Preamble
I am grateful to the symposium participants for their thoughtful and stimulating comments. Regrettably, owing to space limitations, I have had to emphasize disagreement and clarification of position over collaborative exploration of issues.
1. Béatrice Longuenesse
Béatrice Longuenesse claims that in my interpretation of the B-edition first paralogism, Kant’s mention of the relation to intuition ‘disappears altogether’. I disagree. I did discuss this detail in The Fiery Test of Critique (Proops Reference Proops2021; hereafter FTC, 113), though it features less prominently in my account than two other ideas. These are first, the contrast between a claim about things (or objects), on the one hand, and a claim about representations, on the other, and second, the question of where the I-representation can and cannot go in a judgement. (By ‘the I-representation’, I mean the judgemental component expressed by the word ‘I’ when this word is taken to refer specifically to one’s thinking rather than one’s embodied self.)
In the first paralogism in B, the major premise talks about things (i.e. objects) in contrast to representations. It asserts of every object that if it cannot be conceived of as a substance1, then it cannot exist as a substance1. (By a ‘substance1’, I mean: ‘a necessarily non-inhering subject of inherence’.) Amongst the objects this premise speaks about is the self or ‘thinking I’ as it is given in intuition or, equivalently, as it appears to itself in inner sense.Footnote 1 That is where ‘the relation to intuition’ comes in. The minor premise, by contrast, talks not about things or objects but about representations. It says that the ‘I’ representation must occur in the subject position in a judgement. This has the consequence that it cannot occur as the sole occupant of the predicate position. The note to B411-12 formulates this diagnosis more clearly than the body of the text, for it says that the major premise speaks of ‘things’ while the minor speaks of ‘thoughts’.
For Kant, the key illusion that entices the pure rational psychologist to commit the fallacy involved in the B-edition first paralogism is the illusion that the self as it appears to itself in inner sense is a substance1. This illusion, Kant maintains, is generated partly by reason’s demand for the unconditioned and partly by our inclination to mistake facts about thoughts for facts about things. We are inclined to confuse the known fact that the ‘I’ representation cannot occur as the sole occupant of the predicate position with the unwarranted thought that the self cannot inhere in some other thing (or things). In other words, we are inclined to confuse one impossibility with another. This is why I focus on the impossibility of the ‘I’ representation’s occurring as the sole occupant of the predicate position rather than on the necessity of its occurring in the subject position – though these facts are, I think, two sides of the same coin.
A second point of disagreement relates to the question: Is the change from the A-edition treatment of the paralogisms to the B-edition treatment substantive? I answer: ‘yes’, Longuenesse: ‘no’. She takes Kant at his word when he says:
[In] the propositions themselves and their proofs, and also in the form and completeness of the plan, I have found nothing to alter. (Bxxxvii)
I disagree – both with Longuenesse and with Kant. For I see Kant’s changes as too obviously substantive to be plausibly characterized as concerned merely with ‘exposition’ (Bxxxviii). The conclusion of the fourth paralogism, for example, changes from an epistemic claim in A to a metaphysical claim in B. To suppose, as I do, that Kant was lying when he denied making substantive changes is both liberatory and charitable. It is liberatory because it enables us to trust our impression of Kant’s inconsistency across the two editions. It is charitable because it enables us to avoid adjudging him oblivious to the substantive nature of his changes. What is it that motivates Kant’s moment of dishonesty? Roy Sorensen and I have argued that the culprit is his organicist conception of his system. (For details, and for a sustained rebuttal of the ‘no substantive changes’ claim, see Proops and Sorensen Reference Proops and Sorensen2023.)
Longuenesse’s third main criticism relates to Kant’s baffling description of a paralogism at A402-3. Here Kant seems to present the paralogisms as involving a double equivocation: equivocation on the middle term – yes – but also on the predicate, thus:
Major: All M T are P T
Minor: The I, as thinking being, is M E
Therefore, The I, as thinking being, is P E
(T = ‘transcendental use’; E = ‘empirical use’.)
This is what I claimed. But, as Longuenesse astutely notes, it is mistaken. After all, it is the same category that is said to be used both in the minor premise and in the conclusion. However, although this objection is well taken, a problem remains. For the kind of ‘equivocation’ Longuenesse posits – ‘same concept, different use’ – is foreign to Kant. When Kant explains equivocation, he always does so in the expected way: a single word has multiple meanings (e.g. A373; Log-Blomberg, 24: 294).
What, then, is going on? The following is what I now think. We are very easily primed by Kant’s official description of a paralogism at A402-3 to expect an explanation of the equivocation he has just alluded to by labelling the paralogism a ‘sophisma figurae dictionis’. But this priming is in fact unintended (and misleading). Kant’s contrast between ‘transcendental’ and ‘empirical’ uses of a category relates not to the equivocation involved in a paralogism but to a difference in logical form. A concept is used transcendentally when it figures as the predicate in a universal judgement and empirically when it figures as the predicate in a singular judgement.
As odd as it may seem, then, Kant is not describing an equivocation when he draws the contrast between empirical and transcendental uses. He is rather claiming that the paralogisms – and presumably he means the first three non-anomalous ones – are arguments that, though they are in fact invalidated by equivocation, all purport to have the following form:
All M are P
a is M
So,
a is P
(Where ‘a’ is a singular term, e.g. ‘I’, ‘the soul’.)
In logic lectures delivered around 1780, Kant draws precisely the distinction I have in mind – albeit using different terminology. ‘Every use of a concept’, he says,
is either in abstracto or [in] concreto. I cognize a concept in concreto when I apply it to that which is contained under it. I think a concept in abstracto when I think it in general. (Log-Wiener, 24: 908)
In the minor premise of a paralogism and in the conclusion, the category is used in concreto because that premise purports to apply the category to a particular which falls under it. In the major premise, by contrast, it is used in abstracto because no such purported application is made.
Insofar as each paralogism is known to have true premises, it can count as a sophisma figurae dictionis because it features equivocation somewhere – and possibly just on the predicate. This happens in the A-edition first paralogism, where the conclusion gets inflated after having been drawn. The inflation amounts to an equivocation on the predicate of the pre-inflated conclusion. This illustrates a point that is easy to miss, namely, that Kant has a broad use of the term ‘sophisma figurae dictionis’ as meaning a fallacy involving equivocation, but not necessarily on the middle term (see, e.g. Corr, 11: 45).
2. Patricia Kitcher
Patricia Kitcher’s discussion contains much with which I agree, but also much with which I would take issue. In the latter category are, first, her claim that one of my interpretative proposals lacks textual support and, second, several straightforward mischaracterizations of my views – mischaracterizations that sometimes make those views look rather silly.
2.1 What is the confusion that Kant exposes in the Paralogisms chapter?
Kitcher takes me to hold that the relevant confusion is linguistic in nature: a confusion about the word ‘I’. This is not my view. I maintain, rather, that the confusion concerns a judgemental component, namely, the non-linguistic representation expressed by the word ‘I’. In my view, Kant is claiming that in the first paralogism, the paralogist confuses the known fact that this representation cannot occur as the sole occupant of a judgement’s predicate position with the merely imagined fact that what it represents cannot be a modification of another thing. Kant is not doing philosophy of language. He is rather drawing attention to a subtle confusion about thought, a confusion he regards as an instance of the mind’s broader tendency to confuse facts about mental representations with facts about things.
I did sometimes write, for the sake of brevity, as if I were talking about language. But I did so with frequent reminders of the more correct, if long-winded, way of speaking. I spoke, for example, of ‘The representation “I” (or “me”) – or, strictly, the judgmental component expressed by these words’ (FTC, 109, emphasis added). Having thus indicated that I would sometimes be speaking about words when what I strictly meant were judgemental components, and having repeated this warning in another context, I added: ‘Hereafter, I will suppress this qualification’ (ibid.). I meant that the qualification would remain silently in place. I did not mean that it would be revoked.
2.2 Does my interpretation of the B-paralogism lack textual support?
Kitcher claims that I provided no evidence to support my account of the confusion in the B-edition first paralogism. She speaks of a ‘lack of textual support’. I find this criticism puzzling. I thought I had been clear that the evidence in question consisted in Kant’s otherwise baffling talk in the note to B411 of my inability to use myself otherwise than as a subject (FTC, 107–8). This language, I argued, can only be understood if we accept my reading. Kant’s point – more clearly expressed in the A-edition – is that the representation expressed by ‘I’ is incapable of being used in any role other than that of subject-term in a judgement. The rival reading, according to which Kant is saying that in thinking my existence, I must be the subject of my judgement, cannot so happily account for Kant’s employment of the word ‘use’. How, after all, do I ‘use’ myself simply by being the subject of my thoughts?
Controversially, I maintain that the most plausible reason for the impossibility to which I draw attention is that the representation in question is singular. But the crux of my reading – and so also the nub of my disagreement with Kitcher – is independent of this claim. At its core, my reading requires only (a) that Kant should have indeed recognized that the representation expressed by ‘I’ was incapable of occurring as the exclusive occupant of the predicate position in a judgement. And (b) that this fact is plausibly confusable with the idea that I cannot inhere in another thing as a worldly ‘predicate’ (i.e. as a ‘modification’ or ‘mode’ of that thing).
I freely admit that my reading is not definitive. But it is not, I would insist, without textual support. That said, Kitcher’s discussion does bring out a useful point. It suggests that the heart of our disagreement lies in our different ways of resolving Kant’s ambiguous talk of the necessity of using myself as a ‘subject’ in thinking my existence in the footnote to B411. While Kitcher takes ‘subject’ to mean ‘thinker’, I take it to mean ‘representation that functions as the non-linguistic correlate of the “subject-term” in a judgement’.
2.3 Do I believe that Kant is sympathetic to the paralogist’s project of pure rational psychology?
Referring to what I say at FTC p. 80 about Kant’s prefatory defence of pure rational psychology, Kitcher portrays me as regarding Kant as somewhat sympathetic to pure rational psychology. This is not my view.
Granted, I do think that Kant has sympathy with rational psychology sans phrase. After all, I argued that he takes the empirical part of this discipline (i.e. impure rational psychology) to contain a nugget of silver in the form of the doctrinal belief in an afterlife. But the discussion to which Kitcher alludes (FTC, 80–2) relates only to the project of pure rational psychology, a project whose viability I take Kant unequivocally to deny.
Kant’s discussion of putting oneself in the place of another is part of his attempt to ensure that his critique of pure rational psychology is properly charitable. Not wanting to attack a straw man, he endeavours to first cast pure rational psychology into its most defensible form. But Kant engages in this exercise solely for the purpose of preparing a worthy target of critique.
His procedure here illustrates the aspect of his methodology I had termed ‘Kantian charity’, a summary of which runs as follows:
The germ of the contestations [of the dogmatic metaphysicians] which lies in the nature of human reason, must be extirpated; but how can we extirpate it if we do not give it freedom, indeed even nourishment, to send out shoots, so that we can discover it and afterwards uproot it? Thus, think up for yourself the objections [to your own position] which have not yet occurred to any opponent, and even lend him the weapons or concede him the most favorable position he could desire. (A777/B805–6, emphasis added, quoted in FTC, 24)
So, to be clear: I do not believe that Kant has any sympathy with the project of pure rational psychology. What he does have sympathy with is one component of impure rational psychology, namely, David Fordyce’s (1711–51) argument for an afterlife, an argument that includes empirical propositions amongst its premises. As I noted in the introduction (FTC, 5), on two occasions, Kant locates this argument squarely within rational psychology (Met-Volckmann, 28: 441–2; Met-L2/Pölitz, 28: 591–2). Consequently, in Kant’s view, rational psychology sans phrase yields a valuable residue when subjected to the fiery test of critique.
2.4 Are the paralogistic inferences inevitable?
Kitcher further mischaracterizes my position when she portrays me as holding that ‘Paralogism about the soul is inevitable’. On the contrary, I maintain that while Kant sees transcendental illusion as inevitable for human beings, he sees transcendental paralogism, along with any other species of error customarily arising from transcendental illusion, as in principle preventable. As I interpret him, Kant’s view is that, while transcendental illusion is necessary for humans insofar as they inevitably feel the pull of it, it is not necessary in the sense that they must inevitably be taken in by it (FTC, 41). Indeed, much of the point of the Paralogisms chapter is to show us how to avoid being taken in by one species of this illusion.
3. Allen Wood
Allen Wood and I agree on many matters. I agree with Wood about the Dialectic’s centrality. But I would go further. For several reasons, I take one part of the Dialectic to have preeminent importance for Kant, namely, the antinomies.
First, the antinomies occupy a unique place for Kant amongst the inferences that transcendental illusion invites us to draw in functioning as the smelling salts that provoke reason’s critical self-examination (Prol, 4: 341n; Corr, 12: 257–8). The other dialectical inferences (in the Paralogisms and Ideal), being the result of a merely ‘one-sided’ illusion, create no crisis. Second, Kant says that had it not been for his wish to respect ‘school’s rights’, he would have opened the Critique with the colourful antinomies (Corr, 10: 269–70). By ‘respecting school’s rights’, he means engaging in a systematic exposition, one guided by the topic-ordering of an eighteenth-century logic text: namely, concepts, judgements, inferences, and method.Footnote 2 Since the Transcendental Dialectic concerns inferences, school’s rights will be respected only if the antinomies are located far into the book. Third, Kant takes the antinomies to constitute the best method for the critique of metaphysics (R 4454 (17: 557), Prol, 4: 338). Fourth, they enact Kant’s favourite method in philosophy, the so-called ‘sceptical method’ (see FTC, 15–24). Fifth, their resolution is a precondition of reform in metaphysics (A497/B525). Sixth, insofar as they provide the materials for Kant’s indirect proof of transcendental idealism (though not, of course, the proof itself), the antinomies, in Kant’s view, constitute the most valuable residue of traditional speculative metaphysics (i.e. the gold).
This last point calls for clarification. Some readers (but none of the critics here) have taken me to hold that Kant’s indirect proof of transcendental idealism is sound and known to be sound. This, however, is not my view. I hold rather that the indirect proof fails of knowable soundness because for no antinomy are the arguments for the thesis and antithesis positions both evidently sound on the assumption of transcendental realism. The gold that Kant thinks he finds in the crucible turns out to be illusory. Regrettably, I did not spell this point out until the ‘Closing Reflections’ (‘Kant regarded etc’, FTC, 453, emphasis added). Instead, I relied on the reader to glean it from my identification of numerous flaws in the reasoning for thesis and antithesis (e.g. FTC, 222, 233, 235, 243). So, to spell it out now: my view is that Kant believes that the fiery test of critique yields both gold (consisting of the materials for the indirect proof of transcendental idealism, i.e. the arguments for the thesis and antithesis of each antinomy) and silver (consisting in two ‘doctrinal beliefs’), though I myself believe it yields neither gold nor silver.
If I did not make this point sufficiently vividly, it was because when drawing attention to the importance of the ‘fiery test’ as one of Kant’s guiding metaphors, I was more focused on his ambitions than his accomplishments. In particular, I was concerned to challenge Moses Mendelssohn’s now-famous characterization of Kant as the ‘all-crushing’ destroyer of traditional metaphysics – a characterization that has caught on despite Mendelssohn’s having coined it while simultaneously admitting that he had not actually read the first Critique (Mendelssohn Reference Mendelssohn, Dahlstrom and Dyck2011: xix). I maintain, on the contrary, that Kant sought to salvage as well as destroy.
To resume the thread: the first point on which I agree with Wood is on the centrality of the Dialectic – though I go further in discerning a core within the core.
Secondly, I agree with his cautionary note to the effect that labels and ‘-isms’ can lead to unhelpful pigeonholing. In the present case, however, I see this worry as overblown. So long as the labels are suitably qualified, they need not induce confusion. We can, of course, abuse them, for we can err in our classifications. And plausibly, it is an especially egregious mistake (indeed a prejudice) to presuppose that Kant’s view must fit neatly into one or other pre-identified camp. But we can reject the pigeonholing without rejecting the labels, and we can use some of them, suitably qualified, to communicate Kant’s views. Wood mentions that Kant never uses the labels himself. But why should that matter? Whether Kant subscribes to an ‘-ism’ depends on his views, not on the labels he himself attaches to them.
Third, in the book I tried to agree with Wood’s claim in his justly famous article that we can make sense of the seemingly paradoxical idea that Kant combines compatibilism with incompatibilism. Unfortunately, however, this time, the agreement did not ‘take’. Wood observes that he had only been tempted to use this characterization. At the end of the day, he wanted to resist the temptation. I admit I had missed the point of Wood’s talk of ‘temptation’. I had imagined its purpose was to register the paradoxical feel of his apparently contradictory formulation of ‘the compatibility of compatibilism and incompatibilism’. I had wrongly supposed that he was advising us to resist the temptation to speak this way simply because, taken literally, it was an overstatement – and, indeed, an incoherent overstatement. The paradoxical language nonetheless charmed me – as it has charmed many others – and I tried to make something of it (see FTC, 322).
Wood and I have a general disagreement about the value of labels in exegesis, but there is, in addition, a particular disagreement concerning the label ‘libertarian’. Wood claims that I deny that Kant is a libertarian. But that unqualified claim does not express my view. My position is rather that Kant is not a libertarian in the sense embraced by several contemporary theorists of free will, theorists who take libertarianism to entail a rejection of determinism (FTC, 317–8). Since I interpret Kant as a determinist (albeit one of a peculiar stripe), I decline to classify him as a libertarian in this sense, though I would readily admit there are other senses in which he is plainly a libertarian (e.g. he believes in free will).
4. Desmond Hogan
4.1 A uniform resolution of all four antinomies?
Does the passage at A506-7/B534-5, which Desmond Hogan labels ‘Proof structure’, contain a general description of Kant’s indirect argument for transcendental idealism? Hogan answers ‘yes’. I answer ‘no’. I maintain that since this passage equates the thesis and antithesis positions with the finitary and infinitary positions, it relates only to the mathematical antinomies. Hogan seeks to defend his answer by marrying it to what he controversially takes to be Kant’s claim of a uniform strategy of resolution for all four antinomies. To support this latter interpretation, he cites what I will call ‘the disputed passage’:
The very same [resolution] holds of the series of causes ordered one above another, or of conditioned existence up to necessary existence, which can never be regarded in themselves as either finite or infinite in their totality, because, as series of subordinated representations, they exist only in the dynamical regress … (A505-6/B533-4)
I do not see this passage as evidence for a uniform resolution, for I regard Hogan’s interpolation of the word ‘resolution’ as unwarranted. The uniformity Kant affirms here is genuine enough, but it does not concern Kant’s strategy for resolving the antinomies. The context makes clear that Kant is claiming that all cosmological series – the dynamical no less than the mathematical – exist, as he puts it, ‘only in the empirical regress’. They are not ‘absolute’ but only ‘relative’ totalities, with the consequence, for Kant, that their magnitude is neither finite nor infinite.
There is a uniformity here, then, but Kant cannot exploit it in his resolution of the dynamical antinomies, for the exploitation is blocked by a crucial asymmetry. While the claims that the world construed as a mathematical series is finite (infinite) are plausible abbreviations of the thesis (antithesis) of the mathematical antinomies, the corresponding claims for the dynamical series are not plausible abbreviations for the thesis and anthesis of the dynamical antinomies. To appreciate this point, it suffices to note that the thesis of the fourth antinomy could be true whether the dynamical series were finite or not finite, for it would be true provided that a necessary being existed outside the empirical world as its cause (A453/B481).
There is, in any case, abundant evidence that Kant used sharply contrasting strategies for resolving the mathematical and dynamical antinomies. See, for example, A529-30/B557-8, A531-2/B559-60, Prol, 4: 343, ‘Progress in Metaphysics’, 20: 291, R 5962 (18: 404), and R 6421 (18: 711). Taken together, these passages make clear that Kant treats the opposed propositions in the mathematical antinomies as apparent contradictories, which transcendental idealism nonetheless reveals to be mere contraries, but those in the dynamical antinomies as apparent contradictories, which transcendental idealism nonetheless reveals to be mere sub-contraries (FTC, 210–11). At a sufficiently high level of abstraction, of course, Kant’s strategy for resolution is uniform: in each of the four antinomies, an apparent contradiction is revealed to be merely apparent. But at the lower level on which Hogan is focused, it is non-uniform.
4.2 Does the antinomy of freedom concern specifically human freedom?
In a note, Hogan implies that it is only in certain writings external to the first Critique – namely, in the correspondence and Prolegomena – that Kant sets things up so that there is an antinomy of human freedom. This, however, is false.
It is true that the argument for the thesis in the antinomy formally presented concerns itself only with an abstractly conceived cosmological freedom. But that does not mean that Kant fails to recognize an antinomy of specifically human freedom in the first Critique. For example, when summarizing what would happen if the oscillation between thesis and antithesis were not stabilized by the interests of reason, he envisages us as facing an antinomy concerning the freedom of the ‘human will’ (A475/B503).
I conclude that if Kant does not omit mention of human freedom from his official presentation of the antinomy merely owing to inaccuracy of formulation, then he must be interpreted as recognizing (if not rehearsing) two distinct antinomies of freedom in the first Critique – one dealing with cosmological freedom in the abstract and another with the concrete freedom of the human will. This cannot be ruled out on architectonic grounds since the elegant idea of there being exactly four antinomies had in any case already been abandoned in the first antinomy, where Kant recognizes two distinct antinomies embedded within the official ‘first’ antinomy: one concerning the world’s age, the other its size. Since we must in any case recognize at least five antinomies, why not six?
5. Anja Jauernig
5.1 Can GUISE be defended?
GUISE is Anja Jauernig’s label for my thesis that Kant introduces the ontological argument (hereafter ‘OA’) under the guise of an argument allegedly establishing a counterexample (viz. the concept of the most real being) to a thesis that Kant wishes to defend. This is the thesis that ‘S is annulled’ is never self-contradictory (or, equivalently, that ‘S is posited’ is never analytic), where ‘S’ is a term for a concept. I will call this ‘the syntheticity thesis’ (‘ST’ for short). My evidence for GUISE is that the part of Kant’s criticism of the OA that carries suasive force occurs in a discussion introduced by the following exchange: ‘You argue that there is indeed one concept … where the non-existence or annulment of the concept’s object is self-contradictory … I answer (Ich antworte) …’ (A596/B624). This is the language of disputation. When one ‘answers’ an objection, one assumes the role of ‘defender’ of the thesis in question. Kant’s opponent (the attacker) challenges the thesis – that is, ST – by adducing the concept of the most real being as an alleged counterexample, relying on the OA to substantiate this allegation.
These points relate to Jauernig’s question about where in the Dialectic Kant’s critique of the OA begins. To get clear about this matter, we need to distinguish between Kant’s claims, on the one hand, and his arguments for those claims, on the other. Kant’s opponent in effect claims that ‘The most real being exists’ is analytic, and Kant denies this on the (thus far undefended) ground that claims of the form ‘S is annulled’ are never self-contradictory. Part of what he is doing in the preliminary material at A594-5/B622-3, I would claim, is establishing that this is the dialectical situation. But he is saving his rehearsal of the argument that constitutes his defence of ST for later. The other thing he is doing in this preliminary material is describing how transcendental illusion entices us to perform the inference involved in the OA, namely, by encouraging us to mistake a conditional (or ‘logical’) necessity pertaining to judgements for an absolute necessity pertaining to things. Witness: ‘this merely logical necessity has proved to have great power of illusion’, etc. (A595/B623).
But – and this is the crucial point – to deploy ST against the ontological argument is not yet to offer a suasive critique of that argument, for, as Kant well knows, his opponent is hardly likely to accept ST merely on his say so. Instead, as Kant himself notes, his opponent will attempt to counterexample ST with the concept of the most real being (A596/B624).
We reach a standoff. And Kant knows that to break it, he must compel his opponent to recognize that he (the opponent) is in fact in no position to advance his alleged counterexample since he is in no position to recognize its supporting argument, the OA, as sound. Kant supplies the needed argument to this effect in the material occurring after ‘I answer’ at A597-600/B625-8. The most familiar part of this material is the 100-thalers argument, an argument that Kant regards as possessing suasive force against his Leibnizian opponent.
5.2 Does the ‘100-thalers’ passage really contain an argument?
Jauernig maintains that the 100-thalers passage contains not an argument, but merely an ‘illustration’ of the thesis that existence is not a real predicate – albeit one amounting (somehow) to a ‘plausibility consideration’.
I remain unpersuaded – not least because the relevant passage contains several words that are plainly expressive of argumentation (the italicized ones):
And thus the actual contains no more than the merely possible. A hundred actual thalers do not contain the least more than a hundred possible thalers. For [Denn], since [da] the possible thalers signify the concept and the actual thalers signify the object and the positing thereof in itself, if the object contained more than the concept, then my concept would not express the entire object and thus would also not be the concept commensurate with this object. (A599/B627, emphases added)
The argument conducted here is an enthymematic reductio of the supposition that existence is a real predicate.
5.3 A bad argument?
Jauernig portrays me as suggesting that Kant intentionally offers a bad argument against his Leibnizian opponent. This is not my view. Rather, I claim that he sees himself as offering an argument that is sound in its disputational context – namely, as directed against Leibniz (and, more generally, against anyone sharing Leibniz’s stance on the Euthyphro contrast). This is because it is an ad hominem argument in the traditional sense of that term, a sense importantly distinct from that of the more familiar ad hominem fallacy (the fallacy of attacking the person rather than their reasoning). One gives an ad hominem argument when one relies on a controversial premise that one’s opponent is nonetheless committed to accepting. Kant regards this kind of ad hominem argument as legitimate:
A justification [Rechtfertigung] kat’ anthropon [i.e., ad hominem] … secures one’s claims against any encroachment and provides one with a titled possession that need not fear any pretensions from others … (A740/B768)
One can refute the opponent by an argument kat’ anthropon in that I accept my opponent’s opinion and from it infer something contrary to him. (Log-Hechsel, p. 118)Footnote 3
Kant once mentions the possibility of using this style of argument (but not the specific argument we are currently discussing) to refute ‘the Leibnizians’:
Maimon’s way of representing is Spinozism and could be used most excellently to refute the Leibnizians ex concessis. (Corr, 11: 50)
‘Argumentum ex concessis’ is another term for ‘argumentum ad hominem’.
As Kant makes clear, the rules of disputation dictate that one is entitled to rely on an ad hominem argument only so long as one is playing the role of a defender (‘Orientation’, 8: 134). And Kant, since he is defending ST against a putative counterexample, is doing just that (see FTC, section 14.10).
Jauernig makes a fair point when she questions my having presented Kant’s objection that existence is not a real predicate as directed only against Leibniz. That, I concede, was an error of formulation. What I had meant to say is that Kant’s objection is directed against Descartes as much as Leibniz but supported by the 100-thalers argument only insofar as it is directed against Leibniz – or, better, the Leibnizians. In consequence, I do not subscribe to Jauernig’s MAIN.
Where does Kant’s discussion leave us? One thing it seems fair to say is that Leibniz did indeed overlook the fact that he could not run the OA while also adhering to his position on the Euthyphro contrast (viz. that God actualizes the best possible world because it is best; it is not the best because God actualizes it), so Kant makes a valuable contribution in pointing this out, even if he does not do so as explicitly as one might have wished.
If, on the other hand, one rejects Leibniz’s stance on the Euthyphro contrast and embraces Cartesian voluntarism, one can quickly argue – as Leibniz does in the Theodicy (section 176) – that the creator is not actually praiseworthy or benevolent, with the consequence that the divine attributes, which include creator, are not compossible.
If this is right, then one can in effect argue against both of Kant’s main opponents – Descartes and Leibniz – by means of an argument by dilemma. And since that is so, the fact that Kant is offering his ad hominem argument only against Leibniz (and friends) does not diminish the interest of his point: there is no escape for Descartes – or, at least, none given the traditional assumption that exactly one direction of the Euthyphro contrast must be correct. Either he persists with his voluntarism and so must treat the divine attributes as incompatible, or he abandons it and so must accept the 100-thalers argument. Either way, the Cartesian OA fails.
5.4 Was my reconstruction too complicated?
Jauernig complains that my reconstruction of Kant’s presentation of the ontological argument is too complicated. It is true that in endeavouring to portray Kant as charitably presenting his opponent’s argument, I did supply several suppressed premises. Jauernig’s rival reconstruction achieves greater simplicity, but only at the cost of introducing an invalidity not recognized by Kant: the inference from 1 and 3 to 4a in Jauernig’s reconstruction is not logically valid. Compare the evident invalidity of the following argument:
The richest person has all the yachts.
Morning Cloud is a yacht.
Therefore,
The concept of the richest person contains Morning Cloud.
5.5 What do I mean by a ‘version’ of the ontological argument?
I maintain that Kant focuses on Leibniz’s version of the OA. I do so for several reasons. First, this assumption yields a charitable interpretation of the 100-thalers passage. Second, Kant explicitly recognizes Leibniz as a proponent of the OA (‘Real Progress’, 20: 349; Met-K2, 28: 782) – and, indeed, as having employed it in the very way that Kant envisages it as being employed by the dogmatic metaphysician, namely, as completing the cosmological argument (Met-L2/Pölitz, 28: 599). Third, the argument Kant criticizes resembles Leibniz’s in incorporating the needed logical possibility assumption. Fourth, Kant explicitly describes Leibniz as having wanted to improve on Descartes’s version by composing the argument out of two steps – the second step being the needed logical possibility proof (28: 782).
I take the fact that Leibniz adds a premise to Descartes’s argument – together with an argument for that premise – to be sufficient to constitute it a distinct version of the OA. The grain of truth in Jauernig’s ‘versions’ worry is that I did not distinguish clearly enough between (a) the idea that a token of the OA counts as ‘Leibnizian’ because it incorporates an explicit logical possibility premise (plus supporting argument) and (b) the idea that it counts as ‘Leibnizian’ because it is envisaged as being rehearsed against the background of certain Leibnizian commitments. When I claim that Kant’s criticism works against Leibniz’s version of the argument, I am focusing on (b).
Acknowledgements
My thanks to Katherine Dunlop, Martin Lin, and Jon Litland.
