of this article appeared in 2012 in the now defunct online journal Kant Studies Online .

[2] Even apart from explanatory problems with the meaning of ‘analyticity’, I have often wondered why someone would declare him- or herself specifically an analytic philosopher. To me, to do so is more a political or ideological gesture, so as to mark out one’s turf, or at any rate to make it clear that one is not a continental philosopher, whatever that means (with all the institutional benefits that brings). If it is not political, then certainly the moniker ‘analytic philosophy’ is pleonastic. As if philosophy could be anything other than analytical. Isn’t philosophy about analysis tout court? On the other hand, one may doubt if an ‘analytically oriented’ approach to Kant’s Critical philosophy is actually appropriate. Kant used the phrase in an early Anthropology lecture (in the Winter Semester of 1775–76); he is reported to have said the following: «Indeed, there exist sciences of the kind, and this is analytical philosophy [analytische Philosophie], in which one sheds light on obscure representations by uncovering them» (V-Anth/Fried, AA 25:480, Lectures on Anthropology, trans. and ed. A. Wood et al [Cambridge: Cambridge University Press], p. 56). In the later Anthropology Mrongovius (1784–85), Kant says: «In analytic philosophy [Analytischen Philosophie], I simply make obscure representations in the soul clear» (V-Anth/Mron, AA 25:1222; Lectures on Anthropology, p. 353). Given these definitions, what Kant says in an introductory passage in the Transcendental Analytic, in the run-up to the Metaphysical Deduction would seem to interdict an ‘analytically oriented approach’ to transcendental philosophy, which would subvert the latter if it were applied to it; Kant says namely that «I understand by an analytic of concepts not their analysis, or the usual procedure of philosophical investigations, that of analyzing the content of concepts that present themselves and bringing them to distinctness, but rather the much less frequently attempted analysis of the faculty of understanding itself, in order to research the possibility of a priori concepts by seeking them only in the understanding as their birthplace and analyzing its pure use in general; for this is the proper business of a transcendental philosophy; the rest is the logical treatment of concepts in philosophy in general» (A65–6/B90–1; my underlining). 

[3] By way of anecdotal evidence, in Brian Leiter’s recent survey “The Most Important Western Philosophers of All Time”, so as recent as last April (!), voters rated Quine as the 4th most important 20th century philosopher after Frege, Wittgenstein and Russell, and the 22nd of all time, above Epicurus and Rousseau! It is striking that Husserl and Heidegger didn’t get enough votes to end up in the top-30. The other 20th-century philosophers in the top-30 are, predictably, Kripke, Lewis, Rawls and Carnap, the latter by far the more interesting and important of the bunch.

[4] I concur with the central thrust of Clinton Tolley’s (2012) argument that Kant’s general or formal logic and transcendental logic are “domain-coincident”. I suggested a similar approach in Schulting (2012). On Tolley’s views in this regard, see further below.

[5] Similarly, I could think of myself as not now existing as a logical possibility, where “not now existing” is the semantic content of my actual self-referring thought, though not actually to exist while thinking is obviously a metaphysical impossibility.

[6] I thank Clinton Tolley for directing me to these passages in Kant. Tolley discusses them in Tolley (2006) in the context of an illuminating account of why Kant’s general logic should be seen as constitutive, rather than normative. If Kant’s general logic is seen as constitutive, then the laws of logic, such as the principle of non-contradiction, are not laws that we ought to obey, but ones which through their sui generis “bindingness” (Tolley 2006:397) we in fact do obey in any act of the understanding, and so in any thought. On this account, it is reasonable, against the backdrop of the textual evidence, for Tolley to conclude that, in Kant’s view, we do not have the freedom to think what is illogical (2006:389). Though Tolley stresses the importance of self-consciousness in acts of the understanding (cf. 2006:405n.54), and his elaborate comparative discussion of free will and spontaneity in the theoretical and practical contexts is extremely interesting and highly relevant, I believe that too much stress on the ‘constitutive’ rather than ‘normative’ nature of theoretical thought risks losing sight of the essentially spontaneous, self-legislative character of thought, quintessentially expressed by the transcendental principle of apperception as a principle of the self-ascription of all of one’s own thoughts. See also my extensive discussion of spontaneity in judgement in Schulting (2017), ch. 3. There, I discuss the views of Henry Allison, Karl Ameriks and Robert Pippin in this regard, as well as those of Wilfrid Sellars, whom Tolley refers to (2006:405n.54).

[7] For an account, see Schulting (2012) and Schulting (2017), ch. 4.

[8] For more on the Leitfaden, see Schulting (2012), ch. 5 and Schulting (2017), esp. chs 2, 3 and 5.


Allais, L. (2016), ‘Conceptualism and Nonconceptualism in Kant: A Survey of the Recent Debate’, in D. Schulting (ed.), Kantian Nonconceptualism (London and New York: Palgrave Macmillan), pp. 1–25.

Golob, S. (2016),Why the Transcendental Deduction is Compatible with Nonconceptualism‘, in D. Schulting (ed.), Kantian Nonconceptualism (London and New York: Palgrave Macmillan), pp. 27–52.

Heidemann, D. (2012), ‘The “I think” must be able to accompany all my representations. Kant’s Theory of Apperception and the Unconscious’, in P. Giordanetti et al (eds), Kant’s Philosophy of the Unconscious (Berlin: de Gruyter), pp. 37–59.

Schulting, D. (2010), ‘Kant, non-conceptuele inhoud en synthese, Tijdschrift voor Filosofie 72,4: 679–715.

——— (2012), Kant’s Deduction and Apperception. Explaining the Categories (Basingstoke and New York: Palgrave Macmillan).

——— (2015), ‘Probleme des „kantianischen“ Nonkonzeptualismus im Hinblick auf die B-Deduktion’, Kant-Studien 106,4: 561–80.

——— (2017), Kant’s Radical Subjectivism: Perspectives on the Transcendental Deduction (London and New York: Palgrave Macmillan).

Tolley, C. (2006), ‘Kant and the Nature of Logical Laws’, Philosophical Topics 34,1/2: 371–407.

——— (2012), ‘The Generality of Kant’s Transcendental Logic’, Journal of the History of Philosophy 50,3: 417–46.

© Dennis Schulting, 2017.

Dennis Schulting is a former Assistant Professor of Philosophy at the University of Amsterdam. His work concentrates on Kant and German idealism. His most recent journal publications were published in KANT YEARBOOK, KANT-STUDIEN, the PHILOSOPHICAL REVIEW and KANTIAN REVIEW. He is the editor of Kantian Nonconceptualism (Palgrave Macmillan 2016). His second monograph, entitled Kant’s Radical Subjectivism: Perspectives on the Transcendental Deduction, will be published in June 2017.