Bodies in Prolegomena §13: Noumena or Phenomena?

Abstract

This article discusses Kant's transcendental idealism in relation to his perplexing use of ‘body’ and related terms in Prolegomena §13. Here Kant admits the existence of bodies external to us, although unknown as what they might be in themselves. It is argued that we need to distinguish between a phenomenal and a noumenal use of ‘body’ to make sense of Kant's argument. The most important recent discussions of this passage, i.e., Prauss (1977), Langton (1998) and Bird (2006), are presented and shown to suffer from both systematic and exegetical shortcomings. The article is a contribution to understanding the nature of Kant's transcendental idealism, defending the view, especially against Prauss and Bird, that Kant is committed to the existence of things in themselves.

I Prolegomena §13: the dual sense interpretation

One of the most contentious issues about Kant's philosophy is the character and justification of his idealism. Some of his opponents were quick to assimilate it to Berkeley's idealism, for instance Garve or Eberhard, and we know how angry Kant was about this association, for he considered Berkeley's enthusiastic idealism absurd and as the reductio proof of any doctrine entailing it, such as Descartes's problematic idealism (A368f.). But part of the problem lies in the multifarious labels and senses of idealism (transcendental, critical, formal, empirical, material, problematic, enthusiastic, dreamy), as understood even by Kant's supporters, indeed by Kant himself. Witness the exchange between Kant and his student Beck in 1792. In reporting to Kant about the ongoing Garve-Eberhard accusations, Beck explains that even if it were true that the Critique should not have mentioned the distinction between things in themselves and appearances, it should have still focused, as in fact it did, on the conditions under which objects are given to us. He adds: ‘Appearances are objects of intuition, and everybody means just these, when he speaks of objects surrounding him, and Berkeley denied the existence of these very objects, which the Critique argued against.’¹ Kant's reply is as follows:

‘Hrn Eberhards und Garven Meynung von der Identität des Berkleyschen Idealisms mit dem critischen, den ich besser das Princip der Idealität des Raumes und der Zeit nennen könnte, verdient nicht die mindeste Aufmerksamkeit: denn ich rede von der Idealität in Ansehung der Form der Vorstellung: jene aber machen daraus Idealität derselben in Ansehung der Materie d.i. des Objects und seiner Existenz selber. — Unter dem angenommenen Nahmen Änesidemus aber hat jemand einen noch weiter gehenden Scepticism vorgetragen: nämlich daß wir gar nicht wissen können ob überhaupt unserer Vorstellung irgend etwas Anderes (als Object) correspondire, welches etwa so viel sagen möchte, als: Ob eine Vorstellung wohl Vorstellung sey (Etwas vorstelle). Denn Vorstellung bedeutet eine Bestimmung in uns, die wir auf etwas Anderes beziehen (dessen Stelle sie gleichsam in uns vertritt).’²

The two men talk past each other. For Beck the difference between Berkeley and Kant consists in the former denying the existence of outer, spatial objects (appearances), whatever one is to think about the distinction between things in themselves and appearances. Kant, however, is more interested in stressing that Berkeley is denying the existence of things in themselves, while all he is denying, is the transcendental reality of space and time as forms of representation, but not the reality of things in themselves.³ That he is talking about things in themselves here, and not just appearances, follows from the fact that the transcendental ideality he asserts does not apply to the ‘Object’ and its existence, and hence that, transcendentally speaking, the ‘Object’ really exists.⁴

This is reinforced by Kant's rejection of Schulze's Änesidemus by means of a kind of ‘semantic’ argument committing us to things in themselves.⁵ Again, ‘Object’ means here things in themselves, not appearances, for appearances are nothing but Vorstellungen (A250), and in the Änesidemus passage above he is contrasting Object with Vorstellung. This does not mean that Kant did not also attribute to Berkeley the denial of the existence of outer, spatial objects, for he did, in the Refutation of Idealism (B274). Rather, what this suggests is that Kant is operating with different notions of idealism even with respect to Berkeley. While Kant has specific labels for Descartes’ (‘problematic’, B274) and for his own idealism (‘transcendental’, ‘critical’, ‘formal’), he does not have two different labels to distinguish between a Berkeley who denies the existence of outer, spatial objects (phenomena), and a Berkeley who denies the existence of things in themselves (noumena). We can call the first Phenomenal-Berkeley, and the second Noumenal-Berkeley.

Kant distinguished his idealism from Berkeley's implicitly and explicitly in the Critique, (A368, B71, B275), but especially in the Prolegomena §13, Remarks II and III, and in the appendix. The most striking and perplexing passage occurs in Remark II (I number the relevant sentences):

Q1: ‘(1) Der Idealismus besteht in der Behauptung, daß es keine andere als denkende Wesen gebe, die übrigen Dinge, die wir in der Anschauung wahrzunehmen glauben, wären nur Vorstellungen in den denkenden Wesen, denen in der That kein außerhalb diesen befindlicher Gegenstand correspondirte.

(2) Ich dagegen sage: es sind uns Dinge als außer uns befindliche Gegenstände unserer Sinne gegeben, allein von dem, was sie an sich selbst sein mögen, wissen wir nichts, sondern kennen nur ihre Erscheinungen, d.i. die Vorstellungen, die sie in uns wirken, indem sie unsere Sinne afficiren.

(3) Demnach gestehe ich allerdings, daß es außer uns Körper gebe, d.i. Dinge, die, obzwar nach dem, was sie an sich selbst sein mögen, uns gänzlich unbekannt, wir durch die Vorstellungen kennen, welche ihr Einfluß auf unsre Sinnlichkeit uns verschafft, und denen wir die Benennung eines Körpers geben; welches Wort also blos die Erscheinung jenes uns unbekannten, aber nichts desto weniger wirklichen Gegenstandes bedeutet.

(4) Kann man dieses wohl Idealismus nennen? Es ist gerade das Gegentheil davon.’⁶ (AA 04: 288.34ff.)

The picture presented here appears straightforward. Idealism, taken as the ontological doctrine denying that anything else exists apart from thinking beings, and their representations, to which nothing real corresponds (1), is contrasted with a doctrine according to which there are, indeed there must be things corresponding to representations (cf. AA 04: 355.05-08). What these things are in themselves is unknown to us, but they do affect our senses, causing (‘wirken’) us to have representations of them, and we thus know them at least through their representations. The possessive construction ‘their representations’ (‘ihre Erscheinungen’), which occurs several times (AA 04: 287.28, 288.28, 289.05, 293.02-03), is important, since it suggests that things are not identical to their appearances; the passage is at any rate counterevidence to the double-aspect reading of ‘things’ as ‘things as they are in themselves’ and ‘things as they appear to us’ (see below). This also conforms with the confusion between appearances, as arising from an unknown cause (the transcendental object), and this unknown cause itself, i.e. the confusion (subreption) between appearances and their noumenal ground discussed extensively in A386-391.

The doctrine espoused in Q1 sounds, but is not really Lockean indirect realism, for as Kant explains in the ensuing paragraph (04: 289.15ff.), he counts even what is called primary qualities as mere appearances, mere modifications of the senses. Does this make appearances relational properties of the things unknown as they are in themselves? No, for there are no properties left for such a claim. All properties constituting the intuition of a body are properties of appearances (289.30f). Properties of appearances may be construed as relational properties, but only insofar as appearances are mere modifications of our sensibility (A490-1/B518-9). This does not make them relational properties of those things unknown in themselves which affect our sensibility. But if no properties, what else is ‘known’, or rather accepted, presupposed about things as they are in themselves? The one feature of a thing which is not a property: its being. For being is not a real predicate (A598/B626).⁷ And this is precisely Kant's own qualification of his idealism: admitting that all properties are phenomenal does not contradict ‘the existence of the thing which appears’ (‘die Existenz des Dinges, was erscheint’, AA 04: 289.32).⁸ Just as he will explain to Beck in 1792: Kant's idealism asserts the ideality of the form of the representation of the thing, not its matter. Remark II is thus about Noumenal-Berkeley, not Phenomenal-Berkeley. This reading is consistent with other parts in the Prolegomena, especially §13, Remark III:

Q2: ‘(1) Denn dieser von mir sogenannte Idealism betraf nicht die Existenz der Sachen (die Bezweifelung derselben aber macht eigentlich den Idealism in recipirter Bedeutung aus), denn die zu bezweifeln, ist mir niemals in den Sinn gekommen, sondern blos die sinnliche Vorstellung der Sachen, dazu Raum und Zeit zu oberst gehören, und von diesen, mithin überhaupt von allen Erscheinungen habe ich nur gezeigt: daß sie nicht Sachen (sondern bloße Vorstellungsarten), auch nicht den Sachen an sich selbst angehörige Bestimmungen sind.

(2) Das Wort transscendental aber, welches bei mir niemals eine Beziehung unserer Erkenntniß auf Dinge, sondern nur aufs Erkenntnißvermögen bedeutet, sollte diese Mißdeutung verhüten.’⁹ (AA 04: 293.20-30)

Here too Kant stresses that his idealism does not concern the existence of things in themselves, but only their sensory representation. To assume that they have a sensory representation is to assume they exist. He does not use the more standard phrase ‘Dinge an sich’, but his use of ‘Sachen’ and ‘Sachen an sich’ is an unproblematic rhetorical substitute.¹⁰ Equally, he makes clear that his transcendental idealism does not involve any relation to things, and thus no relational properties of them, but only to the faculty of cognition.

There are, however, two problems with Q1, pertaining to sentences (2) and (3). I focus on (3) first. How is Kant's use of ‘outside’ or ‘external’ (‘außer’) and ‘bodies’ (‘Körper’) to be understood? It seems that he means by ‘bodies’ appearances of the outer sense (‘die Benennung eines Körpers […] welches Wort also blos die Erscheinung […] bedeutet’; ‘which we call bodies; a term, therefore, signifying merely the appearance’). A related problem concerns his use of relative clauses, especially in Q1(3). When he claims that there are bodies outside of us (‘außer uns Körper’), i.e. things, which, while unknown as they are in themselves, we know through representations, and which (‘denen’) we give the designation ‘body’, a term that only means the appearance of the unknown thing, what does he mean by the ‘which’? Syntactically, especially in the original, it seems to refer to bodies = unknown things, but this is inconsistent with his indication that by ‘bodies’ he means appearances of the outer sense. Hence, ‘which’ refers to bodies = representations = appearances.

This interpretation is consistent with various other passages about bodies, in fact with the paragraph preceding Q1, in which, reflecting the conclusion of the Aesthetic, it is explicitly stated that bodies are in space, and are together with space mere appearances in us (288.30-32). A similar view is expressed in B69, in A387, and in Kant's reply to Eberhard:

‘[…] die Körper [sind] gar nicht Dinge an sich selbst und ihre Sinnenvorstellung, die wir mit dem Namen der körperlichen Dinge belegen, [ist] nichts als die Erscheinung von irgend etwas […], was als Ding an sich selbst allein das Einfache enthalten kann, für uns aber gänzlich unerkennbar bleibt.’¹¹ (AA 08: 209.16-20)

Clearly, it is the sensory appearances which are designated as ‘bodies’ or ‘bodily things’. Kant says something similar in a note in the Nachlass: ‘Die Frage, ob die Körper außer mir etwas Wirkliches sind, wird so beantwortet: Körper sind außer meiner Sinnlichkeit keine Körper (phenomena), und also sind sie nur in der Vorstellungskraft empfindender Wesen.’¹² (AA 14: 586.14-17, Nr. 4536)

But if ‘bodies outside of us’ really refers to appearances of the outer sense in Q1, then the reading given above is mistaken, and really Remark II is about Phenomenal-Berkeley – as are then Remark III and many other passages in the Prolegomena and beyond explaining transcendental idealism. An ontological understanding of transcendental idealism, as a doctrine committed to the existence of things in themselves, would be at peril, if Phenomenal-Berkeley is granted presence in Remark II.

Moreover, a purely phenomenal interpretation makes Q1(3) inconsistent. Kant affirms that there are bodies outside of us, things about which we don't know what they are in themselves. If these bodies were indeed appearances, it would immediately contradict the fundamental claim of critical philosophy, i.e. that if anything is known to us, it is appearances. Q1(2) has the same problem, indeed it is even more pressing, since it claims that of things as objects of our senses outside of us (‘Dinge als außer uns befindliche Gegenstände unserer Sinne’) we don't know what they are in themselves. Since Kant usually equates objects of our senses with appearances, it seems that Q1(2) claims that we don't know what appearances are in themselves.

Kemp Smith's solution to Q1(3) is to declare Kant's use of ‘body’ without justification (1930: 306). We could accept this, point out, with Georg Kullmann (1922), that the Prolegomena are Kant's worst edited book (see Pollok 2001: liii), and conclude that the whole passage is muddled and polemical. But we can give a better answer. ‘Outside of us’ (and ‘inside of us’) is not used univocally by Kant, as he explains in the Fourth Paralogism (A373) and his discussion of the fallacy of subreption (A386-391). The issue is actually addressed later in the Prolegomena. In §49 he is concerned with proving that there are objects outside of us in the empirical sense. Yes, there are such objects, for to be empirically outside me is to be spatial, and to be spatial is to be transcendentally ideal, to belong to the sphere of representations (AA 04: 336.10ff.), and of those we are immediately aware and certain. This allays the Cartesian dream argument, which presupposes space and time as conditions of appearances. It also refutes Phenomenal-Berkeley.

This empirical sense of ‘outside’ has to be distinguished from a second, a transcendental sense, concerning that which exists outside of my faculty of representation (‘außer meiner Vorstellungskraft’, AA 04: 337:27).¹³ It is true that at AA 04: 337:17-18 Kant claims that ‘außer uns’ means only existence in space. But if this were so, many of his rejections of transcendental realism would make no sense (or involve a category mistake), e.g. ‘all objects of any experience possible to us […] have no independent existence outside our thoughts’ (B518-9). Equally, ‘body’ has two senses as well, meaning ‘nicht blos die äußere Anschauung (im Raume), sondern auch das Ding an sich selbst […], was dieser Erscheinung zum Grunde liegt’ (AA 337.20-21; ‘not only external intuition (in space), but the thing in itself, which is the basis of this phenomenon’, Kant 1977: 73). This dual sense is immediately employed (‘_N’ specifies a noumenal, ‘_P’ a phenomenal gloss): ‘so kann die Frage, ob die Körper_P (als Erscheinungen des äußern_P Sinnes) außer_N meinen Gedanken als Körper_N in der Natur existiren, ohne alles Bedenken verneint werden’ (AA 04: 337.22-24; ‘and so the question, whether bodies_P (as phenomena of the external sense) exist as bodies_N in nature apart from my thoughts, may without any hesitation be denied’, Kant 1977: 73).¹⁴ Note how Kant commits himself, once again, to the existence of the thing in itself, as he always does when faced with the accusation of idealism.

Thus Kant provides himself the means to solve the puzzle of Q1, as ‘outside’ and ‘body’ have a phenomenal and a noumenal sense. In Q1(3) these two senses are not explicitly distinguished and in addition obscured by the relative clauses. But we can untangle the sentence by introducing indices for the two main strands of anaphoric reference:

‘(3) Demnach gestehe ich allerdings, daß es außer_N uns Körper_N gebe, d.i. Dinge₁, die₁, obzwar nach dem, was sie₁ an sich selbst sein mögen, uns gänzlich unbekannt, wir durch die Vorstellungen₂ kennen, welche₂ ihr₁ Einfluß auf unsre Sinnlichkeit uns verschafft, und denen₂ wir die Benennung eines Körpers_P geben; welches Wort also blos die Erscheinung jenes uns unbekannten, aber nichts desto weniger wirklichen Gegenstandes bedeutet.’

‘(3) Consequently I do grant that there are bodies_N outside_N of us, i.e. things₁ which₁, although quite unknown to us as to what they₁ are in themselves, we yet know by the representations₂ which₂ their₁ influence on our sensibility procures us, and which₂ we call bodies_P; a term, therefore, signifying merely the appearance of that object unknown to us, but nevertheless actual.’

We can now also make sense of Q1(2). For reasons already specified, the things as objects of our senses outside of us (‘Dinge als außer uns befindliche Gegenstände unserer Sinne’) cannot be phenomena in space. Hence, they are things in themselves, and ‘outside’ has here a noumenal gloss. Admittedly, this puts a certain strain on ‘objects of our senses’, since we are inclined to understand this as ‘appearances’. But as with so many other important phrases in Kant, this one has also a dual sense, empirical and transcendental. Taken transcendentally, the phrase refers to things in themselves, standing for the unknown ground or ‘cause’ of our appearances as a whole. In fact, the phrase occurs in the Critique on several occasions, most prominently in the discussion of subreption already mentioned: ‘that which is the true (transcendental) object of our outer senses cannot be the cause of those representations (appearances)’ (A390, see also A392).

While the interpretation given here contains the slight complication of claiming a dual sense for the terms ‘body’ and ‘outside’, it is conforming to the overall Kantian doctrine and terminology. It also demonstrates that the reality of the thing in itself is taken for granted by Kant, and is not to be confused with the appearances it is a ground of, lest we want to commit the fallacy of subreption.

II Prauss, Langton and Bird on Prolegomena §13

II.I Prauss: the double-aspect view

While the difficulty of Q1 has been discussed in the literature, the disambiguating reading offered above is new. It does, however, agree with some of the commentaries offered in the past at least in one respect: the assumption that Q1 is principally concerned with expressing Kant's acknowledgement that his idealism does not deny the existence of things in themselves. As Sidgwick wrote against Caird's interpretation (in Caird 1889): ‘I do not see how Mr. Caird can hold that Kant when he wrote these passages regarded as “meaningless” the question “whether there is an existence of things in themselves independent of consciousness”‘ (1879: 410).¹⁵ Similar verdicts were reached by Apel (1908: 135, 144), Erdmann (1878: lxviii), Kemp Smith (1930: 306), Turbayne (1955: 238, 240), Walker (1978: 134), van Cleve (1999: 136), to mention but a few commentators. The main dissenting view was put forward by Prauss (1977). His interpretation of the passage is consistent, indeed taken as evidence for his overall interpretation of the problem of things in themselves, i.e. that Kant was not concerned with a two-world view distinguishing between appearances and things in themselves, accompanied with a double-affection doctrine (Adickes), rather with two levels of transcendental reflection on the empirical objects of our experience (and nothing else), regarding them as appearances in the first reflection, and as things in themselves (‘an sich selbst betrachtet’) in the second reflection.¹⁶

Accordingly, textual passages which appear to offer evidence in favour of the two-world and double-affection doctrine view are partly due to Kant's own confusion about these two levels (Prauss 1977: 199). Prolegomena §13.II is a particularly good example for this, in Prauss’ view (1977: 201-3). Q1(2) is apparently clearly about outer appearances, objects of our senses (‘Gegenstände unserer Sinne’), and we are only concerned with a transcendental reflection on them when we say that we don't know anything about what they are in themselves, which does not imply any noumenal affection. By contrast, Prauss argues, Q1(3) muddies the waters, because its final wording suggests there is an unknown object in itself (‘jenes uns unbekannten, aber nichts desto weniger wirklichen Gegenstandes’).

But it is really Prauss’ reading of this passage, whatever the merits of his overall reading of Kant, which is problematic. If Q1(2) were about outer appearances, we would end up with an unacceptable position: bodies as outer appearances, which are, as all appearances, modifications of my faculty of representation, and as such representations, affect our senses and cause representations in us. The double-aspect view turns into a double-representation view, with one set of representations (outer appearances) causing another set of representations (inner appearances?).¹⁷ On this view we would not even know what outer appearances are, since ‘we only know their representations’ (‘kennen nur ihre Erscheinungen’), a phrasing which clearly suggests that we must distinguish between the second set of representations and their cause. We would thus not be able to escape even scepticism about outer appearances, i.e. the kind of scepticism Kant assumes that follows from Descartes’ empirical idealism (see A367ff.). But as we have seen, the ‘objects of our senses’ Kant has in mind in Q1(2) are not at all the outer appearances Prauss claims they are, rather, according to the dual sense of ‘outside’, ‘bodies’ etc., things in themselves.

II.II Langton: the dual properties view

More recently, two other interpretations of Q1 were offered, which contradict the dual sense interpretation, but also the traditional reading. They are also not compatible with each other. The first was advanced by Rae Langton (1998), the second by Graham Bird (2006). Like Prauss, Langton accommodates her interpretation of Q1 within her overall reading of Kant, which states: things in themselves are substances that have intrinsic properties; we don't (cannot) know these intrinsic properties; appearances are relational properties of substances; we (can) know these relational properties, because our knowledge depends on receptivity, i.e. on how things affect, and thus relate to us (1998: 20-23). Prolegomena §13, Remark II is to be understood as extending Locke's category of secondary qualities to all qualities, including primarias.¹⁸ But this does not make Kant a Berkeleian: all properties of phenomenal objects are secondary qualities, in Locke's ontological sense that they are relational properties of things, and this does not deny the existence of things and their intrinsic, if unknowable properties. And properties of phenomenal objects are at the same time primary qualities, in Locke's epistemological sense that they are the properties of the bodies science investigates, which might well be hidden, and not manifest, unlike colours.

So the view proposed in Remark II is: ‘the properties of phenomenal objects are those relational properties ascribed to bodies by science’ (Langton 1998: 156). Following this view, Langton then shows how Q1 can be absolved of paradox: Q1 presents us with a series of equivalences which leads to the identification of bodies with (a) things unknown to us as they are in themselves, (b) with appearances, (c) with representations, which would give us the incoherent identity ‘representations = things unknown to us as they are in themselves’ (Langton 1998: 158f.), an incoherence also noted in section I above. ‘Something has to give’ to save Kant from incoherence, i.e. we have to modify the equivalences to yield a more plausible position. She suggests we modify the equivalence ‘appearances = representations’ implied in Q1(2) (‘ihre Erscheinungen, d.i. die Vorstellungen’, ‘their reps, i.e. the representations’) according to what is implied in Q1(3), i.e. that bodies are known through representations, as opposed to that they are representations. ‘Body’ means ‘the appearance of the real thing’ (Langton 1998: 159), i.e. the term refers to the relational properties of the real thing.

This solution has the advantage of simplicity, but does not ultimately succeed. The view that appearances are relational properties of the things unknown as they are in themselves was already rejected in section I. But there are other problems as well. Langton's interpretation does violence to the text. Kant says clearly that we know the representations of things, i.e. we know appearances, and he also says that we know things (bodies_N) through their representations. I can know a photo of Jerusalem and I can know Jerusalem through the photo. This distinction is in the text and should stay there. In addition, Langton's reading puts a strain on an important anaphoric chain in Q1(3). Kant says that we know things through the representations which (‘welche’) their influence on our sensibility provides for us, and to which (‘denen’) we give the name of bodies. On a natural reading both ‘which’ refer to ‘representations’, but on Langton's reading this is not possible, since this would mean that we give representations the name of bodies, precisely what she wants to avoid. Hence, she must assume that the second ‘which’ (‘denen’) refers all the way back to ‘things’ (‘Dinge’). This turns into self-refutation, because it means that Kant is saying we are giving the name of bodies to things, and precisely not to appearances, as Langton wishes to maintain.

Even if we were to accept Langton's view that ‘appearance’ stands for the relational properties of things, we could not coherently hold both that ‘body’ means, univocally, ‘the appearance of the real thing’ and also ‘the real thing’. The properties of a thing are never identical with it, especially if they are relational properties, especially on Langton's view, which admits of intrinsic properties of things as well.¹⁹ Hence, ‘body’ cannot mean at the same time and in the same respect ‘the real thing’ and ‘the appearance = the relational properties of the real things’. But this is just what Langton's reading is committed to. Explaining how her reading clarifies Q1 she writes: ‘We know nothing of what those things may be in themselves. […] Those things are bodies outside us. […] “Bodies” means the appearance of the real object [thing], which is unknown to us as it is in itself’ (Langton 1998: 159). In other words: things = appearances. Evidently, the chain of implausible equivalences has only been given a different form here, but has not been severed.

There are further problems with Kant's position thus understood. There appear to be three elements in the chain now: things, their appearances (bodies), representations. We know things insofar as we know their relational properties, i.e. bodies, which are known through representations. But how so? It may well be that ‘representations’ are all there is. We cannot get to the real things anyway, as we cannot know their intrinsic properties. We supposedly know only their relational properties, as appearances, i.e. as real things relate to us, our sensibility. But we don't know even these appearances directly. As seen, Langton excludes that appearances are representations. Since representations are certainly end-products of some causal chain, either representations are caused by appearances or representations are caused by things.²⁰ If representations are caused by things, then, given that representations are not identical to appearances, we have a problem with explaining the precise relation between representations and appearances. If representations are caused by appearances, we slide into exactly the kind of Cartesian idealism/scepticism Kant wishes to avoid in the Fourth Paralogism, where he explicitly declares outer objects to be nothing but representations of which we are directly aware, which forestalls the need to give a genuine (inferential) proof for the existence of outer objects (A371). The identification of appearances with representations is a recurrent theme of great systematic importance in the Critique (see A129, A250, A490-1/B518-9), and a lot would have to give if it were denied. Also, if representations were caused by appearances, this would install yet another veil of ideas, this time between appearances and representations. The only other way to avoid Cartesian idealism/scepticism would be to reapply Langton's dual properties view to the representations/appearances divide itself, i.e. to say that representations give us at least the relational properties of appearances, if not their intrinsic properties. We would have to profess partial ignorance again, this time about appearances. From this (intrinsically absurd) position an infinite regress ensues. Despite being based on very different premises, Langton's interpretation of Q1 suffers from consequences and shortcomings similar to Prauss’. The solution lies in assuming the correlated ambiguities of the terms ‘body’, ‘outside’ etc., as offered in the dual sense interpretation above.

II.III Bird: the noumenally non-committal view

The most recent detailed interpretation of Prolegomena §13 was offered by Graham Bird (2006: 207-19), who offers both a nuanced discussion of the problematic passages and a critical discussion of Langton's view. I will focus on the former.²¹ His interpretation of §13 is based on his overall understanding of Kant's project, which has some affinities with Prauss's views. One of the most important aspects of this is the conviction that the contrast between things in themselves and appearances is a purely formal, not an ontological contrast. Kant's idealism does not claim our ignorance about a transcendent, nevertheless real world, and our knowledge about the natural domain as ‘constructed’ by our (noumenal) mind. Such a contrast would confuse the transcendent with the transcendental (Bird 2006: 5). ‘Transcendental’ designates a kind of enquiry, namely into the possibility of a priori knowledge of first-order objects (2006: 53, 83ff., also B25, B80-1). It is a descriptive, not a normative enquiry, giving us an inventory of the a priori structure of our experience (2006: 15).²² Accordingly, ‘things in themselves’ does not designate real things, but mere conceivables, which play no positive role in describing the a priori nature of immanent experience, but are part of Kant's diagnosis of the rationalists’ temptation to conjure up chimeras as objects of proper philosophical knowledge (Bird 2006: 210). In the light of this, Prolegomena §13 cannot make any claims committed to the reality of things in themselves. Bird demonstrates this with respect to a passage already occurring in Remark I: ‘that our sense representation is not a representation of things in themselves but of the way in which they appear to us’ (‘daß unsre sinnliche Vorstellung keinesweges eine Vorstellung der Dinge an sich selbst, sondern nur der Art sei, wie sie uns erscheinen’, AA 04: 287.05-06). This would be inconsistent if it meant that our representations do not represent things in themselves, but things in themselves as they appear to us. What Kant rather means is that ‘objects are represented to us as appearances, but [not] as “things in themselves”‘ (Bird 2006: 208). The double quotation marks are meant to indicate a predicative use, devoid of existential commitment. Bird saves the passage from inconsistency by appealing to Kant's twofold conception of objects of experience as appearances and as things in themselves, only the former of which is justified.

Bird presents this as an admittedly complex reading, but also as natural and correct (ibid.). This is problematic. The passage claims quite straightforwardly that representations do not represent things in themselves, but the manner (‘Art’) in which they (‘sie’) appear to us. The stress is on the manner of appearance, as this is the content of the representation. The inconsistency arises only if one takes ‘as they appear to us’ (‘wie sie uns erscheinen’) to be an extensional modifier on ‘things in themselves’. In fact, it is an intensional modifier (cf. ‘The ghost of Banquo enters, and sits in Macbeth's place’, which does not entail ‘Banquo enters, and sits in Macbeth's place’). As seen from the letter to Beck discussed above, Kant views his idealism as concerning the mode of representation of things, not their existence, which is unquestioned.

Bird's claim that objects are represented to us as appearances also faces the following problem: given his rejection of a commitment to things in themselves, objects can only be appearances, in which case the claim becomes vacuous or untenable, as it entails that appearances are represented to us as appearances. Either this means that appearances are appearances or that appearances can appear (actually, really) as something else as well. Furthermore, it is true that Kant presents a double-aspect view in the Second Preface of the Critique (Bxviii-xxii). But what this view exactly amounts to in this noteworthy passage is unclear. Kant does say, on the one hand, that the project of the Critique can be understood as testing, successfully, the distinction between treating the same objects either as objects of the senses and the understanding or as mere thinkable objects of reason (Bxviii-xix). On the other hand, he shows commitment to the reality of things in themselves, since he also explains that the test confirms that our Vernunfterkenntnis ‘nur auf Erscheinungen gehe, die Sache an sich selbst dagegen zwar als für sich wirklich, aber von uns unerkannt liegen lasse’ (‘has to do only with appearances, and must leave the thing in itself as indeed real per se, but as not known by us’, Bxx).²³

This settles the ground for Bird's discussion of Prolegomena §13.II. He focuses on Q1(3) and Q2. In Q1(3) ‘bodies’ refers to outer appearances, claiming that we have no knowledge of them as what they may be in themselves, i.e. taken ‘as things in themselves’ (Bird 2006: 209), a formulation without existential commitment. Bird admits that the last clause of Q1(3) (‘the appearance of that object unknown to us, but nevertheless actual’) seems to commit Kant to the reality of things in themselves. Three readings are given to deal with this passage. (a) According to the first, Kant identifies what we call ‘bodies’, i.e. appearances, with things in themselves. (b) According to the second, what we call ‘bodies’ are appearances of real, unknown things in themselves. (c) According to the third, bodies are known as appearances, but unknown as things in themselves (ibid.). On all three views bodies causally affect our senses. The first view is dismissed as implausible. The second is more faithful to Kant, but ultimately also problematic, because it still assumes the existence of things in themselves (the ‘of’ of appearances).²⁴ The third is the most faithful reading. Overall, readings two and three express Kant's repudiation of idealism as a claim that there are not outer, spatial objects, but only inner, mental ideas, whereas reading one rejects idealism as a doctrine about the existence of things in themselves. Evidently, Bird takes Kant to be concerned with Phenomenal-Berkeley in Q1, not Noumenal-Berkeley. Bird points to the Refutation of Idealism (B274ff.; see also Fourth Paralogism, A367ff.), since there Kant expressly attempts to prove the existence of outer spatial objects. Q1 is claimed to have the same aim.

The first problem with this interpretation is that it does not pay tribute to the whole text in Q1. In particular, it ignores Q1(2), in which it is not just stated, like in Q1(3), that we know bodies (which one might be tempted to understand as outer appearances) through their influence on our senses, but more specifically that there are things external to us: these are unknown as to what they might be in themselves, but we do know their appearances, i.e. the representations they effect in us. The possessive pronoun does suggest that Kant does not speak here about things unknown in themselves merely in the predicative sense. Since it is these things which Kant calls bodies at the beginning of Q1(3), and which Bird takes to be outer appearances, we have a problem: bodies as outer appearances (Q1(3)) are unknown as to what they might be in themselves, but they have knowable appearances (Q1(2)). This entails an implausible double-appearance view (equivalent to Prauss's double-representation view), for the outer appearances bodies are cannot be the appearances bodies have. As pointed out, the conceptually significant possessive construction occurs frequently in Kant's writings.

Q1(2) is also remarkable, because it classifies appearances as representations, a leitmotif in Kant's critical philosophy. The idealism Kant is concerned with in Q1 is specified in Q1(1): it is the doctrine that thinking beings and their representations are the only existing things. Pointing to the Refutation of Idealism will not help against an idealism thus understood, for the Refutation of Idealism merely proves that consciousness of my temporality, as an inner experience, necessarily entails the existence of external things and their immediate experience (B275-7). As such, these external things are objects of experience, appearances. But as appearances they are after all just representations, modifications of my Vorstellungskraft. Time and again Kant stresses this in key passages: ‘what we call outer objects are nothing but mere representations of our sensibility, the form of which is space’ (A30/B45); ‘[…] all objects with which we can occupy ourselves, are one and all in me, that is, are determinations of my identical self’ (A129). This explains why the inference to the existence of outer objects is precisely not Cartesian, i.e. not a causal, indirect inference (a point also exploited in the Fourth Paralogism, A368). So the Refutation of Idealism does not refute the idealism specified on Q1(1) – it confirms it: there are objects ‘external’ to me, i.e. spatially external and they are all representations, of which I am directly aware, because they are representations.²⁵ Kant needs precisely things in themselves, i.e. non-appearances, which, while they can never become object of my experience and knowledge with respect to their properties, guarantee, through the necessary and unquestioned assumption of their existence that what we deem appearance and experience truly are appearance and experience, as seen from the Änesidemus passage above and Bxxvi-xxvii, A251, B306. As Vaihinger puts it: ‘things in themselves stand firmly like a wall of palisades behind appearances, prevent their dissolution into illusion and forbid the association Kant's with Berkeley’ (1922: 505; my translation). With Noumenal-Berkeley, as we should add.

If this interpretation is along the right lines, then it is correct, with Bird, and indeed Beck (see Vaihinger 1922: 502), to point out that the Refutation of Idealism does not aim to prove the existence of noumena. But it is incorrect to infer from this that Kant does not need or accept the existence of noumena.²⁶ He very clearly does:

‘Die Sinnenwelt ist nichts als eine Kette nach allgemeinen Gesetzen verknüpfter Erscheinungen, sie hat also kein Bestehen für sich, sie ist eigentlich nicht das Ding an sich selbst und bezieht sich also nothwendig auf das, was den Grund dieser Erscheinung enthält, auf Wesen, die nicht blos als Erscheinung, sondern als Dinge an sich selbst erkannt werden können. In der Erkenntniß derselben kann Vernunft allein hoffen, ihr Verlangen nach Vollständigkeit im Fortgange vom Bedingten zu dessen Bedingungen einmal befriedigt zu sehen.’²⁷ (AA 04: 354.05-12)

Notice here how Kant justifies or rather explains his commitment to noumena: not via a proof, but via appeal to the Ideal of Reason (the unconditioned; see A567ff./B595ff.). Kant does not offer, or even thinks there is need for, a proof of the existence of noumena. There is in fact a substantial difference between Kant's talk about the existence of noumena and the existence of phenomena: about the former he merely says, as in Q2, that he never doubted them (AA 04: 293.20-23; see also Walker 1978: 134), about the latter that he is proving their existence (e.g. AA 04: 336.02-05). The relation between noumena and the experiencing subject is not one of possible cognition, but of fundamental presupposition, of ‘granting’ (‘gestehen’): ‘wenn wir die Gegenstände der Sinne wie billig als bloße Erscheinungen ansehen, so gestehen wir hiedurch doch zugleich, daß ihnen ein Ding an sich selbst zum Grunde liege, ob wir dasselbe gleich nicht, wie es an sich beschaffen sei, sondern nur seine Erscheinung […] kennen’ (AA 04: 314.33-315.02).²⁸ Without the necessary (‘unvermeidlich’) presupposition of intellectual beings as the ground of appearances everything would indeed become mere appearance (AA 04: 315.02-10) and we would have no means to escape Noumenal-Berkeley. But the existence of noumena is not doubted, and cannot be doubted. To be a dubitable object or property of an object, the object has to be an object of experience, and the property has to be a phenomenal property (‘beschaffen sei’). For a property to be dubitable it has to be empirically knowable and vice versa. But as pointed above, existence is not a property at all (A598/B626). Hence, the empirical, i.e. qualitative unknowability of noumena is compatible with the necessary presupposition of their existence, necessary for making sense of experience and of appearance (as being mere appearance and not all there is).²⁹ The only sense in which noumena are ‘knowable’, on this account, concerns their existence, as the unquestionable background against which we can think coherently of experience and appearances as such. There is, therefore, a difference to be made between rejecting Noumenal-Berkeley (as absurd) and refuting Phenomenal-Berkeley (as false).

Of course, this does not turn appearances into something identical with things in themselves, or at least with their (relational) properties. Bird is correct to maintain, against Langton, that Kant could not have identified, and did not identify things in themselves with appearances, which is the point of departure for Langton's interpretation of Q1 (see above), since Kant's denial of knowledge of appearances as things in themselves denies precisely that identification (see Bird 2006: 217). Consequently, we can also not be said, pace Langton, to know things in themselves as objects of the senses (as bodies_P, as phenomena), for the former are unknowable and the latter are knowable (ibid.). Note, however, that, as we have seen, there is still a sense in which ‘objects of the senses’ may mean ‘things in themselves’, as in A390 and A392. This identification does not entail the knowability of things in themselves, since ‘objects of the senses’ has a noumenal gloss here. Kant's dual use of ‘objects of the senses’ and related terms must be borne in mind.

Bird quotes A44/B62, where Kant argues, against Leibniz, that our sensibility does not give us some kind of knowledge about things in themselves; rather, it does not give us any such knowledge. But this passage also serves to show that the predicative use of ‘things in themselves’, favoured by Bird, is not, and cannot be, the only one employed by Kant. What this passage precisely tells us is that we don't know the properties (‘Beschaffenheit’) of things in themselves, not that we don't ‘know’, or are not committed to assume their existence. Shortly before, in the same section in the Transcendental Aesthetic, Kant uses the term also in a non-predicative sense, indeed in a way strongly undermining Bird's interpretation: ‘Was es für eine Bewandtniß mit den Gegenständen an sich und abgesondert von aller dieser Receptivität unserer Sinnlichkeit haben möge, bleibt uns gänzlich unbekannt. Wir kennen nichts als unsere Art, sie wahrzunehmen, die uns eigenthümlich ist’ (A42/B59, my italics).³⁰ We might take this formulation as evidence for Langton's view, i.e. that we do, in some sense, perceive things in themselves (as bodies_P), and get entangled again in the problems already discussed. Alternatively, we might take this as a passage congenial to A390 or A392, expressing, in a slightly misleading way, our commitment to things in themselves as objects of our senses, the latter taken precisely not with a phenomenal gloss, but as setting, at least in the minimal sense, the limits to sensibility (A256/B312).

Bird dedicates a detailed discussion to Q2 as well, attempting to refute the view, defended by Kemp Smith (1930: 306) and more recently Walker (1978: 134), the latter a direct target of Bird's arguments, that the passage is concerned with things in themselves. According to Bird, when Kant asserts that it never occurred to him to doubt the existence of Sachen (things), it is far from clear that he means things in themselves.³¹ At best he opens up a contrast between Sachen at the beginning of Q2(1), whose existence he never doubted, and Sachen an sich (things in themselves) towards the end of Q2(1), which space and time are denied to be properties of. In the former occurrence ‘Sachen’ just means outer appearances, and in the latter ‘things in themselves’, but only in a predicative sense, without carrying any commitment to the existence of things in themselves (Bird 2006: 212f.). So really, what Kant never doubted was the existence of outer appearances, and what he did doubt, indeed treated as illusions, were things in themselves. Walker's view that Kant never doubted the existence of things in themselves clashes with Kant's repudiation of traditional idealism (‘Only reason yields truth’) and his endorsement of the opposite view (‘Only experience yields truth’, AA 04: 374.19-20).

This interpretation of Q2(1) is problematic. We have already seen that Kant does need things in themselves and is committed to them. He does indeed not doubt their existence, but it does not follow from this that he claims, like the rationalists, that he has genuine knowledge about them via pure reason, i.e. knowledge about their properties. Lack of doubt about the existence of things in themselves does not place Kant back into the rationalists’ camp, because, unlike them, he has tied genuine knowledge to experience.

Neither the direct nor the surrounding textual evidence favours Bird's interpretation. In Q2(1) there does not seem to be a real contrast between phenomenal Sachen and noumenal Sachen an sich. First, Kant says clearly that he has doubted the sensory representation of Sachen, to which space and time belong, but no more (see also AA 04: 292.35-293.03). If Sachen are indeed outer appearances, Kant is saying here that he has doubted the sensory representation of outer appearances, including space and time. But according to transcendental idealism space and time are necessary forms of the sensory representation of outer appearances, so there is no room for doubt here. Second, Kant claims to have demonstrated that appearances are not Sachen, but modes of representation (Vorstellungsarten). Assuming Sachen are outer appearances makes this statement inconsistent: appearances are not outer appearances. Also, this characterisation of transcendental idealism, i.e. that appearances are not Sachen, correlates with passages in the Critique where Kant specifies that all appearances are mere modifications of our sensibility, as opposed to being things subsisting in themselves (‘an sich subsistierende Dinge’, A491/B519), which are also referred to as ‘Sachen an sich selbst’ (ibid.). When Kant says in Q2(1) that he never doubted the existence of Sachen, this is just the counterpart to saying in A491/B519 that objects of experience are appearances, i.e. mere representations which in the way they are represented, i.e. as extended, have no existence in themselves outside of our thoughts. In other words, what is claimed to be ideal by transcendental idealism is the (spatio-temporal) mode of appearance of Sachen, not their existence, which points us back to the arguments offered in section I above. Kant's ideality claim does not affect the existence of Sachen, at best only the ‘existence’ of their modes of appearance.³² And this is just what is immediately claimed in Q2(2), which explains ‘transcendental’, and hence Kant's idealism, as applying not to the cognition of things (‘Dinge’), but to the faculty of cognition. We see that ‘Dinge’ is just a stylistic alternative to ‘Sachen’ here, and vice versa. Clearly, Kant takes issue with those who think his idealism makes claims, including the denial of existence, about things in themselves, and not just the conditions of experience (AA 04: 292.07-08). Hence, it is not true that Q2(1) presents a contrast between phenomenal Sachen and noumenal Sachen an sich. Kant means by ‘Sachen’ throughout ‘things in themselves’. Walker's interpretation is sound.

III Conclusion

Given the problems which the three most important recent interpretations of Prolegomena §13.II/III run into, and given the relatively small complication of the dual sense interpretation offered in section I, which resolves the perplexities of the text by assuming that a number of key concepts, especially ‘body’ and ‘outside’, in these sections have a dual use, sometimes in the same sentence, a dual use which is corroborated in other parts of the Prolegomena, and in the Critique, it can be concluded that the dual sense interpretation is the most acceptable reading of a prima facie perplexing passage and argument. Prolegomena §13.II/III is essentially directed against Noumenal-Berkeley, i.e. an idealist who denies the existence of things in themselves, of things which are not mere modifications of our mind, transcendentally speaking.