Kant's Radicalization of Cartesian Foundationalism: Thought Experiments, Transcendental Arguments, and Level Circularity in the Paralogisms

Abstract

Kant's critique of rational psychology is a thought experiment that targets no individual or school, but rather the natural tendency of human reason to “hypostatize” the highest intellectual condition of all cognition (the pure ‘I think’) as though it were itself a cognition of the ‘I.’ To do so is to violate the very anti-circularity stricture also at work in Kant's better-known transcendental critiques of Locke and Hume. Along with a new type of circularity (level circularity), this article proposes a conception of transcendental arguments different from that of most contemporary debates regarding empiricism, naturalism, and Cartesian foundationalism.

Résumé

La critique kantienne de la psychologie rationnelle est une expérience de pensée visant ni un individu ni une école, mais une tendance de la raison humaine à « hypostasier » la condition intellectuelle suprême d'une connaissance quelconque (le « Je pense ») en connaissance du « moi ». Cette tendance implique une circularité qui est également la cible des critiques transcendantales bien plus familières qui visent Locke et Hume. De même qu'un nouveau type de cercle (dit « de niveau »), cet article propose une conception des arguments transcendantaux différente de celle présupposée dans la plupart des débats contemporains sur l'empirisme, le naturalisme et le fondationnalisme cartésien.

1. Introduction: Kantian Thought Experiments¹

The Paralogisms chapter of Kant's first Critique is devoted chiefly to exposing the transcendental illusion at work in the fallacious reasoning behind three alleged synthetic a priori principles asserting the substantiality, simplicity, and numerical identity of the soul.² The principles form the conclusions of three syllogisms or dialectical inferences to the soul's absolute unity.³ While Descartes is expressly identified as the target of a somewhat anomalous fourth paralogism, the target (or targets) of the first three is more controversial.⁴ So is the exact nature — whether logical, non-logical, or both — of the fallacies involved, and the precise relations among the key terms ‘transcendental illusion,’ ‘hypostatization,’ and ‘subreption.’ What has gone largely unremarked is the radicalizing critique of Cartesian foundationalism that forms a secondary theme of the chapter.⁵

In the sense intended here, the word ‘radicalizing’ refers to Kant's attempt to ‘get beneath’ that which Descartes and his followers considered the ultimate or fundamental sources (‘roots’) of epistemic justification. In this sense, Descartes himself had already radicalized Scholastic-Aristotelian foundationalism by ‘getting beneath’ its founding dimension (ordinary sense perception) and grounding all cognition of outer things upon inner cognition of the empirical ego cogitans (see Section 2). Kant's critique, however, is more radical. From his new, transcendental standpoint, any attempt to locate the ultimate ground of all object-level cognition in a (or some) empirical (or even pure) cognition belonging to that very level is circular. In positive terms: the foundational dimension of any non-circular justification of all first-order cognition of objects as such must be sought ‘lower down’ (or, varying the metaphor to fit Kant's project, ‘higher up’), in meta-level reflection upon the knowing subject and the conditions of the possibility of any object-level cognition (inner as well as outer, either empirical or a priori) whatsoever. So put, the target of Kant's critique would appear to be identical to that of the fourth paralogism: the Descartes of the Second Meditation. But it applies equally to almost the entire early modern tradition on both sides of the channel, which had absorbed Descartes's foundationalism (with or without his rationalism) almost ‘atmospherically,’ so to speak, much as it would have been imbibed by the pre-critical Kant himself.

Breaking with Cartesian foundationalism, the mature Kant conceived and, in the Transcendental Analytic, executed a novel transcendental foundationalist project designed to ‘get beneath’ the empirical cogito, ergo sum that had been the first principle of Descartes and at least a first principle for all those who can justly be considered his heirs. In the Paralogisms chapter of the Transcendental Dialectic, moreover, he conducted a bold thought experiment designed to show that attempts to ground all first-order cognition in some first-order cognition commit a level mistake that renders them viciously circular. The thought experiment posits two hypothetical forms of foundationalism that neither existed in Kant's day nor had ever existed in the past. The first is presented in the guise of a mixed (both pure and empirical) rational psychology that circularly misconstrues the Ich denke of transcendental apperception, which is the highest intellectual condition of all cognition of objects as appearances, as though it were itself a cognition of an (inner) object or appearance — specifically, an inner perception of the empirical self. This violates the transcendental anti-circularity stricture formulated at A402 as: “I cannot cognize as itself an object that which I must presuppose in order to cognize an object at all.”⁶ Now, as Kant surely realized, neither the historical Descartes, nor anyone else, had had the least conception of an a priori transcendental condition of the possibility of all empirical cognition überhaupt prior to his own Copernican thought experiment.⁷ This ‘refutation’ would therefore be entirely without force if it targeted some particular historical figure or figures rather than the hypothetical discipline imagined in the first part of Kant's thought experiment. Viewed as a thought experiment, however, the anti-circularity stricture turns out to be the operative principle in a transcendental argument that still represents a serious challenge to both empiricist (naturalist) and rationalist efforts to ground all first-order cognition of objects upon some (empirical or a priori) first-order cognitions of objects. It hardly matters whether the candidate explanatory or justificatory ultimates are sense-data, impressions, sensations, ideas (even innate ideas), or inner and outer perception; from the transcendental perspective, any such explanation or justification is circular. Whether the only way to ground first-order cognition non-circularly is through second-order reflection on the transcendental subject and its a priori intuitions and concepts is, of course, quite another matter. Here Kant's procedure in the Critique betrays a Cartesian bias that held sway in modern philosophy at least until Husserl, and that was only seriously challenged by Heidegger (see note 38 below).⁸

The second and main form of foundationalism imagined by Kant in a further (or as part of the same, extended) thought experiment is explicitly said to be a merely hypothetical, or as Kant himself puts it, a “putative (angebliche)” (A342/B400) completely pure science of the soul based exclusively on the a priori “concept — or rather, if one prefers — the judgment I think” (A341/B399).⁹ As Kant well realized, no science of the soul that understood itself in quite this way existed or had ever existed.¹⁰ The point of this further thought experiment is rather that universal human reason is inclined (though not necessitated) to develop such a completely pure doctrine of the soul, the ratiocinative faculty of the mind being naturally constrained (as formal logic shows) to proceed ever higher in search of the unconditioned to any given series of conditions (cf. A322–323/B379). Whereas the Ich denke is no more than the highest intellectual condition of all cognition of appearances, the human ratiocinative faculty, which “relates itself only to the use of the understanding,” aims at “a certain unity of which the understanding has no concept” (A326/B383). The main target of the Paralogisms is this natural propensity of human reason to misconstrue the Ich of the Ich denke as a purely intelligible or noumenal object that can be cognized in an absolutely unconditioned manner through mere categories.¹¹ For, as Kant points out, “[a]pperception is itself the ground of the possibility of the categories” (A401). Thus, the completely pure rational psychology of the second and main part of Kant's thought experiment circularly misconstrues the pure Ich denke, “which grounds the categories” (A421), as though it were itself an a priori cognition of an intelligible object (the pure Ich), applying to it the very categories (‘substance,’ ‘simplicity,’ etc.) that it makes possible in the first place. This grounding in the absolutely unconditioned is not just a transcendental misuse of those categories whose sole legitimate use the Analytic had shown to be empirical; it constitutes a further violation of the above-cited anti-circularity stricture. This time, however, the alleged object is purely intelligible, and not that empirical object of inner perception of which mixed rational psychology purports to have both empirical and a priori cognition.

Only this second part of Kant's thought experiment (which occurs first and remains his dominant concern throughout the chapter) involves a dialectical inference to “the absolute (unconditioned) unity of the thinking subject” (A334/B392), since only such a completely pure rational psychology has its origin in pure reason's requirement that the cognition of a series of conditioned conditions be grounded in that of the unconditioned. By contrast, the error of that mixed rational psychology imagined in the first part of the thought experiment stems from mistaking the ‘I’ of the pure Ich denke, which is the supreme condition of all cognition, for a cognition of the self of empirical apperception. Since there is nothing dialectical about this, it is hardly surprising that Kant's radicalizing critique of Cartesian foundationalism has been eclipsed by the dialectical arguments of the four paralogisms (especially the fourth) in the scholarly literature on Kant's relationship to Descartes. It is nevertheless quite as important as the fourth paralogism both historically and in a systematic perspective.¹²

Its obvious historical significance aside, the first part of Kant's thought experiment epitomizes what he himself would most readily have recognized as a ‘transcendental argument.’ Today, that term is used almost exclusively to describe a type of argument whose prototype is the famous Refutation of Idealism, but which has little to do with the technical sense of ‘transcendental’ in Kant: “occupied not so much with objects but rather with our mode of cognition of objects insofar as this is to be possible a priori” (B25, cf. A11). In Section 6, it will be argued that the radicalizing critique of Cartesian foundationalism is a transcendental argument as Kant defines the term and that this general type of argument has an important bearing on contemporary philosophical debates regarding empiricism (or naturalism), rationalism, and foundationalism — debates in which transcendental arguments modelled on the Refutation have long occupied a pre-eminent place.

The procedure will be as follows. After an initial survey of the ramified family of critiques of Cartesian foundationalism, including those nowadays called ‘transcendental arguments’ (Section 2), the epistemological anti-circularity stricture of the Paralogisms will be contrasted with its better-known ontological counterpart in rational theology (Section 3). Following consideration of three different level mistakes, two of which amount to level circularity (Section 4), the textual basis of the re-interpretation of the Paralogisms as a two-part thought experiment will be presented (Section 5). A rapid glance at Kant's use of similar arguments against empiricist accounts of space and causality will confirm that Kant's epistemological anti-circularity stricture, while lacking in strictly refutative force outside his own two-part thought experiment, still has a certain weight when it comes to the systematic question of whether to naturalize or, alternatively, transcendentalize, foundationalist epistemologies of a Cartesian stamp (Section 6). A brief conclusion follows (Section 7).

2. Foundationalist and Other Critiques of Cartesian Foundationalism

Criticism of a foundationalist epistemology like Descartes's generally takes one of two forms: either it advances (1) an alternative, more radical theory of the same general kind — an alternative foundationalism — or it proposes (2) an anti-foundationalist alternative to foundationalism.¹³ Coherentist theories of justification directly or indirectly inspired by ‘naturalized’ epistemology (cf. Quine, 1953, 1969) are anti-foundationalist critiques of the latter sort; so are those transcendental arguments modelled on the Refutation of Idealism and/or the fourth paralogism that today pass for transcendental arguments tout court (cf. Stroud, 1968, 1979). The Paralogisms, by contrast, form part of an alternative foundationalist strategy aimed at radicalizing traditional Cartesian foundationalism by exposing its circularity.¹⁴

The single most striking example of (1) in the history of philosophy is Descartes's own reversal of what had long been the established order of knowing. With his mens notior corpore doctrine, succinctly formulated in the Principles as “our knowledge of our thought is prior to, and more certain than, our knowledge of any corporeal thing” (AT VIII–1 7: CSM I 195), Descartes gave self-knowledge, as expressed in empirical, contingent propositions (the various values of the variable cogito, ‘I am perceiving,’ ‘I am imagining,’ etc.), an absolute priority over cognition of the external world present to the five bodily senses. Such cognition of material things as can be reliably obtained must be derived by inference from non-inferential cognition of the self, using logical and/or metaphysical “axioms,” “common notions,” “eternal truths,” or “principles of the natural light.”¹⁵ In the Aristotelian tradition, by contrast, knowledge of the self had always been regarded as requiring a special act of consciously directing one's attention away from sensible things (the first and most reliably known objects), a bending back (reflectere) of one's cognitive gaze upon one's actually occurring cognitive acts. As thus dependent upon cognition of ordinary sensible things, empirical cognition of the self is ‘later’ in the order of knowing. Yet though ‘earlier,’ cognition via the outer senses had never been considered a source of principles having the same degree of certainty as the universal, necessary, and hence absolutely first non-existential truths of logic and/or metaphysics. Accordingly, Descartes's (and later Leibniz's) elevation of the contingent existential propositions cogito and sum to the status of absolutely certain first principles was revolutionary: in redefining ‘absolute certainty’ operationally as ‘the ability to withstand the illusion, dreaming, and deceiving God arguments of the First Meditation’ (rather than in terms of universality or necessity), Descartes re-drew the extensional boundaries of ‘principle’ so as to make room for a pair of contingent propositions alongside the necessary principles of the natural light.¹⁶

The Cartesian paradigm confirms that the shared trait of all radicalizing critiques of pre-existing foundationalisms is the effort to ‘go back behind’ those alleged first truths whose greater reliability serves to render all other cognitions more certain. It is true that at least one 20^th century form of foundationalism took the basic truths to be empirical propositions about bodies (the so-called Protokoll-Sätze of 20^th century physicalists); but the influence of Descartes has been such that philosophers have generally looked to empirical first truths about the mental, including the sense-data of the phenomenalists, for foundational or first truths endowed with existential import. Seen in this wider context, the most novel thing about Kant's foundationalism is his insistence that the supreme or ultimate principles must be meta-level cognitions of the conditions of the possibility of both empirical and pure object-level cognitions as such, and not themselves again cognitions of any either empirically (phenomenally) or absolutely (noumenally) real object, not even of the transcendental subject as an object. Behind this lies the conviction that any attempt to found all first-order cognition as such upon that which is itself just an item (or kind) of first-order cognition is viciously circular. The aim of Kant's transcendental foundationalism is thus not just to reach ‘rock-bottom’; if circularity is to be avoided, first or founding principles must be second-order cognitions about the transcendental subject and the possibility of all first-order a priori cognition of objects (whether metaphysical, mathematical, or natural scientific) überhaupt. Thus, Kant not only locates the founding dimension of human knowledge ‘further back’ (or ‘higher up’); he does so in view of a kind of circularity that no one before him had ever even glimpsed. This emerges most clearly in the Paralogisms.¹⁷

3. Ontological and Epistemological Anti-Circularity Strictures

Kant's neglected transcendental critique of Cartesian foundationalism has nothing to do with the familiar Cartesian circle (or circles) to be described presently; it is also a different type of transcendental argument from that exemplified by the Refutation of Idealism. Both points can be clarified by contrasting Kant's transcendental anti-circularity stricture with its better-known ontological counterpart (see note 8).

In a qualified defence of the point of view of “cosmological arguers,” Van Cleve formulates what he calls Kant's “counterprinciple” (Van Cleve, 1999, p. 206) to a principle defended by Hume in the Dialogues Concerning Natural Religion.¹⁸ Hume's principle states (in Van Cleve's terms): “the existence of a totality is always adequately explained when the existence of each member is explained” (Van Cleve, 1999, p. 205). The juxtaposition of Kant's counter-principle to a corollary of this yields Kant's ontological “anticircularity stricture” (Van Cleve, 1999, p. 206). The corollary affirms that “if there is an infinite totality of objects or events each of which is explained by the causal efficacy of some other member(s) of the totality, then the existence of the whole totality is thereby adequately explained — no recourse to anything outside the totality is needed” (Van Cleve, 1999, p. 205). The operative terms here are obviously ‘infinite’ and ‘outside,’ for what Hume rightly affirms of finite classes — that there is no need to go outside them in order to furnish a complete explanation of everything contained in them — no longer holds when the explanans is the existence of the infinite class of all things of a certain kind (say, all contingent things in general — or überhaupt, to use Kant's term). Van Cleve formulates Kant's anti-circularity stricture this way: “The existence of Fs in general (or the fact that Fness is instantiated) can be explained only by appeal to the existence of something that is not an F” (Van Cleve, 1999, p. 206). Elsewhere he substitutes the expression “at all” (e.g., Why are there any Fs at all?) for Kant's überhaupt. The reason that such explanations are circular is simply that the “explanans invokes the existence of Fs, but Fs are the very beings for whose existence an explanation is sought” (Van Cleve, 1999, p. 205). In the simplest terms, the explanans itself forms part of the explanandum.¹⁹

Van Cleve's anti-circularity stricture is straightforwardly ontological; it concerns the possible existence of an infinite class of all contingent entities without something that exists necessarily. What motivated Kant's transition from (his own pre-critical) Cartesian to a new, transcendental foundationalism was an anti-circularity stricture that ruled out attempts to justify — or even just to explain — the cognition of all objects of a certain kind (all x's überhaupt) by recourse to a first or absolutely certain cognition of an object of that very kind (an x). Van Cleve's anti-circularity stricture can be generalized to include cognition in the following way: if y is to serve as the justificatory or explanatory ground of some class of x's, {x ₁, x ₂, x ₃ … x _n} (where n is some finite number), then, while y cannot (without circularity) be a member of that very class, it may still be some other x outside the class, for instance, a ‘first’ x that exists prior to all others, or an x that is more certain than the rest, depending on whether what is in question is the efficient cause of the existence of things or the explanatory and/or justificatory ground of cognitions. Thus, whether Descartes's manner of executing the foundationalist project of the Meditations is premise-circular depends on whether he uses his truth rule (‘whatever I perceive clearly and distinctly is true’) as a premise in the proof of God's existence and veracity, while using God's existence and veracity to establish the certainty of his truth rule; and whether his procedure is rule-circular depends on whether he first deduces that God exists and then uses that conclusion to guarantee the reliability of deduction.²⁰ Finally, whether it is faculty-circular depends on whether he first doubts his reason and then attempts to reason his way out of doubt. However, there is nothing either premise- or rule- or faculty-circular about Descartes's project of placing the cognitions of God and the world on the secure foundation of the cognition of the self. Only where the justificandum is the infinite set of all cognitions of any object whatsoever (überhaupt) does it become viciously circular for the justificans to be itself a cognition (or set of cognitions) of an object, since it then forms part of the justificandum. This type of level circularity has to do with the two level mistakes touched on earlier (see note 5) and to be considered further in the next section.

4. Transcendental Distinction, Level Mistakes, and Level Circularity

The level mistakes noted earlier correspond to two ways of “hypostatizing” (A395) the ‘I’ of transcendental apperception. The (a) Ich of the Ich denke may be hypostatized either as (b) an appearance (an object given in empirical intuition), as in the first part of Kant's thought experiment; or it may be hypostatized as (c) a thing in itself (an intelligible object), as in the second part. Both level mistakes violate the anti-circularity stricture of A402 by taking what is an a priori condition of all cognition überhaupt for a cognition of an object; and both differ from the much more prominent level mistake relentlessly exposed throughout the Transcendental Aesthetic and Analytic, namely hypostatizing (b) appearances, which are at bottom “representations” for Kant, as though they were (c) things in themselves. While Kant understands his “transcendental distinction” (A45/B62) as a simple binary distinction between (b) and (c), certain of the uses to which he puts it (notably in the Paralogisms) require a distinction among three elements: hypostatizing (b) appearances by raising them to the level of (c) things in themselves, which is non-circular; hypostatizing (a) the Ich of the Ich denke by elevating it to the level of (b) an appearance (the empirical ego cogito of Descartes), which is circular; and hypostatizing (a) the Ich by raising it to the level of (c) a thing in itself, which is again circular. The first level mistake is that which Kant regarded as historically actual (indeed pervasive in the transcendental realist tradition), while the other two, on whose circularity he focuses in the Paralogisms, are merely hypothetical and belong to the two parts of his thought experiment.²¹

Within the Paralogisms, the level mistake on which Kant insists throughout the Transcendental Aesthetic and Analytic figures most prominently in the “Observation” (A381–396) devoted to the “three dialectical questions” (A384) regarding the mind-body relationship that form the subject-matter of the second and third parts of rational psychology (see note 2 above). Given its subject-matter, the “Observation” provides Kant with an occasion to reiterate the teaching of Analytic regarding bodies, even though bodies lie outside the scope of the first part of rational psychology (cf. note 2 again), to which the bulk of the Paralogisms are devoted. According to the Analytic, bodies are only outer appearances and hence mere representations; yet in the debate regarding the mind-body relation, Kant notes, “one hypostatizes what exists merely in thoughts, and — assuming it to be a real object outside the thinking subject — takes the same quality, namely extension, which is nothing but appearance, for a property of external things subsisting even apart from sensibility” (A384). Later, he returns to this tendency to “hypostatize outer appearances, no longer relating them to our thinking subject as representations, but rather relating them to it in the same quality as they are in us as things external to us and subsisting by themselves” (A386). The words “relating them to it in the same quality” pick out what, for Kant, is the root error of all “transcendental realism” (A369): hypostatizing things as appearances that are “outside us” in the empirical sense (cf. A373 on the empirical and transcendental senses of ‘outside’) as though they possessed the very same spatio-temporal and categorial features or qualities that they have as appearances (Kant mentions extension and motion) even when considered as things existing outside “our thinking subject” (A386) in the transcendental sense of ‘outside.’ According to the “Observation,” this error is the source “of all the difficulties that one believes one finds in these questions” (A384) regarding the mind-body relation during and after life; but while it may bedevil all three main theories (interactionism, occasionalism, and pre-established harmony), the transcendental realist's “hypostatizing” of (b) outer appearances as though they were (c) things in themselves does not involve circularity.

Circularity arises from hypostatizing (a) rather than (b) in either of the two ways considered in Kant's thought experiment. The error that is Kant's main focus in the Paralogisms is that of circularly elevating (a) the supreme intellectual condition of all cognition through the categories, the ‘I’ of the ‘I think,’ to the level of (c) an intelligible object of purely intellectual cognitions in which those very categories occupy the predicate position. This, in Kant's words, is to mistake the (a) “merely subjective condition [‘on which every experience depends and which precedes it’]” (A354) for (c) “a concept of a thinking being in general” (A354) or “a thinking being in itself” (A360 e.a.).²² It is this unexampled circle that occupies Kant in the second part of his thought experiment. Where he speaks of “making thoughts into things and hypostatizing them” for the third and last time in the Paralogisms, his focus has shifted from that of the “Observation” (regarding the mind-body relation) to this other level mistake whose blatant circularity vitiates the whole first part (concerned with the nature of a thinking being) of that “imagined science” (A395) of the soul which is completely pure. Thus, “everyone either presumes to know something about [purely intellectual, noumenal] objects about which no human being has any concept” (A395), as when outer (b) appearances are hypostatized as (c) things in themselves in those parts of this science dealing with the mind-body relation; “or else makes his own representations into [intellectual] objects” (A395), as when (a) the pure Ich denke is hypostatized as (c) a thing in itself in the part of rational psychology devoted to the nature of the soul. Either way, Kant notes, one only “goes round and round in an eternal circle of ambiguities and contradiction” (A395). But while, strictly speaking, no circle (but only a level mistake) is involved in baselessly presuming to know the human body as it is in itself, Kant's anti-circularity stricture is violated when (a) the ‘I’ of the Ich denke is elevated to the level of (c) an object of a purely intellectual cognition through the very categories that it makes possible.²³

Less noticed than either the first or second is the different level mistake ascribed to the mixed rational psychology of Descartes and his followers in the first part of Kant's thought experiment. This third level mistake involves “hypostatizing” that which is only (a) the subjective and necessary intellectual condition of all cognition of appearances whatever (the Ich denke) as though it were itself (b) a cognition of the ‘I’ as an appearance. Like the second, then, the third level mistake concerns (a) the Ich denke as the “sheer spontaneity of combining the manifold of a merely possible intuition” (B428) which is the highest intellectual condition of all appearances whatsoever; this time, however, the ‘I’ of the ‘I think’ is hypostatized as though it were (b) “the subject of consciousness as an appearance” (B428 e.a.), that is, an empirical object that is not “outside us” at all, not even in the empirical sense of A373, but ‘in us’ as the object of inner perception or the empirical self. For that which is “obviously not an experience, but rather the form of apperception on which every experience depends” (A354), namely (a) “the formal proposition of apperception, I think,” (A354), cannot (without circularity) be said to be (b) an experience of an object (the Cartesian ego cogitans), that is, an instance of the very experience that it makes possible.

Now according to the doctrine of transcendental idealism, to represent myself via the ‘I think’ of pure apperception is to “represent myself to myself neither as I am [in myself] nor as I appear to myself” (B429 e.a.). Despite Kant's frequent insistence that the Paralogisms are wholly devoted to a completely pure rational psychology, these words make it clear that he is also targeting that level mistake which he supposes to be the original sin of a mixed rational psychology of the type inaugurated by Cartesian foundationalism. But since neither the historical Descartes nor his rationalist and empiricist successors can be said to have hypostatized, as though it were an inner appearance, a pure or transcendental ‘I think’ of which they had not the slightest conception, Kant's critique only serves to highlight the fact that all historical forms of Cartesian foundationalism (all mixed and a fortiori all merely empirical psychologies) operate only with an empirical concept of the ‘I’ as an object of inner experience, attempting (non-circularly) to ground all other cognition on that which is itself an ostensibly more secure and reliable cognition — with the aid, in the case of mixed rational psychology, of certain a priori axioms of the natural light. The conclusion of this section is therefore that, while Kant's level-circularity argument is strictly refutative when directed against the mixed varieties of rational psychology envisaged in his thought experiment, the Paralogisms do not appear to lay the groundwork for any telling criticism of the actual Cartesian project of founding all other first-order cognition on a first-order cognition. That the circularity stricture has nevertheless a certain force against both historical and contemporary forms of empiricism and naturalism will be argued in the Section 6. First, however, its actual use in the text of the Paralogisms must be examined.

5. The Anti-Circularity Stricture in the Thought Experiments of the Paralogisms

Kant gives his anti-circularity stricture its most general — and eye-catching — formulation at the end of the “Observation” (appended to the four paralogisms of the first edition). This will be referred to as the A passage (A402). It is there immediately applied to a mixed doctrine of the soul of the sort first espoused by Descartes. A similar application of the anti-circularity stricture is found in (what will be called) the B passage (B421–422) appended to the condensed restatement of the four paralogisms in the second edition. A further, but different, use of the same stricture is found much earlier, at the very outset of the Paralogisms, in a passage that was taken over unaltered in the second edition (A346/B404). In the A/B passage (as it will be referred to), the target is a hypothetical completely pure science of the soul that hypostatizes the ‘I’ of the ‘I think’ as an intelligible object.²⁴

The above-mentioned general formulation appears as the very first sentence of the A passage. In the following sentence, however, Kant narrows the focus to the last of the three level mistakes (or the second type of level circularity) considered in the previous section: hypostatizing (a) the pure Ich denke as (b) an empirical cognition of an object. The passage reads in full (numbering the sentences for ease of reference):

Now it is indeed very evident (einleuchtend) that [1] I cannot cognize as itself an object that which I must presuppose in order to cognize an object at all (überhaupt); and that [2] the determining Self (thinking) is different from the determinable Self (the thinking subject) as cognition is different from its object. Nevertheless, [3] nothing is more natural and seductive than the illusion of taking the unity in the synthesis of thoughts for a perceived [e.a.] unity in the subject of these thoughts. [4] One could call it the subreption of hypostatized consciousness (apperceptionis substantiatae) (A402).²⁵

In (2), the perfectly general anti-circularity stricture of (1) is applied to an imagined mixed rational psychology based on the cognition of the empirical ego cogitans or “thinking subject”; equally as “evident” as (1), Kant maintains, is the fact that (2) “the determining Self,” which he here refers to as “thinking (das Denken)” instead of ‘the pure Ich denke,’ “is different from the determinable Self,” or ‘the thinking subject’ qua empirical object of inner perception, “as cognition is different from its object” (A402 e.a.). Now the latter would indeed be evident if it meant — as it seems it must, coming hard on the heels of the general circularity stricture — that the determining Self or ‘I’ of pure apperception differs from the determinable Self of empirical self-consciousness as (a) the highest intellectual condition of the cognition of all appearances überhaupt differs from (b) the inner cognition of the self as an appearance or object of empirical introspection. To hypostatize the former as though it were the latter is precisely the third level mistake considered in the previous section. But to say, as Kant does here, that they differ “as cognition … from its object” is, frankly, not evident at all. Instead of “cognition” and “its object,” one would have expected a contrast between a condition of all empirical cognition überhaupt and an item of empirical cognition. After all, Kant draws just such a contrast, using the same terminology, in the B edition, when he writes:

All modi of self-consciousness in thinking as such (im Denken an sich) … provide thought with no object at all, and hence do not present my Self as an object to be cognized. It is not the consciousness of the determining Self or Ich denke, but only that of the determinable Self, i.e., of my inner intuition … that is the object.” (B407, e.a. and ‘self’ capitalized)

In light of its reoccurrence in this passage, the substitution of das Denken for das Ich denke at A402 is less troubling; and the words “the unity of consciousness in which alles Denken consists” just a few lines earlier (B406) makes it even less so. By contrast, the ill-chosen opposition of “cognition” and “its object” is made more troubling by the fact that, in sentence (3), Kant immediately goes on to juxtapose that which is only a condition of all cognition with the empirical cognition of an object as inner appearance: “Nevertheless nothing is more natural and seductive than the illusion of taking the unity in the synthesis of thought,” that is, the merely intellectual unity represented in the determining Self or ‘I’ of pure apperception, “for a perceived unity in the subject of these thoughts” (A402 e.a.). What is in question in both (2) and (3), then, is precisely the third of the level mistakes distinguished earlier, circularly hypostatizing (a) the highest intellectual condition of all cognition of appearances in general as (b) a cognition of an empirical object or appearance. Taking (a) the ‘I’ of pure apperception for (c) the cognition of an intelligible object existing in itself is, of course, just as gross a level mistake (the second in the earlier enumeration), and just as evident a violation of the general anti-circularity stricture stated in (1). But that violation is not Kant's focus here, even if, in (4), he turns the reader's attention to a “subreption of hypostatized consciousness (apperceptionis substantiatae)” (A402) that presumably covers both kinds of level mistake or level circularity.

To turn now from the A to the B passage, the first thing to note is that, although its opening sentence refers only to a “misunderstanding,” Kant's attention turns to circularity already in the second sentence, coming to rest on that fallacy in the penultimate and final sentences, even though the word ‘circle’ is absent. The final sentence also provides an example of a different error analogous to that with which most of the passage is concerned: construing (d) that which is only a pure sensible condition of all empirical intuition of objects qua appearances existing in determinate time relations as though it were (e) an object empirically intuited as existing in determinate time relations. While (e) is just a specific instance of (b) or an appearance, (d) is only the sensible analogue of (a) “the unity of consciousness” or highest intellectual condition of the categories. The passage reads (without the final sentence):

[1] From all this one sees that [mixed] rational psychology has its origin in a mere misunderstanding. [2] The unity of consciousness, which grounds the categories, is here taken for an intuition of the subject as an object, and the category of substance is applied to it. [3] But this unity [of transcendental apperception] is only the unity of thinking, through which no object is given; and thus the category of substance, which [i.e., whose legitimate application] always presupposes a given intuition, cannot be applied to it, and hence this [transcendental] subject cannot be cognized at all. [4] Thus the [transcendental] subject of the categories cannot, by thinking them, obtain a concept of itself as an object of the categories; for [5] in order to think them, it must take its pure self-consciousness, which is just what is to be explained, as its ground. (B421–422 e.a.)

The final sentence gives what Kant presumably considers an intuitively more obvious example of the same sort of circle (again, without using the word):

[6] In just such a manner (Ebenso), the subject, in which the representation of time originally has its ground, cannot thereby determine its own existence in time, and if the latter cannot be, then the former, as the determination of its self (as a thinking being in general) through categories, can also not take place. (B422)

That mixed rational psychology is intended in (1) is indicated by the words “an intuition of the subject as an object” in (2) and by “given intuition” in (3); it would be very surprising if an intellectual rather than an empirical intuition were meant here. Were it not for the easily overlooked allusion in (2) to the circularity of applying the category of substance to the purely intellectual “unity of consciousness, which grounds the categories” (B421 e.a.) themselves and all cognition by their means, it might seem that the only point being made here were the familiar one highlighted in (3) and (4): since no such object as the pure ‘I’ of the ‘I think’ is actually given in empirical intuition (but only hypostatized as so given), the category of substance cannot be employed to obtain the sort of synthetic a priori cognition of the empirical self that has always been supposed to be attainable in the pure part of mixed rational psychology. After all, the words “which always presupposes a given intuition” clearly evoke the restriction placed on the use of the categories in the Transcendental Analytic. But, in fact, both the circularity alluded to in (2) and the limitation of the categories to given objects of intuition in (3) are intended to support the conclusion drawn in sentence (4): the (a) pure ‘I’ cannot itself be the object of a synthetic a priori cognition by means of the categories. For so to understand it would be a misemployment of the categories, to be sure, but it would also be an example of the “subreption of hypostatized consciousness (apperceptionis substantiatae)” mentioned at the end of the A passage. It is the clause numbered (5) that shifts attention from the limitation imposed by the Transcendental Analytic to the circle inherent in the procedure of the rational psychologist who applies the categories to the ‘I’ of the ‘I think’ of transcendental apperception: “for [5] in order to think them [the categories], it must take its pure self-consciousness, which is just what is to be explained, as its ground.”

Using “just what is to be explained” (the explanandum) as (part of) the explanation (the explanans) is roughly what is understood by a logically circular explanation. As Kant describes the imagined rational psychologist's procedure here, a category (‘substance’) is predicated of the unity of consciousness (the ‘I’ of the pure ‘I think’) that is itself the “ground” of all cognition of objects through categories. This is not just the misapplication of the categories proscribed by the Transcendental Analytic; according to (5), it is also circular. The tenor of the whole passage, with its frequent references to intuition, suggests that what Kant has in mind here is the hypostatization of (a) the pure ‘I’ as though it were (b) an empirical object given in inner intuition or perception. One might therefore speak of a ‘recuperation’ of the essential content of the deleted A passage here in the B edition, though with the unhelpful talk of a “misunderstanding” now doing duty for the precise formulation of the anti-circularity stricture in the A edition.

That the anti-circularity stricture of A402 is really at the heart of this passage is made even clearer by sentence (6), which draws a parallel with the circle that occurs when (d) the transcendental subject whose pure intuition of time is the “ground” or condition of all temporal determination of objects überhaupt is said to be itself (e) a temporally determined object or inner appearance. Kant apparently considers this an even more evident instance of the sort of level circularity with which he is concerned in the A and B passages; otherwise, he would hardly have resorted to it to reinforce the point that (a) the “determining Self” of pure apperception cannot be said to be determined as an appearance with respect to the category of substance-accident. As all efforts to determine (a) “the nature of our thinking being” (A345/B403) through the categories are circular, so too, according to (6), is the attempt to determine (d) the transcendental subject whose pure intuition is the source of all temporal determination of objects überhaupt as itself (e) a temporal object or inner appearance.²⁶

So much for the B passage. The third, much earlier A/B passage does not refer to a “misunderstanding” but explicitly to “a constant circle.”²⁷ It is not, of course, that this circular reasoning is the only thing that renders otiose the natural tendency of reason to develop a completely pure scientific rational psychology based on the Ich denke alone. This earliest of the three passages is only the opening gambit in the complex etiology of error that occupies the entire chapter. Still, the circle that Kant highlights in this introductory passage is enough to render null and void all the reasonings of this “imagined science” (A395) based on “nothing but the simple and in content for itself wholly empty representation I.” Since a completely pure science is in question, the A/B passage contains no such references to being given in intuition or perception as are found in both the A and B passages.

The third passage begins by juxtaposing (a) the “wholly empty representation I,” that is, the “mere consciousness … that accompanies every concept” and (c) “a concept” through which the ‘I’ is cognized (understood: as a purely intellectual object). The former is “wholly empty” because the ‘I’ of pure apperception is no more than the merely logical form of all concepts of objects überhaupt; it can only be described at all in terms of the various forms of a merely possible synthetic unity of representations, that is, those unifying logical functions of the understanding which characterize discursive thinking in general.²⁸ These subjective forms of thinking underlie the categories, which are (or can become) concepts or cognitions of objects, provided the latter are not just thought, but also given in intuition. By themselves, however, the “modi of self-consciousness in thinking are … mere functions, which provide thought with no object at all, and hence do not present my Self as an object” (B407 e.a.).

Nevertheless, the immediately following sentence in the German text of the A/B passage misleadingly states that the ‘I’ is, after all, erkannt, albeit only indirectly, “through the thoughts that are its predicates, … in abstraction [from which thoughts] we can never have even the least concept” of it. There are two difficulties here. For one thing, if Kant is not to be said to violate his own anti-circularity stricture, erkannt must be understood in its other, perfectly ordinary, German meaning of ‘recognized’ (so in the translation of Guyer/Wood), not as ‘cognized’ (Dyck 2014, p. 78) or ‘known’ (Kemp Smith). The sense of ‘recognized’ here can be glimpsed through a consideration of the second difficulty. Despite the distinct echo of a famous passage in Descartes's Principles (I, 52), the second sentence cannot be a replica of Descartes's observation that “we cannot initially become aware of a substance merely through its being an existing thing, since this alone does not of itself have any effect on us,” although we can “easily come to know a substance by one of its attributes,” since the latter do affect us (AT IX–A 25: CSM I 210). Since Descartes immediately goes on to gloss ‘attribute’ with “properties or qualities” (AT IX–A 25: CSM I 210), his point, as regards thinking substance in particular, is that the ego cogitans can only be cognized indirectly via the modi cogitandi through which it affects itself, not directly, through its mere existence. Now since the ego of which Kant is speaking is plainly not this empirical, but rather the transcendental ‘I’ of pure self-consciousness, the “thoughts” or “predicates” via which it can be “recognized” (not “cognized”) can only be those “modi of self-consciousness in thinking” (B407), which are “mere functions” or “forms” of all thinking in general. If this is correct, then the Ich of the pure Ich denke can only be “recognized … through the thoughts that are its predicates” in the second-order, transcendental reflection of the Critique. Like formal logic, the Critique is not a science “properly and objectively so called” (B ix e.a.), that is, not a science containing first-order cognition of objects, but second-order cognition of the subject itself.²⁹ And here it is the logical table of judgements that provides second-order reflection with the requisite “clue (Leitfaden)” (A76/B102) to the pure forms of synthetic unity that underlie the tables of judgements and categories alike. However, the ‘I’ that is “recognized” through these forms cannot (on pain of circularity) be said to be cognized as an object of first-order cognition, not even cognized indirectly through these “predicates.” For, as Kant notes elsewhere in the Paralogisms, “this I is no more an intuition than it is a concept of any object; rather, it is the mere form of consciousness which accompanies both sorts of representations” (A382). While it “can elevate them to cognitions … insofar as something is given in intuition, which provides the material for the representation of an object” (A382 e.a.), the pure ‘I’ or Self can never be “elevated” to a concept or first-order cognition of an object. Still, just such a circular “hypostatizing” of “the transcendental subject of thought = x” takes place in the objective (see note 29) science of pure rational psychology toward which pure reason, in its pursuit of an unconditioned cognition to complete the series of conditioned cognitions, is naturally driven in its search for a stopping-point. In this way, a condition of all intuitive or discursive cognition of objects whatsoever is mistaken for “a representation distinguishing a particular object,” the latter being, as Kant notes in the final sentence of the passage, what is meant when a representation is called “a cognition” of an object.

With the two main difficulties addressed in advance, it remains only to discuss the “constant circle” of the last three sentences of the A/B passage, which reads in its entirety:

[1] At the ground of this doctrine [a hypothetical, completely pure rational psychology] we can place nothing but the simple and in content for itself wholly empty representation I, of which one cannot even say that it is a concept, but a mere consciousness that accompanies every concept. [2] Through this I, or He, or It (the thing) which thinks nothing further is represented than a transcendental subject of thoughts = x, which is recognized (erkannt) only through the thoughts that are its predicates, and of which, in abstraction, we can never have even the least concept; [3] around which (um welches) we therefore turn in a constant circle, since we must always already avail ourselves of the representation of it at all times in order to judge anything about it; [4] we cannot separate it [“this I, or He, or It”] from this inconvenience (eine Unbequemlichkeit, die davon nicht zu trennen ist) because consciousness in itself is not even a representation distinguishing a particular object, but rather the form of representation in general, insofar as it [the representation] is to be called a cognition; for [5] of it [namely, of cognition] alone can I say that through it I think anything [i.e., an object]. (A346/B404)³⁰

The circularity problem here is obviously akin to that addressed in both the A and the B passage, though it is now posed with a completely pure rational psychology in view: we must inevitably “turn in a constant circle” should we “judge anything” about the ‘I’ of the pure ‘I think’ as an intelligible object. This is because “this I, or He, or It” is that which makes every concept and every judgement as such possible. While this is perfectly general (“judge anything”), it concerns chiefly those bogus synthetic a priori judgements that are the conclusions of the first three paralogisms and whose predicates are the categories of relation, quality, and quantity, respectively.³¹ These pure concepts are themselves made possible by those “wholly empty” forms of discursive thinking through which the pure ‘I’ can be “recognized” in second-order, transcendental reflection on the table of judgements. The pure ‘I think’ cannot therefore itself be an intelligible object of first-order synthetic a priori cognitions through the very categories that it makes possible. For, in the words of the general formulation of the transcendental anti-circularity stricture of the A passage: “I cannot cognize as itself an object that which I must presuppose in order to cognize an object at all” (A402).

So much for the three passages. It remains to consider in more detail why the argument of the first two lacks all force against historical forms of Cartesian foundationalism, whether empiricist or rationalist, and why this type of transcendental argument may nonetheless have greater weight in the context of contemporary discussions of empiricism (naturalism) and rationalism than the type to which that title is nowadays commonly assigned.³²

6. Kantian Transcendental Arguments

Contemporary transcendental arguments modelled on Kant's Refutation of Idealism seek to demonstrate the absurdity of the bodiless, worldless, solipsistic ego of Cartesian scepticism by showing that the truth of those beliefs about one's body, the external world, and other minds that Descartes declared doubtful is entailed by those beliefs about the self that he deemed certain beyond the rational possibility of a doubt, outer being a necessary condition of inner cognition in general.³³ That the Refutation should be considered a transcendental argument is at least somewhat odd, since outer is not a transcendental condition of the possibility of inner cognition; indeed, it is so far from being an a priori and merely formal subjective condition of inner cognition überhaupt that it is rather itself a cognition, and an empirical one at that. In this section, a more recognizably Kantian candidate for the honourific title ‘transcendental argument’ will be examined, first in an historical, and then in a systematic perspective.³⁴

It is a characteristically Kantian retort to the broadly foundationalist (or in Hume's case, sceptical) epistemologies of his empiricist and rationalist predecessors to insist that x cannot itself be a y (or be justified or explained by any number of y's) if x is that which makes all y's überhaupt possible. Sometimes he couches his arguments in fittingly modal language, for instance, in this remark from the Transcendental Aesthetic (directed against the Lockean and Leibnizian views that ‘space’ is an empirical idea acquired through sensation): “That within which sensations can alone be ordered and placed in a certain form cannot itself be in turn sensation” (A20/B34). This version of the rule of A402 figures as a suppressed premise in the first space argument of the Metaphysical Exposition, where the a priori character of space is demonstrated.³⁵

Hume made certain that Locke's foundationalism would founder on the shoals of scepticism by substituting impressions for ideas as the sole explanatory and/or justificatory ultimates. A transcendental argument similar to that of the Metaphysical Exposition is at work in Kant's emblematic reply to Hume in the Prolegomena: “[T]heir possibility,” he writes of the categories in general and cause-and-effect in particular, “is founded solely in the relation of the understanding to experience; not, however, in such a way that they are derived from experience, but that experience is derived from them, a completely reversed type of connection that never occurred to Hume” (4:313, cf. A301/B357). Thus, Hume too commits the error of “hypostatizing” what is for Kant (a) an a priori condition of all experience of outer events überhaupt, circularly elevating it to the level of a merely subjective impression or feeling of necessity produced in the mind by (b) a regular sequence of outer perceptions of A-like events followed by B-like events, when in fact (a) is that which makes (b) — the perception of events as distinct from enduring states of affairs — überhaupt possible.

There is no overlooking the parallel between these better-known transcendental arguments and the neglected thought experiment by which the Kant of the Paralogisms sought to radicalize Cartesian foundationalism. As he there introduces his own conceptions of space and causality in place of those of his opponents, so the Kant of the Paralogisms blithely substitutes his own ‘I think’ and radicalizing foundational project for those of Descartes, as though the former were at issue in the mixed rational psychology of his predecessors or contemporaries and not just in his own thought experiment. The key sentence reads: “[T]his proposition [Ich denke] is indeed certainly not an experience, but rather the form of apperception on which every experience depends and which precedes it” (A354). The unstated reason that the apperception in question cannot be an inner experience or empirical cognition of the self is that it is instead a transcendental condition of the possibility of all experience of objects (whether inner or outer) as such. Yet far from rigidly designating the universal and necessary form of thinking that “precedes” (A354) all particular experiences and on which experience überhaupt (as a form-matter composite) “depends” (A354), the self-consciousness expressed in Descartes's cogito is a variable whose concrete values are all those particular inner experiences of the self that he considered the first and most reliable cognitions of an empirical object. In substituting his own Ich denke for the Cartesian cogito and introducing the novel idea of a necessary condition of all empirical cognition or experience überhaupt, Kant, one is sorely tempted to say, is not so much laying bare a fallacy of the mixed rational psychology of Descartes and the tradition as simply changing the subject.³⁶

Given that his predecessors not only lacked, but probably could have made no sense of, an Ich denke that was not a reliable cognition of the self on which to ground the cognition of other things, it would obviously be question-begging on Kant's part to tax any existing doctrine of the soul with “hypostatizing” his own Ich denke as though it were a first-order cognition of a sensible or intelligible object. However, if Kant is engaged in an imaginative thought experiment involving two merely hypothetical sciences of the soul that he conceived in terms of the very level mistakes he ascribed to them, then, however question-begging the critique of Descartes and his followers in an historical perspective, the Paralogisms still retain a certain interest as a rational suasion in favour of a transcendental radicalization of Cartesian foundationalism. For like the ontological, Kant's epistemological anti-circularity stricture has a positive sense that can be put this way: if there is to be stopping-point in the process of justification, if all object-level cognition überhaupt is to be non-circularly grounded in something that is not itself part of the explanandum, then what is required is a higher-order cognition of the a priori conditions of the possibility of any object-level cognition as such. To the extent that Kant himself suggests that he is engaged in a critique of an existing doctrine or doctrines in the Paralogisms, he has both given historical scholarship a false scent and done his own thought experiment a disservice. For his reflections on level-circularity serve above all to reinforce the thesis, developed throughout the Critique, that just as (1) empirical requires grounding in pure cognition if it is to be ‘science’ in the sense that Kant and his rationalist contemporaries still gave that word (roughly: a discipline that is either completely a priori or at least rests on a priori first principles), so (2) all first-order cognition must be grounded in something “higher” (B131) if science itself is to have a stopping-point that renders the justificatory process non-circular and (so Kant believed) absolutely secure against the relentless attacks of the sceptics.³⁷

Since the failure to ascend to meta-level cognition of the subjective conditions of the possibility of object-level cognition is only logically circular where the justificandum or explanandum is the infinite totality of first-order cognition überhaupt, it is easy enough to evade the force of Kant's argument by denying that such ultimate grounding of cognition überhaupt is even possible, let alone necessary. Still, as a normative injunction to push epistemological inquiry from the object- to the meta-level, Kant's epistemological anti-circularity stricture deserves the same respect that its ontological counterpart has always received in rational cosmology and natural theology. It is not without reason that Van Cleve endorses the demurrer of “cosmological arguers” who baulk at efforts to explain the whole natural and contingent order of things without recourse to an ultimate existent of a different and higher (transcendent) order. The force of the epistemological anti-circularity stricture has been similarly felt (and continues to be felt) by what may be called ‘transcendental arguers,’ beginning with Kant. At a minimum, Stroud's picture of naturalism, on the one hand, and transcendental arguments (in his sense), on the other, as the only promising lines of attack upon Cartesian foundationalism requires supplementation by an alternative foundationalism based on the more characteristically Kantian sort of transcendental argument examined here.³⁸

7. Conclusion

The conclusion of all this is that anticipated at the end of Section 4. Since Kant's level-circularity arguments are strictly refutative only with respect to the peculiar forms of rational psychology envisaged in his thought experiment, the Paralogisms fall far short of a decisive critique of the historical Cartesian project of founding all other first-order cognition on a first-order cognition. Nevertheless, the circularity stricture formulated there is highly relevant to the question of whether it is not perhaps more desirable to transcendentalize than to naturalize Cartesian foundationalism.

In this regard, it is worth recalling that Gadamer once claimed (personal recollection) to have learnt from Heidegger that what matters in philosophy is not going beyond, but rather going back behind, one's predecessors. Heidegger himself will have learnt this lesson from Husserl, and Husserl, in turn, from Descartes and Kant. While Kant no doubt believed himself to have gone beyond his predecessors, and even to have shown all previous foundationalisms to be viciously circular, refutations that do not beg the very question at issue are rare in philosophy. Kant did, however, go back behind all his predecessors, opening up a deeper-lying, transcendental dimension to philosophical research, and therein consist both the historical importance and the systematic relevance of his Paralogisms today.