2. Monistic approaches

Kant’s proof of the existence of God attempts to establish that only God could be the ground of all possibility. There are reasons to believe that this way of arguing for the existence of God is one of Kant’s historical innovations (Adams Reference Adams2000). But the view itself, the claim that all possibilities are grounded in God, is not new to Kant. It has enjoyed some popularity throughout the history of philosophy. And the accompanying question of how God grounds possibilities has received a multiplicity of answers. My focus in this article will be entirely on this question and not with Kant’s argument for the view that God grounds all possibility.

One answer to the question of how God grounds all possibilities, which is given by Descartes, is that possibilities are possible because God wills them to be such.Footnote ⁴ Had God willed a different set of possibilities, then those would have been possible. This counterfactual, we can note, has an impossible antecedent: it is impossible for God to have willed a different set of possibilities, since none of the possibilities that he in fact willed include his willing a different set of possibilities.Footnote ⁵ With a sufficiently sensitive treatment of counterfactuals, this Cartesian picture can avoid inconsistency, but it is nevertheless guilty of a form of arbitrariness. Since there were no possibilities prior to God’s willing, there was nothing to constrain what God could have made possible in his willing of the possibilities. This is a position that Descartes accepts, though one that other philosophers in his time found indefensible.

Leibniz, who took this arbitrariness concern quite seriously, did not accept a voluntarist account of the grounds of possibility. As a proponent of the principle of sufficient reason, Leibniz was unwilling to accept both ungrounded possibilities and insufficiently grounded possibilities. Instead, Leibniz accepted an intellectualist account of how God grounds possibilities. For Leibniz, God does not ground possibilities by willing them to be possible, but simply by considering them.Footnote ⁶ It is in the act of thinking these possibilities that God makes them possible. Which possibilities there are is not the result of any arbitrary choice, but the view does not fully avoid concerns of arbitrariness, as it is hard to see why certain possibilities are thinkable while others are not. Could God have made a thinkable proposition unthinkable? What grounds the thinkability of propositions? It is difficult to find a non-arbitrary answer to these questions.Footnote ⁷

It is generally not thought that Kant is a Cartesian or a Leibnizian about the grounding of possibilities.Footnote ⁸ Nicholas Stang (Reference Stang2010) proposed that Kant’s account might have more in common with the account of possibility-grounding offered by Christian Crusius. According to Stang, Crusius takes God to ground all possibilities by means of his power. In particular, God grounds all possibilities by having the power to make those possibilities actual. Hence, it is neither God’s actual will nor actual intellect that ground the possibilities. It is God’s potential for bringing about some state of affairs that makes that state of affairs possible.Footnote ⁹

This capacity-based approach has much going for it philosophically. It does a much better job than the voluntarist and intellectualist approaches at avoiding the charge of arbitrariness. It seems right to say that God has the power to bring about unicorns, but not the power to bring about square circles. And what God’s powers are is the kind of thing that should follow from God’s nature. So, there is a non-arbitrary answer to the grounding question: God grounds possibilities in virtue of his power, which flows from his nature. No arbitrary decisions must be made, and there is no further question of what grounds the thinkability of possibilities.

The capacity-based approach also has a philosophical cost. Since the notion of a ‘capacity’ or a ‘power’ is an implicitly modal notion, the approach gives up on providing a reductive analysis of modal terms into non-modal terms. It is not obvious that this is a problem, however, since it is not clear that a reductive analysis is what Kant is after. After all, in inquiring into the grounds of possibility, Kant is deploying the notion of a ‘ground’, which is itself a modally loaded notion.Footnote ¹⁰ I take the question of how the possibility of instantiation of predicates is grounded to be compatible with a non-reductive analysis of modal notions in general.

A cursory look at the textual evidence suggests that Kant’s early thinking about the grounding of possibilities took something like this capacity-based form.Footnote ¹¹ Of particular interest is Kant’s example, in the Only Possible Argument, of how a quotidian possibility, the possibility of a flowerbed or an avenue, could be grounded:

The point of Kant’s discussion is to establish that God, as the ground of all possibility, must be ‘a mind’ (2: 87). This is because only a mind is able to have the kinds of powers that would allow for the kind of ‘order’ required in creating a flowerbed. This discussion not only suggests that Kant accepted a Crusian conception of possibility-grounding, it also shows why Kant is not a Leibnizian about the grounding of possibility. Kant does agree with Leibniz that possibilities are grounded in God only if God is able to think about them. But this is because thinking is a component of intentional action. Ultimately, it is God’s capacity for intentional action, not his capacity for contemplation that grounds all possibility.

There appears to be, however, an insurmountable problem for the capacity-based interpretation of Kant: it fails to give an account of how God’s own possibility is grounded. This concern is raised against his old view by Stang (Reference Stang2016), who quotes the following passage from the Only Possible Argument:

As Stang notes, Kant in this passage commits himself to the following thesis: God is not only the ground of external possibilities, he is also the ground of his own possibility, which includes the possibility of his wisdom, goodness, and most importantly power.Footnote ¹² But the claim that God is the ground of the possibility of his own power is unacceptable for the capacity-based approach. Such an approach would have to claim that God grounds the possibility of his own power by having the power to formulate and execute a plain that brings about the existence of his power. In short, God would ground the possibility of his own power by having the power to bring about his own power.

Stang takes this problem to be decisive against a capacity-based approach and instead concludes that Kant’s ultimate philosophical position is an agnosticism about how God grounds possibilities (Reference Stang2016: 118). As we will see, this defeatism is not warranted in the interpretation of the pre-Critical Kant.

3. Predicate dualism

In a series of papers, Andrew Chignell has defended a reading of how possibilities are grounded which differs significantly from those discussed above and which Chignell more closely associates with Spinoza.Footnote ¹³ The core of Chignell’s view is that God grounds possibilities by instantiating them. This is motivated, in part, by the need to ensure that Kant be an actualist about possibilities: that he hold that what is possible is grounded in what is actual (OPA, 2: 79). And it pushes the actualism as far as it can go: the possible is grounded in the actual by being actual.

Yet this first-pass answer cannot be right as an account of how all possibility is grounded, as it would require God to instantiate too many possible predicates. On a naïve reading, it collapses into incoherence. This is because, for example, it is possible for something to be red, and it is possible for something to be non-red. If both possibilities are grounded by being instantiated by God, and God is the ground of all possibility, it would follow that God is both red and non-red. Hence, this naïve treatment of the instantiation account will not work.

To avoid this problem, Chignell distinguishes between two kinds of predicates: fundamental predicates and derivative predicates. A fundamental predicate is a predicate that is both ‘positive and unanalyzable’ (Reference Chignell2012: 640). A predicate is positive just in case it is ‘possessed of some genuine content of its own – that is, content that is not merely the negation of some other predicate’, while an unanalysable predicate is one that ‘cannot be “constructed” or derived from simpler predicates via operations like negation, disjunction, conjunction, and so forth’ (ibid.).Footnote ¹⁴ Since all unanalysable predicates are positive, then, it follows from Chignell’s definition that all unanalysable predicates are fundamental.

In distinguishing the fundamental from the derivative, Chignell is able to give a dualist account of how possibility is grounded. The fundamental predicates are grounded directly in God, by being instantiated. The derivative predicates are constructed via logical operations from the fundamental predicates (2012: 657).Footnote ¹⁵ These derivative predicates are still grounded in God, but only indirectly.Footnote ¹⁶ They are grounded in the fundamental predicates that are used to construct them. One textual motivation for this view comes from Kant’s own claim that:

We can lay out Chignell’s interpretation of this claim as involving the following two claims:

• Fundamental Instantiation Grounds: if F is a fundamental predicate, then the possibility of F is grounded by God in virtue of the fact that God instantiates F .

• Derivative Instantiation Grounds: if D is a non-fundamental (derivative) predicate, then the possibility of D is grounded by God in virtue of the fact that God instantiates the fundamental predicates out of which D is constructed.

As it stands, these two theses are insufficient to characterize the nuances of Chignell’s reading. This is because Chignell also thinks that some logical constructions out of fundamental predicates are impossible due to being ‘repugnant’ (Reference Chignell2009: 172). But the grounding question at issue here is not which possibilities there are, but rather how those possibilities are grounded in God. So, I take the two theses above to be sufficient for characterizing the aspects of Chignell’s view that have import for the question at hand.

With these two theses in mind, we can see how Chignell’s reading can avoid the charge of incoherence. On this account, God need only instantiate the fundamental predicates. As long as all the fundamental predicates are jointly consistent, incoherence is avoided. As we will discuss below, there are reasons to suspect that the incoherence rearises. But for now, it is important to focus on a structural difference between Chignell’s instantiation account and the accounts mentioned in the previous section. This is the fact that Chignell’s account rejects the idea that all possible predicates are grounded in the same way. The way in which the fundamental predicates are grounded is by instantiation, while the way in which the derivative predicates are grounded is by construction. I think this is an improvement, though perhaps a diagonal one. I agree that a dualistic account of grounding is the right way to interpret Kant, but will later reject both the idea that such an account will involve two ‘steps’ of grounding, and that it will involve two different kinds of predicates.

For now, it should suffice to note that Chignell’s account fails to make sense of how Kant takes certain possibilities to be grounded. Return to the case of the gardener and the flowerbed. The gardener is supposed to be the ground for the possibility of the flowerbed. Now, since the gardener is not himself a flowerbed, the only way in which he could be the ground of the possibility of the flowerbed would be by instantiating some predicates which can be used to construct the predicate ‘flowerbed’. But which predicates could those possibly be? It is hard to see which predicates that apply to a gardener could be put through logical operations in a way that generates the predicate ‘flowerbed’. Perhaps the best try would be to say that the fundamental predicates out of which the gardener is constructed (extension, force, etc.) are also the fundamental predicates out of which the flowerbed is constructed. But then we lose an account of why it is the gardener, rather than any arbitrary material object, that is the ground of the possibility of the flowerbed; or why it is the flowerbed, rather than any arbitrary material object, whose possibility is grounded by the gardener.

Chignell cannot account for general, quotidian grounding of possibility. Yet Kant not only believes that quotidian possibility-grounding is a phenomenon, he also thinks that this kind of possibility-grounding gives us a way to understand the way in which God grounds possibilities. The point of the ‘flowerbed’ example is to demonstrate that God must be a mind, since ‘in the absence of understanding, no power or generative force, nor any other data of possibility, are adequate to render the possibility of such order complete’ (OPA, 2: 88). By ruling out the quotidian grounding of possibility of which humans are capable, Chignell blocks Kant’s central argument for theism (as opposed to mere deism): without an account of how humans can ground the possibility of flowerbeds, we are left without an argument for the claim that God acts intentionally.Footnote ¹⁷

This is an important concern, but putting it aside, there is a second and deeper problem that arises from Chignell’s account. Chignell thinks that there are two kinds of possibility-grounding because there are two sets of possibilities. There are the possibilities involving fundamental predicates, which are grounded by instantiation, and then there are the possibilities involving the derivative predicates, which are grounded by means of logical operations. But as Stang (Reference Stang2016) stresses, Kant believes that there are certain fundamental predicates that are not had by God. Importantly, Kant says:

A straightforward reading of this passage takes Kant to treat both extension and pain to be ‘positive and unanalysable’. Yet Kant takes God to lack both. This contradicts Chignell’s reading, according to which every fundamental predicate is instantiated by God. Chignell could, of course, claim that although for Kant extension and pain are positive predicates, they are not unanalysable predicates. But both of these cases are problematic. First, Kant explicitly claims in the passage above that pain is not a deprivation. This means that constructing ‘pain’ out of a negation of pleasure will not work.Footnote ¹⁸ This makes it difficult to see how pain is to be constructed at all. When it comes to extension, this is even harder, as Kant seems to at least allow the possibility that extension is unanalysable earlier in the Only Possible Argument:

In this passage, Kant does not commit outright to the claim that extension is unanalysable. But he seems content with the fact that it may turn out to be so. Chignell’s interpretation of Kant’s God makes God monstrous: he instantiates mutually incompatible predicates. Chignell seems content with ascribing this position to Kant, but this is a deeply unsatisfactory conclusion to draw.

In recent work, Uygar Abaci has attempted to carve out a version of predicate dualism that does not commit Kant to this ‘monstrous’ ground of possibility.Footnote ¹⁹ For Abaci, there is also a distinction between fundamental predicates and non-fundamental predicates. Unlike Chignell’s approach, however, the distinction between fundamentality and non-fundamentality is itself characterized in modal terms:

Given this account of what it is to be a (relatively) non-fundamental predicate, we then define the two kinds of predicates that can be grounded:

• Fundamental Predicate Grounds: if F is fundamental, then x grounds the possibility of F in virtue of the fact that x instantiates F .

• Non-Fundamental Predicate Grounds: if D is derivative, then x grounds the possibility of D in virtue of the fact that x instantiates some fundamental predicates Fs such that D is derivative of the Fs.

There is a problem with this account. For Abaci, what it is for something to ground the possibility of some non-fundamental predicate is for it to have a property that grounds the possibility of that non-fundamental predicate. This is not entirely uninformative: it tells us that the grounding of non-fundamental predicates goes by way of the instantiation of fundamental predicates, but not in a way that requires the instantiation of the non-fundamental predicate itself. In a sense, this is progress over Chignell.

But this also leaves the question of the grounding of possibility unanswered. What does it take for a fundamental predicate to ground the possibility of a derivative predicate? For Chignell, the answer is ‘logical construction’. Abaci does not accept this answer because he agrees that it is too restrictive. But he offers no replacement. All he says is that a fundamental predicate grounds the possibility of a derivative one by being such that, if the fundamental predicate did not exist, the derivative predicate would not be possible. This is as good as no answer at all. Yet this is something Abaci is happy with, as it allows him to show that there is something inherently wrong with the pre-Critical Kant’s view, and it allows him to present the Critical Kant as coming to the rescue:

We have again arrived at a defeatist position. Perhaps this approach is tempting because it allows us to say that the Critical Kant ‘wins’ over the pre-Critical one. But this is, in general, not a great strategy in the history of philosophy. It cannot be the case that every mistake the pre-Critical Kant makes is one that he made because he was not yet the Critical Kant. Such approaches concede defeat too quickly and prevent us from seeing the richness of the views in question.

In sum, Chignell’s dualism draws a clear and informative distinction between fundamental and derivative predicates, but fails to give an extensionally adequate account of which of these predicates belong to God. Abaci’s dualism saves extensional adequacy by giving an uninformative account of the distinction between fundamental and derivative predicates. Neither of these options is tempting.

4. Grounding dualism

So far, we have seen two ways in which one could give an account of how God grounds possibilities: one can give a monistic account, and claim that all predicates have their possibility grounded in the same way, or one could give a non-monistic account and claim that the possibility of some predicates is grounded differently than that of others. The voluntarist, intellectualist, and capacity accounts are all monistic. Chignell’s and Abaci’s are not. But these accounts are non-monistic as a result of taking there to be two sorts of predicates which demand to have their possibility grounded in different ways. This view runs into trouble precisely for this reason: Kant takes there to be predicates that are not instantiated by God, yet are just as positive as those predicates that are instantiated by God. Chignell embraces this problem as a feature of Kant’s view. Abaci avoids this problem by fiat: he postulates that the predicates that are not instantiated by God are ‘non-fundamental’ and then defines non-fundamentality as that whose possibility is grounded in the fundamental predicates (those instantiated by God).

But this is not the only way to give a non-monistic answer to the grounding question. One could instead say that there is only one kind of predicate, but that there are two kinds of grounding relations.Footnote ²⁰ On such an account, some possibilities are directly grounded by means of one grounding relation, while other possibilities are directly grounded by means of another. And, importantly, this is not because the possibilities in question are of intrinsically different kinds, as Chignell and Abaci would have it. No distinction between fundamental and derivative predicates is drawn here. The difference in grounding relations is simply a result of how the predicates in question relate to their ground.

To flesh out this proposal, we need to return to Crusius’ account of ground. Despite noting Crusius’ influence in Kant, extant discussions of Crusius fail to fully capture the details of his position. As I will show, Crusius’ view on ground is dualistic in precisely the way described above.

Crusius notoriously distinguishes between two senses of ground: real grounds and ideal grounds. Real grounds bring about an object, whereas ideal grounds merely bring about cognition of an object. Fire is a real ground of warmth, but warmth is only an ideal ground of fire.Footnote ²¹ Yet this is not the place where Crusius’ distinctions end. Much less notorious is the fact that Crusius divides the category of real ground into two distinct and equally fundamental kinds of real grounding relations. In his Sketch, he explains:

For Crusius, there are two different ways in which something can bring something about. First, there is familiar notion of an active force (tätige Kraft), where the ground brings about the consequence through the manifestation of a power. Second, there is the notion of a metaphysical (‘existential’) ground (Existentialgrund), where the ground brings about its consequence through its existence alone. ^{Footnote 22,Footnote 23} While fire is the efficient cause of warmth, the parts of an object (together with their relations) are the existential real ground of the geometrical properties of the object. Crusius is clear that both of these are forms of real grounding relations (Realgründe), rather than mere grounds of knowledge (Idealgründe) (Sketch, §§34-6). And neither one of these real grounding relations is more real than the other. These are simply two different, equally real ways in which something can bring about something else.

In pseudo-Aristotelian fashion, Crusius is endorsing the claim that why-questions have irreducibly different kinds of answers.

As Crusius notes, furthermore, a real ground can both ‘bring about’ (hervorbringt) something or it can ‘make [it] possible’ (möglich macht). It is important to note that this distinction cross-cuts the distinction between efficient causes and existential grounds. Something can ground a possibility by means of its power: the unburnt wood grounds the possibility of fire. And something can ground a possibility by means of its existence: the shape of a stamp is an existential ground of the possibility of the shape of the printed seal that it would leave (Logic, §141).

The interpretation of Kant that I am proposing is that Kant joins Crusius in accepting two different forms of real grounding, and that he also joins Crusius in accepting the correlative two different forms of the real grounding of possibilities. A close reading of Kant’s discussion of ground in his 1755 New Elucidation reveals some deep similarities between the ways in which both of these thinkers approach the question of how possibility is grounded in general. It also supplies us with an answer to the question of how God grounds all possibility.

5. Kant’s dualism

Kant’s first systematic treatment of the notion of ground, his New Elucidation, engaged significantly with Crusius’ metaphysical views. Not only is Crusius directly mentioned frequently, including in an extended discussion of the metaphysics of freedom, but the content of the essay itself involves constant responses to the objections to Leibniz and Wolff raised in Crusius’ writings.Footnote ²⁴ In fact, it is very common for the young Kant to attempt to take Crusius’ ideas and re-cast them within his own conceptual framework. For example, while Crusius defines ground in terms of ‘bringing about’ (hervorbringen), Kant defines ground in terms of making intelligible (1: 392).Footnote ²⁵ Similarly, many of Crusius’ distinctions receive translations into Kant’s new vocabulary. And the distinction between efficacious and existential grounds is no exception. In introducing the notion of an antecedently determining ground, which Kant also characterizes as a ‘ground of why’ or a ‘ground of being’, Kant adds a subtle footnote:

The similarity between Kant’s notion of an identical ground and Crusius’ notion of an existential ground is striking. In fact, Kant’s example of an identical ground is a simplified version of an example Crusius himself gives when illustrating the notion of an existential ground:

Similar to Crusius, Kant holds that there are two ways in which one predicate could answer a ‘why’ question about another: either by being prior to it or by being ‘identical’ with it. Despite the symmetry exhibited in the notion of an ‘identical ground’, the identity here at issue is not numerical identity. In Kant’s example, the concept three-angled is not numerically identical to the concept three-sided. Rather, these concepts are, if I may use a metaphor, metaphysically simultaneous: the instantiation of one is sufficient for the instantiation of the other, but neither of them is the efficient cause of the other. They are mutual existential grounds. Kant, like Crusius, is a dualist about grounding relations.

It is important to note that, unlike the dualisms proposed by Chignell and Abaci, which of these grounding relations obtains does not depend on the nature of any single one of its relata. As Crusius puts it:

Kant’s example of ‘a triangle has three sides’ is a case in which the predicate ‘has three sides’ is identically grounded in the subject ‘triangle’. But we can draw inspiration from Crusius and give a different example: if I were to draw a triangular figure on a blackboard and say ‘the figure I drew has three sides’, this will not be a case in which the predicate ‘has three sides’ is identically grounded in the subject ‘the figure I drew’. In this latter case, nothing in the concept of the subject contains the predicate, and hence this truth requires a further determining ground. One and the same predicate (e.g., ‘three-sided’) can have an identical ground in one context and a non-identical ground in another. What is relevant is not what the predicate is, but how the predicate relates to the subject.

This is a point that extant dualist readings miss. The attempts to define the two ways in which possibility can be grounded have presupposed that there are two kinds of predicates: ones whose possibility is grounded in one way, and ones whose possibility is grounded in another. But Kant’s (and Crusius’) dualism about ground does not take this form. And we can see this rather clearly if we return to Kant’s example of a flowerbed:

For Kant, a gardener’s understanding and will are the grounds of the possibility of flowerbeds. But it is not the case that, as a general fact, the understanding and will of something are more fundamental than the thing that they create. Hence, just as the gardener is the ground of the possibility of a flowerbed, so were Kant’s parents the grounds of Kant’s possibility. Insofar as the gardener can bring about the existence of the flowerbed through its understanding and will, so can Kant’s parents bring about Kant’s existence through their understanding and will. Yet it would be absurd to think that, as generations continue to be born, human beings are becoming less metaphysically fundamental.Footnote ²⁷ The right thing to say is that the way in which the gardener’s will and understanding are ‘prior’ to the possibility of the flowerbed has nothing to do with fundamentality and everything to do with the kind of relation (that of being a potential cause) that holds between these two objects.

We now turn to the implications of this conception of ground on the question we started with. We will use this Crusius-inspired dual notion of ground to understand how God grounds all possibility.

6. Grounding possibilities

In the discussion of Kant’s argument for the existence of God, it is common to acknowledge, but not to explore, the way in which this argument appears in Kant’s New Elucidation. This is because the argument in the New Elucidation takes up two paragraphs of text, while the argument as presented in the Only Possible Argument stretches over several pages, and spells out its steps in significant detail.

But the focus on the later version of Kant’s argument has led readers to overlook the fact that Kant gives a straightforward answer to the grounding question in the New Elucidation:

This passage might seem puzzling at first, as it is difficult to see what it would mean to say that something is both prior and identical to something else. But given the Crusian background, this idea is not strange at all. God’s existence is ‘prior’ to his possibility in that his existence is a real ground of his possibility. God’s existence is ‘identical’ to his possibility in that the grounding relation between God’s existence and his possibility is that of an identical ground. Kant’s claim here is that God grounds his own possibility in the same way that trilaterality grounds triangularity: the former can help us understand the latter, despite being metaphysically ‘simultaneous’ with it.

This passage is quite informative. It tells us, as explicitly as Kant ever tells us anything, that the way God grounds his own possibility is by being its identical ground. Of course, this does not mean that this is the way God grounds all possibilities. After all, identical grounding is not the only kind of grounding. Rather Kant follows Crusius in thinking that there is more than one way in which God can ground a possibility: God can be an identical ground of a possibility, or he can be the non-identical (antecedent) ground of a possibility.

As with standard cases of this distinction, what kind of ground God is does not depend on what kind of predicate we are discussing. It is not the case that some predicates are more suited than others to have identical grounds. Rather, the grounding relation at issue depends on the relation that holds between the predicate and the subject. The predicate ‘has three sides’ has an identical ground with respect to the subject ‘triangle’, because the subject contains the predicate in it. The same predicate has non-identical ground when the subject is ‘the figure I drew’, because the subject does not contain the predicate in it.

This general thought can be carried over to cases of grounding possibility. In order to see how, it may be valuable to return to the Crusian, capacity-based approach. We can formulate this approach more precisely as follows:

• Capacity Grounds: x causes F just in case the exercise of x ’s power results in the existence of F .

• Capacity Grounds of Possibility: x grounds the possibility of F just in case x has the capacity to cause F .

This broadly Crusian approach to grounding possibilities is, in fact, a particular instance of a more general principle: the principle that the ground of possibility is the possible ground of actuality. We can write down that principle as follows:

• Possibility-Ground Link: x grounds the possibility of F just in case it is possible that x grounds the existence of F .

It is worth reminding the reader that capacity-based accounts do not attempt to give reductive analysis of modality into non-modal terms. But there is still a kind of analysis of modality here: de re modality is analysed into a particular kind of de dicto modality. And the Possibility-Ground Link explains how the gardener grounds the possibility of the flowerbed. He does so because it is possible that he is an antecedently determining ground of the existence of the flowerbed. This approach can also be extended, once we introduce the notion of an identical ground:

• Identical Grounds: x identically grounds F just in case the existence of x is sufficient for x being F .

• Identical Grounds of Possibility: x identically grounds the possibility of F just in case it is possible that x identically grounds F .

If we assume that sufficiency is a modally robust notion, we should conclude that the identical ground of something is also the identical ground of that thing’s possibility (though we should not assume the converse). If the triangularity of a figure grounds its trilaterality, then it is also what makes its trilaterality possible.

Taking on the Possibility-Ground Link, we can turn back to the problem of how possibilities are grounded. Recall the original capacity-based approach. On this approach, God grounds all possibilities by means of what I have labelled Capacity Grounds: by having the capacity to cause those possibilities to be actual. The problem for this approach was that it failed to give an account of how God grounds his own possibility. Most pressing, it failed to give an account of how God’s own capacities (God’s power) are grounded. It simply cannot be the case that God grounds his own capacities by having the capacity to bring them about.

On the dual-grounding view, this is not a problem at all. To use Crusian terminology, God is not the efficacious ground of his own nature, he is the existential ground of his own nature. To use Kantian terminology, God’s existence is the identical ground of his essential properties. Since all identical grounds are also identical grounds of possibility, it follows that God’s existence is the identical ground of God’s own possibility. And hence, God’s existence is ‘prior to, or, if you prefer, identical with possibility’. Recall one of the key motivating passages for the predicate dualist view:

As we can now see, this passage is entirely consistent with the grounding dualist view. What Kant is here claiming is that there are two ways in which possibilities are grounded. This in no way favours the predicate dualist reading over the grounding dualist reading. We can then formulate Kant’s approach to the grounding of possibility in the New Elucidation as follows:

Grounding Dualism (NE): God is the ground of all possibility because

1. 1. If God is F, God is the identical ground of the possibility of F-ness.

2. 2. If God is not G and G is possible, God is the antecedent ground of the possibility of G-ness.

Why, we might wonder, does Kant think that God is the only being whose possibility has an identical ground? The answer can be found in Kant’s later discussion of contingency, where Kant argues that ‘nothing which exists contingently can be without a ground which determines its existence antecedently’ (1: 396, italics added). That is, all contingent beings must have their existence grounded in something external to them. Via the Possibility-Ground Link, this means that all contingent beings must also have their possibility grounded in something external to them. But this does not apply to God, since God is the only being ‘the existence of which is prior to the very possibility both of itself and of all things’ (1: 395). As such, God is the only being whose existence does not have an antecedently determining ground, and therefore the only being capable of having an identical ground of its own possibility.Footnote ²⁸

When it comes to the possibility of concepts that are not in God’s essence, the story is different. These concepts are such that their instantiation must be brought about causally. Consider, for example, extension. Kant does not think that God is extended. Nevertheless, God has the understanding required for conceiving of a plan to create extended beings, and he has the will and power to execute such a plan. Hence, God is capable of causally bringing about the existence of extended beings. Via the Possibility-Ground Link, it follows that God is the (causal) ground of the possibility of extension. The same reasoning here can apply to every other non-divisible predicate. And the result is that all possibility is grounded directly in God, though it is not all grounded by means of the same grounding relation. When it comes to the concepts that constitute God’s essence, their possibility is identically grounded in God; when it comes to all other concepts, their possibility is antecedently grounded in God.

One concern that a reader might have about this approach is that it focuses too heavily on an off-hand remark by Kant in a footnote to the New Elucidation. The notion of ‘identical’ grounds no longer appears in Kant’s later writings, and, most importantly, it does not appear in Kant’s major defence of his possibility proof, the Only Possible Argument. In this later essay, Kant does away with the distinction between antecedent and identical grounds, and instead draws a distinction between logical and real grounds. Why, then, should we expect Kant’s solution to be given in terms of this very specific notion of an identical ground?

I think the correct attitude to have about this is that the distinction between identical and antecedent grounds eventually transforms into the distinction between logical and real grounds.Footnote ²⁹ As a result, we have a straightforward way of formulating the grounding dualism interpretation as follows:

Grounding Dualism (OPA): God is the ground of all possibility because:

1. 1. If God is F, God is the logical ground of the possibility of F-ness.

2. 2. If God is not G and G is possible, God is the real ground of the possibility of G-ness.

This interpretation of OPA has not, as far as I am aware, been considered. And the reason for this is that Kant seems to insist, in OPA, that he is looking exclusively for the real ground of all possibility. But this is not what Kant actually says. What Kant says is

Kant here is making it very clear that the ‘real ground’ of a possibility is that which grounds the internal possibility of realities other than its own. Its own reality, on the other hand, is simply grounded as a determination ‘existing within’ it. It is not a stretch to suppose that Kant’s view here is that the possibility of God’s predicates is logically grounded in God’s existence, while the possibility of all other predicates is really grounded in it. This is simply a natural extension of the view expressed in the New Elucidation, in which God is ‘prior to, or, if you prefer, identical to possibility’.

7. Conclusion

As has been frequently discussed, Kant’s Critical attitude towards his Only Possible Argument is not as optimistic as his pre-Critical view. From the perspective of the Critical Kant, something goes wrong with his proof.Footnote ³⁰ There are two sorts of assessment one can make about what the Critical Kant finds questionable. First, we could think that Kant’s main concern is with the soundness of the argument in question, and that the Critical Kant is only indirectly concerned with whether the conclusion of the argument is true. Second, we could think that Kant’s problem is primarily that the conclusion of the argument is unacceptable, and that the argument in question must therefore be unsound. The skeptical and defeatist approaches to ‘how does Kant’s God ground all possibility?’ are attempts to point to this second assessment. What Chignell, Abaci, and Stang all want to show is that there is something problematic about the claim that God grounds all possibility, and that this problem is highlighted by the issue of how God grounds his own possibility.

To illustrate the way in which this approach avoids these difficulties, it might be helpful to consider an objection from Abaci. When discussing Stang’s (Reference Stang2010) reading, of which ours is a descendant, Abaci notes:

Abaci’s objection has two parts. First, he notes that it is not clear in what way God’s fundamental predicates are able to ground their non-fundamental derivatives. Second, it is not clear why any predicate other than omnipotence could be fundamental, since it is ultimately through God’s power that all other predicates are grounded.

These are good objections to the predicate dualism implicit in Stang’s account. But, importantly, this set of objections assumes that the relationship between God’s predicates and the other predicates is a relation between the fundamental and the derivative. As I have argued, this is not the right way to think about either Crusius or Kant. It is not the case that ‘omniscience’ is a fundamental predicate from which ‘not-wise’ and ‘somewhat wise’ are derivative. This is not at all how possibilities are grounded via the Possibility-Ground Link. The possibility of ‘not-wise’ is grounded in the possibility of creating and executing a plan that results in the existence of something that is not wise. Creating such a plan might require omniscience, but ‘not wise’ is not special in this respect. This grounding story is the same as the grounding story for any other predicate that God does not instantiate. Any predicate that is not internal to God’s essence derives its possibility from God’s capacity to bring about the actuality of that predicate through the conception and execution of a plan.

The same thought goes for the second objection Abaci raises. Abaci wonders why ‘omnipotence’ is not the only fundamental predicate, since it is the only predicate needed to bring about all other possibilities. The answer to this question is that the notion of fundamentality is entirely out of place in a discussion of how possibilities are grounded. The distinction between fundamental and derivative predicates was an artifact of a particular attempt to give a non-monistic account of how possibilities are grounded. The Crusian account is a non-monistic account that has no need for such artifacts. And hence the question of which predicates are fundamental, whatever exactly that might mean, is not needed for a satisfactory account of the modal issues at hand.Footnote ³¹

What we can see, then, is that there is nothing intrinsically wrong with Kant’s pre-Critical account. There is no problem, from within the philosophical framework of the early Kant, in answering the question of how God grounds possibilities. God grounds possibilities either by being identical to them or by being a non-identical ground (i.e., a potential cause) of them. These two grounding relations are distinct and irreducible to one another. There is nothing incoherent or unintelligible in this answer. There is no need to claim that God’s method of grounding possibility is unknowable. God grounds possibilities in the same way that anything else does. Whatever Kant’s problems might be in the Critical period, they are primarily concerns with the argument, not with the coherence of its conclusion.