2. Initial worries about MAIN

MAIN directly gives rise to several worries. My first two worries concern Proops’s formulation of MAIN as the claim that Kant’s ENRP objection is intended to have force only against ‘Leibniz’s version’ of the Cartesian ontological argument.

First, Proops does not tell us what exactly he has in mind when he talks about different ‘versions’ of an argument, but it seems inappropriate to me to characterise Leibniz’s argument as a ‘version’ of OA that differs from Descartes’ own ‘version’. Leibniz repeatedly states explicitly that he likes OA as presented by Descartes just fine and regards it as correct as far as it goes.Footnote ² His only complaint is that Descartes failed to provide a defence of the implicit premise that God is possible, a defence that he (Leibniz) proceeds to supply. To my mind, calling this enhanced argument a different ‘version’ is a stretch.

Second and more importantly, it seems fair to say that an objection O is directed at version A of an argument but not at version B only if O is directed solely at those parts of version A with respect to which it differs from version B; otherwise, O would be directed at both. Applied to the case at hand, the ENRP objection would be directed at Leibniz’s ‘version’ of OA and not at Descartes’ only if it were directed solely at Leibniz’s defence of the premise that God is possible. For that is the only bit with respect to which the two ‘versions’ differ. But Leibniz’s proof of God’s possibility is clearly not the target of the ENRP objection. Kant rejects this proof in the penultimate paragraph of the OA section, as a kind of afterthought. But, by that point, the ENRP objection has already been done and dusted. The ENRP objection targets the part of OA with respect to which Leibniz and Descartes agree, namely, the premise that existence is a perfection and thus a reality.

So, regardless of how ‘version’ is understood exactly, it is simply false to say that Kant’s ENRP objection targets only Leibniz’s version of OA. At most, what could be said is that the ENRP objection is directed only at Leibniz and people like him who are committed to all of the claims on which the objection and Kant’s support for it depend. This is how I will understand MAIN in the following.

Another worry about MAIN is that it is prima facie implausible. Kant presents the ENRP objection as one of his main objections to OA in the OA section, which bears the bold title ‘On the Impossibility of an Ontological Proof of the Existence of God’. If it turned out that this central objection targets only a particular group of proponents of OA who share certain additional commitments, one could justly complain that Kant’s presentation is highly misleading and he is greatly overselling his case.

A related further worry about MAIN is that there is no direct textual evidence to support it. Indeed, as Proops himself points out, when Kant associates a specific philosopher with OA, it is almost always Descartes who is mentioned. (He also mentions Anselm a few times.) This strongly suggests that Kant’s objections are intended to be effective against Descartes and anybody else who endorses (a ‘version’ of) the Cartesian proof. If Kant intended his ENRP objection to be effective only against Leibniz, he could rightly be blamed for putting his readers on the wrong track by failing to explicitly tell them about this sudden and unusual restriction of his target.

In reply to a similar kind of worry, Proops notes that (a) ‘Kant sometimes uses the phrase “the Cartesian proof” as an umbrella term for any version of the modern, post-Anselmian ontological argument’ and ‘speaks of philosophers other than Descartes as running “the Cartesian proof”’ and that (b) ‘the very context in which the label “the ontological (Cartesian) proof” occurs suggests that Kant’s immediate target in that part of the Ideal is Leibniz’ (p. 341).

This reply strikes me as ineffective. Regarding (a), an author who uses the phrase ‘the Cartesian proof’ in the indicated way and speaks of philosophers other than Descartes as running it should be especially eager, in order to prevent misunderstandings, to explicitly alert his readers if he is presenting an objection that is meant to attack only one of these philosophers. Regarding (b), this claim is highly implausible. As already noted, Kant criticises Leibniz’s attempted proof of the possibility of God in the penultimate paragraph of the OA section. But the label ‘the ontological (Cartesian) proof’ occurs in the final paragraph, indeed, the final sentence, of the OA section, where Kant concludes that ‘the famous ontological (Cartesian) proof of the existence of a highest being from concepts is only so much trouble and labour lost …’ (B620/A602). Since this is the very last sentence of the section, it is eminently plausible to think that Kant uses it to summarise the conclusion of the section as a whole and not merely of the preceding paragraph that deals with Leibniz’s possibility proof. The section as a whole is concerned with the Cartesian ontological proof and has the result that this proof is a dismal failure.

3. GUISE is false

Recall GUISE: Kant introduces OA under the guise of a putative objection to his own thesis that any concept can be consistently supposed to lack exemplification. According to Proops, this introduction of OA happens at B624/A596 in a passage that, following him, I will call the ‘expository passage’, to be quoted below. As I see it, the expository passage is not where Kant introduces OA in the OA section. Indeed, it would be surprising if it were, given that the passage constitutes the seventh paragraph of the section, which occurs about halfway through the discussion. On my reading, Kant introduces OA in the fourth paragraph. Furthermore, he not only introduces OA in the paragraphs before the expository passage but also launches a first objection to it. The expository passage provides a restatement of the argument under the guise of a response to this objection.

Several factors obscure this. Kant evidently (and reasonably) assumes that his readers are familiar with OA and introduces it by providing only a brief reminder of its key moves. Moreover, he packs this introduction into a long, complicated (and not entirely grammatical) sentence, which also includes his diagnosis of the source of the argument’s deceptive power – which consists in a confusion of two kinds of necessity, the logical necessity of judgements and the absolute necessity of things, discussed in the third paragraph – as well as the beginning of his attack on the argument.

With some goodwill, one can discern the following presentation of OA in the italicised sentence, especially if one takes into account how Kant continues in the following paragraph and what his first objection turns out to be:

1. 1. The concept ‘God’ is internally consistent and contains the concept ‘existence’, just as the concept ‘triangle’ is internally consistent and contains the concept ‘triangularity’.

2. 2. The claim that God does not exist is contradictory, just as the claim that a triangle does not have three angles is contradictory. (From 1)

3. 3. God necessarily exists, just as a triangle necessarily has three angles. (From 2)

That Kant most likely has this argument in mind here is confirmed by the fact that it closely corresponds to Descartes’ presentation of OA in the early parts of the Fifth Meditation, down to the parallel with the triangle case.Footnote ³

Kant articulates his first objection to OA in the fifth paragraph of the OA section that follows right on the heels of the previously quoted passage:

Kant’s objection depends on a distinction between two kinds of claims, which I will call ‘existential’ and ‘essential’ claims. Existential claims express that an object that falls under the subject concept is ‘posited’, or exists in reality. Essential claims express that the predicate concept is contained in the subject concept, and thus entail that, if an object falls under the subject concept, it necessarily falls under the predicate concept as well. For example, the claim that a triangle exists expresses that an object that falls under the concept ‘triangle’ is posited; the claim that triangles have three angles expresses that the concept ‘triangle’ contains the concept ‘triangular’, and thus entails that any object that is a triangle necessarily has three angles. The objection turns on the thesis that no negative existential claim is contradictory. While denying an essential claim leads to a contradiction, denying an existential claim does not. The latter amounts to no more than ‘cancelling’ the object that falls under the subject concept, or expressing that there is no such object in reality. So, where should the contradiction come from?Footnote ⁴ Not even negative existential claims with internally contradictory subject concepts are contradictory. Since there is no possible object that falls under an internally contradictory concept, such negative existential claims are trivially true.

In light of these clarifications, Kant’s first objection to OA can be presented in the form of the following brief argument:

1. 1. No negative existential claim is contradictory.

2. 2. The claim that God does not exist is a negative existential claim.

3. 3. The claim that God does not exist is not contradictory. (From 1 and 2).

This argument shows that the thesis in line 2 of OA as formulated above is false. Hence, OA fails. Note that proponents of OA must admit that the claim that God does not exist is a negative existential claim. For, apart from being rather implausible, the suggestion that this claim is essential would commit them to holding that the claim that God exists is also essential. But this would amount to conceding explicitly that the latter claim expresses only that the concept ‘God’ contains the concept ‘existence’, and thus entails no more than the hypothetical claim that, if an object is God, then it exists, a concession that would be tantamount to throwing in the towel.

In the sixth paragraph of the OA section, right before the expository passage, Kant considers a possible rather lame reply to his first objection. The reply consists in the question-begging assertion that there are some very special beings that cannot possibly be cancelled. Since this reply is question-begging, it can quickly be set aside.

So, by the time the expository passage comes around, Kant has already introduced OA and started his criticism of it by way of launching his first objection. The expository passage does not introduce OA but restates it. More specifically, the passage provides a restatement of the argument under the guise of another reply to Kant’s first objection. As such, it fits right in with the lame reply just considered. Both are dialectical analogues of ways for proponents of OA to stamp their feet.

This reading conforms well with how Kant starts the expository passage. Note that Proops’s rendering of the opening sentence of this passage is not entirely faithful to Kant’s text. Here is how Proops puts it (on p. 339):

Here is what Kant actually says:

‘All of these general conclusions’ are the conclusions we just rehearsed, in particular, the conclusion that OA fails because it runs afoul of properly distinguishing between existential and essential claims. The imagined challenge of the proponents of OA is focused on the claim that there is no concept such that cancelling its object leads to a contradiction – a claim that is equivalent to the claim that no negative existential claim is contradictory, or, as Proops puts it, that any concept can be consistently supposed to lack instances – because, as we just learned, this is the main premise of the objection to OA that Kant develops in the early paragraphs of the OA section. By offering a restatement of their argument, proponents of OA hope to show that there is one special being after all of the kind mentioned in the previous lame reply, namely, of the ‘not cancellable’ kind.

So, GUISE is false. Kant does not introduce OA under the guise of a putative objection to one of his own theses. What Proops calls a putative objection to Kant in truth is a putative reply to Kant’s first objection to OA, as presented in the early paragraphs of the OA section. And since GUISE is false, THALERS also does not draw additional support from being able to explain the otherwise puzzling GUISE in light of the rules of disputational conduct. If there is no puzzle, no explanation is needed.

4. THALERS and MAIN are false

Kant’s ENRP objection targets OA as presented in the second part of the expository passage:

The heart of this argument is the already familiar point that, due to the containment of ‘existence’ in the concept ‘God’, the supposition of God’s non-existence leads to a contradiction. The present presentation of OA differs from the earlier one in that the contradiction is derived in a slightly different way and an explicit justification for the containment claim is supplied. Proops’s reconstruction of the argument (pp. 342–3) strikes me as much too complicated. In addition to the three premises explicitly stated by Kant (1–3), it features three suppressed premises and seven intermediary conclusions. For the record, I favour the following alternative reconstruction:

1. 1. The most real being has all reality.

2. 2. The most real being is possible.

3. 3. Existence is a reality.

4. 4a. The concept ‘the most real being’ contains existence. (From 1 and 3)

5. 4b. If the most real being is possible, then it exists. (From 4a)

6. 5. The most real being does not exist. (Reductio assumption)

7. 6. The most real being is not possible. (From 4b and 5)

8. 7. Contradiction. (From 2 and 6)

9. 8. The most real being exists. (From 7, discharging the reductio assumption in 5)

Apart from being much more straightforward and much closer to Kant’s text than Proops’s reconstruction, my reconstruction also has going for it that, in his discussion of OA in the New Essays, Leibniz comments on the importance of the intermediary conclusion 4b of my reconstruction for the argument, an intermediary conclusion that plays no role in Proops’s reconstruction.Footnote ⁵ Since it is fairly safe to say that Kant read the New Essays, and since I agree that Leibniz is one of Kant’s targets in the OA section, I regard this comment as a piece of evidence in favour of my reconstruction.

Despite our disagreement about how best to reconstruct OA as presented in the expository passage, Proops and I agree that Kant’s ENRP objection is directed at claim [3], the thesis that existence is a reality, which features as a premise in both of our reconstructions. The objection consists in the straightforward complaint that, since ‘existence’ is not a real predicate (ENRP), which is equivalent to the claim that existence is not a reality, claim [3] is simply false and the argument unsound.

The force of this objection depends on the plausibility of ENRP. Kant supports ENRP in the following passage, which I will call the ‘hundred thalers passage’:

On Proops’s reading, Kant offers an argument for ENRP in this passage, the hundred thalers argument. Proops develops a detailed reconstruction of this argument over the course of several pages (pp. 354–60), a reconstruction that features three premises and four intermediary conclusions. For our present concerns, there is no need to examine this reconstruction in its entirety. We will focus our attention on the main premise, which Proops calls ‘M (strong)’:

As Proops points out, M can be rejected. Kant’s opponent could simply insist that it is possible for there to be two deposits of a certain kind, where one is more real than the other, such that they have the same monetary magnitude and differ only in that one of them is actual while the other is merely possible. Denying this possibility would be nothing less than ‘simply to beg the question in favour of the view that actuality is not a reality’ (p. 360).

So, is Kant guilty of offering an embarrassing question-begging argument for ENRP? Proops submits that ‘there’s a more charitable interpretation’ (p. 360), namely, his interpretation that is captured by THALERS and MAIN. M (strong) is the premise of which THALERS says that Leibniz is committed to it, while Descartes is not. So, if we read Kant as having intended the ENRP objection to have force only against Leibniz, as stated in MAIN, the hundred thalers argument for ENRP does not constitute a liability or embarrassment for Kant since Leibniz is committed to its main premise.

Before articulating my misgivings about Proops’s interpretation, I would like to note that I agree with his claim that Leibniz seems to be committed to M (strong). On Leibniz’s view, God’s choice of which one of infinitely many possible worlds to create is based on which one is best or most perfect. The degree of reality of a world is part of what determines its perfection. So, Leibniz is committed to a certain position on (what Proops calls) the ‘Euthyphro contrast’, a position according to which ‘differences in the degree of reality between worlds must exist prior to God’s creative choice and must serve to ground that choice; they cannot be consequent upon it’ (p. 360). But this means that Leibniz must accept the claim that there are no possible objects such that one is more real than the other merely by dint of being actual while the other is merely possible, a claim that directly entails M.

Now, Proops’s reading that the hundred thalers passage contains a question-begging argument for ENRP, which, however, is not a liability or embarrassment for Kant since he intends the ENRP objection to work only against Leibniz who is committed to the argument’s question-begging premise, seems problematic to me for several reasons. First, that the person to whom a question-begging argument is addressed agrees with the question-begging premise does not make the argument any better. A question-begging argument is a bad argument, and Kant should be embarrassed about proposing it even if Leibniz were forced to accept it. (Also note that Leibniz could reject the argument qua argument, even if he is committed to its question-begging premise.) So, if we are looking for a charitable interpretation, THALERS and MAIN do not really fit the bill.

Second, there is no doubt in my mind that Kant endorses ENRP, the conclusion of the hundred thalers argument. But, assuming charitably that he is not confused, why would he offer a bad argument for a conclusion that he himself endorses? Would it not be much better to offer a good argument and address it to everyone – if, indeed, he thinks that an argument is needed? Surely, it would be bizarre to suppose that Kant intentionally offers a bad argument for a conclusion that he himself endorses because he intends the objection against OA that is based on this conclusion to be effective only against one particular person and his followers who alone are forced to accept the bad argument. The ENRP objection is clearly effective against OA. Why would Kant not want to use it against all proponents of OA? So, in light of the fact that Kant endorses ENRP, Proops’s reconstruction of the hundred thalers argument as well as MAIN appears highly implausible. Or does Proops mean to suggest that Kant does not endorse ENRP? This suggestion seems rather difficult to square with the text.

Third, another way of saying that Leibniz is committed to the question-begging premise M (strong) of the hundred thalers argument is to say that he is committed to ENRP. And, indeed, ENRP directly follows from the position on the Euthyphro contrast that Proops ascribes to Leibniz. The thesis that the reality of worlds and of the things in them is fixed prior to God’s creative choice directly entails that existence is not a reality. So, if there is good reason to think that Kant would have known of Leibniz’s position on the Euthyphro contrast, as Proops claims (p. 361), then there is also good reason to think that Kant would have known that Leibniz is committed to ENRP. But, in that case, there is a much more straightforward objection than the ENRP objection backed by the hundred thalers argument that Kant could have raised if he had wanted to target Leibniz in particular. He could have simply said that Leibniz cannot endorse OA as presented in the expository passage on pain of inconsistency. For, on account of his position on the Euthyphro contrast, Leibniz is committed to ENRP, but ENRP is the denial of the argument’s most important premise, claim [3]. That Kant does not raise this objection reinforces the conclusion that he does not want to target only Leibniz but all proponents of OA.

Fourth, on my reading, the considerations articulated in the hundred thalers passage are not intended to be understood as an argument for ENRP, strictly speaking. Rather, they are meant to illustrate it and make it plausible. Note that this reading coheres well with the finding that, understood as an argument for ENRP, the considerations in the hundred thalers passage turn out to be question-begging. That is just what one would expect to happen if one tries to read what is meant as an illustration and plausibility consideration as if it were meant as an argument. Kant’s plausibility consideration turns on the observation that, if existence were a reality, we would have to use different concepts to completely describe two individual objects that differ only in that one is actual while the other is merely possible. In the hundred thalers passage, Kant invites his readers to fully appreciate what this consequence means and agree with him that it is implausible. More specifically, I take it that Kant wants his readers to imagine the following kind of scenario. In preparation for creating a heap of a hundred thalers, God surveys various complete concepts of different possible individual hundred thaler heaps and eventually decides that he wants to actualise the possible individual heap that falls under concept C. (A complete concept of an individual completely describes it as the very individual it is down to the minutest detail.) God then goes ahead and creates this heap. Would it not be implausible to assume that C is not a complete concept with respect to the actual individual hundred thaler heap that God just created? After all, the possible heap and the actual heap are exactly alike. There is simply nothing left to say in order to characterise the actual heap as the very heap that it is that is not already captured in C. Of course, the indicated plausibility consideration will probably not sway all of Kant’s readers. But there is no reason to assume that he does not address it to all of them in the hope of convincing them of the plausibility of ENRP.

Also note that the reason why, of all examples that he could have chosen, Kant settled on a hundred thalers to illustrate ENRP in the indicated way is not that thalers or deposits are especially suited to the point he wanted to make, as Proops’s reconstruction of the hundred thalers argument suggests. As I read him, Kant took advantage of the opportunity to show his witty side. There is nothing witty about the locution that, say, an actual tree contains not the least bit more than a possible tree that is exactly like the actual one. But it is at least somewhat witty to say that a hundred actual thalers contain not the least bit more than a hundred possible thalers that are exactly like the actual ones. For, in some sense, a hundred actual thalers are, of course, much more than a hundred possible thalers, namely, with respect to one’s ‘financial situation’, as Kant himself explicitly notes in the very next sentence.

Fifth, the ENRP objection is continuous with Kant’s first objection discussed earlier, an objection that is repeated in slightly different terms in the eighth paragraph of the OA section, just before the ENRP objection is introduced. Kant there points out that if proponents of OA understand the claim that God exists as analytic (or essential in the earlier terminology), they would have to admit that it is nothing but a ‘miserable tautology’, whereas if they understand it as an existential claim, they would have to admit that it is synthetic and thus can be denied without contradiction. That ‘existence’ is not a real predicate accounts for the distinction in kind between existential and essential claims, which underwrites Kant’s first objection. Since the first objection is directed against all proponents of OA, it thus stands to reason that this holds for the ENRP objection as well.

On the basis of these considerations and the findings from section two that MAIN is prima facie implausible and lacks direct textual support, I thus conclude that Kant neither argues for ENRP in the way suggested by Proops nor intends his ENRP objection to be effective only against Leibniz and his followers. That is, I conclude that THALERS and MAIN are both false.