Kant’s Pre-Critical Argument for the Existence of God

Most philosophers are aware that Kant criticized the main arguments for God’s existence

in his Critique of Pure Reason and can probably give at least the outline of his critique of

the ontological argument, upon which he believes the other main arguments for God’s

existence depends. For most philosophers, that is Kant’s last word on the subject: there

can be no proofs for God’s existence, which he avers is not necessarily a bad thing. The

wings of reason must be clipped in order to make way for faith. Thus, whatever we may

make of his attempts to justify belief in God on moral grounds, the notion that there

might be a successful proof of God’s existence is something that, for Kant, must be

regarded as a closed book. It will come as a surprise to many philosophers that despite

the official doctrine of the First Critique, Kant was working himself, at one point in his

career, toward a different, more Aristotelian account of being or existence, one from

which a new and unique argument for God’s existence could be mounted. Kant did in fact

offer his own proof for the existence of God, in one of his lesser-known works called On

the Only Possible Basis for a Demonstration of the Existence of God. In my opinion, this

argument deserves more attention than it has received in the literature and in this essay, I

hope to provide some of that deserved attention.

In this essay, published in 1762, Kant presents essentially the same critique of the

ontological argument that he presents in the First Critique, with the intention of

challenging the reigning notion of possibility as merely logical possibility, for which the
criterion is apparent coherent conceivability, which in turn reduces to apparent coherent

imaginability.1 On this view, famously associated with Descartes and shared by all the

major philosophers of the early modern period including Hume, for something to be

possible is simply for it to be imaginable without contradiction. Certainly, square circles,

two-sided triangles, things that are red and green all over at the same time or taller or

older than themselves are impossible beings, because they are inconceivable, not just for

us, but in and of themselves.2 It is therefore natural to assume that anything that is

conceivable without contradiction must be possible. However, there are different senses

in which this is the case, as comes to the fore when we investigate the connection

between possibility in this sense and the notion of existence. In his proof, Kant was

working his way toward an analysis of this topic that has both been neglected and that I

believe can still be useful to us today.

Many philosophers will be inclined to dismiss the Beweisgrund (as it is called) as

irrelevant to our understanding of Kant because it is a pre-critical work and one we

suppose that Kant would have repudiated in light of his Critical “Copernican revolution”

as a false start, perhaps to be excused as a bit of juvenilia. However, when we consult the

historical record, this does not appear to be the case. Kant republished the essay: twice, in

1770 and 1794 as a separate book, and as part of a collection of his essays and occasional

pieces, in 1798, without altering the text. Further, the doctrine adumbrated in the

Beweisgrund also reappears in the First Critique and in his Lectures on the Philosophy of

Religion, recorded by students in 1783-4. The doctrine of possibility and actuality

1
I will be at pains to distinguish merely logical possibility, the only kind known to Kant, which I here call
formal possibility, from logical possibility in the full and proper sense.
2
In the terminology I have developed elsewhere, here we have not just inconceivability in the conceiving,
but inconceivability in the conception; in the latter case it is what we are attempting to conceive, rather
than our conceiving of it, that is impossible.
presented there both illuminates and is illuminated by the discussion of the categories of

quality in the section entitled Postulates of Empirical Thought. He returns to the

argument in the Ideal of Pure Reason in the First Critique and in the Lectures on

Philosophical Theology, in both places retaining the argument without, however,

allowing that it is a demonstration of God’s existence. On the one hand, the strictures of

the Critical Philosophy prevent him from affirming any substantive claims about

noumena; on the other, he still finds the argument persuasive at some level. However, he

is unable to determine how to make a place for it in the Critical philosophy, offering

different accounts of its significance in different contexts. Throughout, Kant insists that

we can have no other account of the nature of possibility (as he characterizes it) than the

one provided by this argument, and that it is the origin of our idea of God as ens

realissimum. He thus regards it in one place as stating a necessary condition for the very

exercise of human reason and thus a “subjective” ground for belief in God. In another

place he describes it as a transcendental argument, which falls short of proving God’s

existence but which establishes that no thought about anything as existent is possible

without presupposing God’s existence as the “ground of all possibility.” In yet another

place, he treats it, like the apprehension of the teleology of nature in the Third Critique,

as a “necessary illusion” of reason without which theoretical inquiry would be impossible

for us.3

Kant’s “possibility proof” for God’s existence remains a significant element of

classical, pre-critical philosophy in Kant’s system, and derives directly from reflection on

the Classical and Scholastic tradition summarized in the rationalist textbooks from which

3
For corroboration of these claims, see Allen Wood, Kant’s Rational Theology, Ithaca, NY, Cornell
University Press, 1978, 70-79.
Kant continued to lecture his entire academic career. It both criticizes and critically

appropriates elements from that tradition in an attempt to work out a better theory of

possibility than that on offer in those books. The failure of Kant (and his successors down

to the twentieth century) to justify the revolutionary claims of the “Copernican

revolution” in philosophy permits those of us who embrace the Classical tradition in

philosophy he rejected to allow Kant’s arguments to develop in accordance with their

inner dynamism and reach the conclusions to which they genuinely lead. These

conclusions will be a rather different set of conclusions than those reached by Kant

himself, stunted and distorted as they were by his artificial limitations on what we can

know using reason understood as a faculty of theoretical inquiry. Having removed Kant’s

Humean blinders, we will find that much that is obscure, difficult and apparently

inconsistent in Kant will disappear, not however without sacrificing much of what people

have thought was novel and admirable in his thought. For this, I make no apology.

At the same time, I have not been able to reconstruct Kant’s argument as it

appears in his texts as a valid proof or, at least, an argument that is both valid and

plausible. This appears to be par for the course where Kantian argumentation is

concerned. However, I remain convinced that Kant has interesting ideas about possibility

and existence, somewhat unfamiliar to contemporary philosophers, that are worthy of

being more widely known and from which a successful proof of God’s existence can be

constructed. What follows is my attempt, using modern modal ideas, to construct such a

proof; I make no claim to exclusivity in this endeavor, and invite other philosophers to

explore this sort of proof in greater detail. However, before turning to that, let me discuss
a distinction drawn by Kant in the Beweisgrund that is centrally important to the proof

that I shall give.

Two Senses of “Possible” in Kant In the Beweisgrund and elsewhere, Kant

distinguishes between two sense of “possible.” The first sense of “possible” is a familiar

one, which we might call formal possibility.4 For something to be possible in this sense

merely means for the concept of that thing to be conceivable without contradiction. 5

According to Kant, nothing follows from possibility in this sense concerning the actual

existence of anything. The main error of the Ontological Argument consists in its

supposition that, if the idea of God as a necessary being is coherent, and thus possible in

that sense, that conceptual analysis alone can arrive at the substantive claim that God

actually exists. This does not mean, however, that the pre-Critical Kant believed that no

conceptual proof of God’s existence was possible. However, the notion of possibility

used in his proof is a different and more robust conception of possibility than that used in

the forms of the ontological argument rejected by Kant. We might call this material or

“real” possibility.6

4
Kant calls this logical possibility, see (e.g.) Immanuel Kant, Lectures on Philosophical Theology, Allen
Wood and Gertrude M. Clark, trans., Ithaca, NY, Cornell University Press, 1978, 46-47. However, while my
account of what I call formal possibility corresponds to and expresses Kant’s account of logical possibility,
I reserve this latter term for the sort of possibility envisaged in contemporary modal logic, according to
which logical possibility de dicto means “true in some possible world” and logical possibility de re means
“exists in some possible world.”
5
However, we need to note in passing something not mentioned by Kant, i.e. for something to be
conceivable without contradiction is not just a matter of what is apparently coherently conceivable.
Something can be apparently coherently conceivable without its being the case that it is logically possible
– witness the possibility that Goldbach’s conjecture is false. It is apparently coherently conceivable that
that Goldbach’s conjecture is false. However, since Goldbach’s conjecture is a mathematical truth, it is
either necessarily true or it is necessarily false. So, the mere fact that I can apparently coherently conceive
of the possibility that Goldbach’s conjecture is false because I perceive no contradiction in that assertion
is not enough to secure the even the logical possibility that Goldbach’s conjecture is false. It is sufficient
only for what I have elsewhere called intrinsic epistemic possibility – see Duncan, The Proof of the External
World, Eugene, OR, Wipf and Stock, 2008, . As such, apparent coherent conceivability, while sufficient for
what I am here calling formal possibility, is at best necessary, not sufficient, for logical possibility in many
cases.
6
Kant uses this latter expression in the Lectures, op cit., 47.
 Furthermore, Kant puts an interesting spin on the notion of possibility. We often

suppose that inconceivability, and thus impossibility, is the baseline notion, given that it

seems to be intuitively self-evident that (to use Kant’s example) a four-sided triangle is

formally impossible. Indeed, we often define what we might call formal possibility

negatively in terms of formal impossibility, when we say that the possible is that which is

conceivable without contradiction. However, Kant maintains that formal impossibility is

not simple or unanalyzable in the way that the foregoing suggests. To the contrary,

formal impossibility is parasitic on formal possibility, just as inconceivability is parasitic

on positive conceivability. My judgment that a four-sided triangle is impossible is a

consequence of my recognition that no plane figure can instantiate both the property of

being a triangle and that of having four sides. My ability to recognize the incompatibility

involved in a single plane figure exemplifying both of these properties thus requires that I

clearly and distinctly conceive the natures of these properties as properties of plane

figures prior to that recognition. In turn, this requires that these properties should be

conceivable as formally possible prior to our recognition of their incompatibility in some

other case. Indeed, the very fact that we typically define the formally possible as that

which is not formally impossible, i.e. inconceivable, testifies to the fact that this concept

is not further analyzable and thus the product of a basic, intuitive grasp.7

Kant never gives a definition of possibility in this second sense, but it seems to be

a consequence of his account of existence in part one of the Beweisgrund. Existence, says

Kant, is not a predicate. Anything predicated of something else is posited relatively, in

7
As will emerge later, I regard formal possibility as a necessary condition for logically possibility, so that
nothing that is formally impossible is logically possible, and whatever is formally possible is logically
possible so far forth. However, formal possibility is not always sufficient for full-blown logical possibility,
as is amply shown the example in footnote 3, as well as that just discussed above. How these two notions
are related in a manner relevant to the current case will emerge in due course.
relation to something on which it depends and thus exists only in a derivative sense.

However, when we posit something absolutely and thus not in relation to anything else,

we use the existential “is,” which expresses a simple idea of positing not further

analyzable. For a thing to exist, then, is for it to be the object of a true existential

judgment; this is what it means for something to be or exist. To make this clearer, we

may say that for something to exist is for it to be real or actual outside of its causes,

although Kant would object that “real” and “actual” here are synonyms for “exists” in

this context, and thus equally unanalyzable as the word exists itself. However, it will

have to do, since Kant refuses to give us anything better.8

If for a thing to exist is for it to be real or actual outside its causes, and thus to that

extent self-subsistent, then material possibility can be specified as follows. P is materially

possible if it is capable of being real or actual at T. In turn, P will be capable of being real

or actual at T only if there is some specifiable set of conditions such that, if they were to

obtain at T, would bring it about that that P is real or actual at T or some time subsequent

to T. This is a very strong sense of the notion of possibility, which presupposes what I

have elsewhere called metaphysical possibility but supersedes this notion. 9 On my

account, something is metaphysically possible if it is intrinsically epistemically possible

(what I am here calling “formal possibility” and elsewhere have called logical possibility

in the broad sense), extrinsically epistemically possible (i.e., not ruled out on logical or

conceptual grounds by anything else we know) and physically possible (not ruled out by

8
This account of existence from Suarez; see his On the Essence of Finite Things, on the Existence of that
Essence and their Distinction, Norman J. Wells, trans., Milwaukee, WI, Marquette University Press, 1983,
103.
9
Or, rather, it adds a dimension to the account of metaphysical possibility that I had overlooked in my
2008 discussion. I gratefully accept the correction, while retaining the notion of metaphysical possibility I
presented there as useful in other contexts.
the laws of nature). What the notion of material possibility adds to this is the further

specification that some set of circumstances, not already excluded by antecedently actual

events, could obtain at T and, if those conditions did obtain, P would exist at T. I believe

that it is this very strong sense of “possible,” where it roughly means, “could actually

exist at T,” that Kant is relying on in his argument. Let us now consider an argument

along Kantian lines with this notion of possibility in mind.10

First Stage of the Argument I begin with a preliminary premises dependent upon

modern modal notions unknown to Kant and thus anachronistic in the context of his

original proof:

1. I exist.
2. If I exist, something exists.
3. Therefore, something exists.
4. If something exists, then it is logically possible that something exists.
5. It is logically possible that something exists.
6. Necessarily, it is possible that something exist.
7. Therefore, something exists in every world.

Kant would no doubt insist that any such proof must be a priori and thus cannot rely on

any empirical claim. However, it is doubtful that any proof of the sort envisaged by Kant

can actually succeed, so I have adopted a compromise. It is true that the claim that I exist

is both contingent and synthetic; nevertheless, I do grasp that claim with extrinsic

certainty via the Cartesian Cogito. As such, it can appropriately be used in this context.

Although this spoils the purely a priori character of the proof, it introduces a much-

needed factual element into the proof that will prove very useful as we proceed. The

10
Whereas the notion of logical possibility is a modal notion, that of material possibility is not; that
something is modally possible in one world does not entail that it is modally possible across possible
worlds. As such, we can only speak of material possibility in a world rather than across possible worlds.
Nevertheless, since nothing can be materially possible unless it is logically possible, in a like manner,
anything that is materially possible is also logically possible. How these two notions are further related will
emerge as we continue to consider the argument.
inferences realized in premises 1-3 and 3-5 are justified by modus ponens. Premise 6

follows from one of the standard axioms of S5 modal logic according to which if P is

possible then P is necessarily possible. This axiom reflects the intuition that the modal

properties of propositions are essential to them and thus invariant across possible worlds.

Premise 7 simply translates 6 into the idiom of possible worlds, the semantics for modal

logic. Thus, if it is true in any world that something exists, and thus true in every world

that something exists, then something exists in every world.11

Second Stage of the Argument

8. No contingent thing exists in every possible world.

9. Further, it is perfectly conceivable that there exists a world in which no contingent
beings exist.
10. Therefore, it is logically possible that there is a world in which no contingent beings
exist.
11. Since it logically impossible for there to exist a world in which nothing exists, in such
a world, a necessary being must exist.
12. If a necessary being exists in any world, that being exists in every world.
13. Therefore, a necessary being exists in every possible world.
14. The actual world is a possible world.
15. Therefore, a necessary being exists.

At this point, I find it necessary to fill some gaps in Kant’s exposition of the argument in

order to make it logically tight. Much of what I assert in premises 8-15 is anachronistic in

the context of Kant’s philosophy. However, nothing that I say here is obviously contrary

to Kant’s philosophical perspective, at least once it is shorn of its “Copernican”

pretensions. At any rate, proving that a necessary being exists is clearly something that

11
This corresponds to Kant’s claim in the Beweisgrund that it is absolutely impossible for nothing to exist:
see Beweisgrund, op. cit., 71. Thus, despite the fact that a possible world in which nothing exists is
formally possible, hence logically possible so far forth, it is not after all logically possible, since its
possibility is excluded by 7 above. It does not follow from this, of course, that there is some particular
thing that exists in every possible world, only that something or other (not necessarily the same thing)
exists in each world. Thus, as Bede Rundle argues, the foregoing claim would be true even if there were
only contingent beings, so long as at least one such being existed in each world. So it is a mistake (of
which Kant may be guilty) to conclude straightaway to the existence of a necessary being from the claim
that necessarily, something exists. I attempt to repair this defect in the next stage of the argument.
Kant has to argue if his proof is to be completed. This is my attempt to accomplish that

end.

By definition, a contingent being is one that exists in some worlds and not others,

such that it exists in at least one world and fails to exist in at least one world. As such, no

contingent being exists in every possible world. At the same time, it seems perfectly

conceivable that there exist a world W in which no contingent beings exist. At the very

least, there is no contradiction in this supposition, so it is formally possible and thus

logically possible so far forth. Further, in the case of the non-existence of a contingent

thing, it seems that formal possibility will be sufficient for logical possibility, since the

only thing required for material possibility in this case is the lack of the necessary

conditions required in order for a contingent thing to exist. This is a condition that could

conceivably obtain in any possible world in which that being is constituted as

metaphysically possible but in which the necessary and sufficient conditions for its actual

existence do not obtain at some particular time T.12

Of course, such a world would not be conceivable if contingent beings were the

only beings that could conceivably exist, since if that were the case then in W nothing

would exist at all, and this as we have seen is impossible. However, there is another

conceivable kind of being, namely a non-contingent or necessary being, one that exists in

all possible worlds. If such a being were to exist in W, then something would exist in W

even though there were no contingent beings in W. Thus, since it is apparently coherently

conceivable that there exists a world W in which no contingent beings exist, and there is

12
By the same token, and for the same reason, the inconceivability of the non-existence of something is
sufficient to constitute it as necessary and thus as existent in all possible worlds. Contra Kant, this shows
that the classical ontological argument is immune to his criticism. On this point, see the companion piece
to this essay, “Kant’s Critique of the Ontological Argument: FAIL” also on this website.
no contradiction in supposing this even if something exists in every world so long as a

necessary being exists in W, W is a possible world. If that is so, then a necessary being

exists in that world. However, a necessary being cannot exist in any world without

existing in every possible world. Therefore, that necessary being exists in every possible

world. Now the actual world is a possible world, so if that necessary being exists in every

possible world, it exists in the actual world. Therefore, that necessary being exists in the

actual world, which is to say, that necessary being exists.

Third Stage of the Argument

16. Whatever exists has a ground for its existence in virtue of which it is materially
possible.
17. That ground must either be in the thing itself of in something outside of that thing.
18. Only a necessary being could be the ground of its own existence.
19. It is not materially possible for the ultimate ground of a contingent being to be
another contingent being or series of such beings.
20. Therefore, only a necessary being could be the ultimate ground of a contingent being.

Again, this takes us beyond the limits of Kant’s own discussion. At the same time,

something along these lines seems necessary to complete the full proof, so I have

attempted to supply something to fill the gap. Since I have no Kantian provenance for this

stage of the proof, it remains open to others to find alternate reconstructions of the

argument.

Everything that exists is both formally and materially possible, as is proved by the

fact that it actually exists, i.e. exists as something real or actual in the actual world.

Therefore, the existence of anything in the actual world presupposes that the materially

necessary conditions for its existence have obtained in that world. Such a being will not

only be materially possible, but such that its material possibility has been realized or

actualized as a real, actual, existing thing, event, or state-of-affairs. The notion of

material possibility, then, presupposes that possibility in that sense is parasitic on prior

actuality, and that this prior actuality both grounds the possibility of what exists, but also

realizes it in actuality as something real and actual. Thus, the materially possible

presupposes the actual as a necessary condition for the realization/actualization of its

material possibility as something real, actual, or existent. 13 Thus, where material

possibility is concerned, the PSR is undeniable – nothing can exist unless the necessary

and sufficient conditions for its existence obtain. As realized in time, i.e. in an ordered

temporal sequence, the PSR takes the form of the causal principle, according to which

whatever comes to exist or occur in time has a sufficient explanation in terms of the

operation of some prior thing, event, or state-of-affairs that produces it, makes it happen,

or constitutes it as real, actual, or existing. As such, every event in any ordered sequence

of real, existent, or actual events is an effect, and every effect has a cause.

Kant, of course, would deny that this applies beyond the bound of possible

experience. However, that is only because he was held back from affirming the

deliverance of his own reason by the demands of his “Copernican revolution” in

philosophy, for which Kant gives no compelling arguments and which I believe does

nothing but distort his philosophical vision. I, then, am fully prepared to follow Kant’s

argument where it leads, without regard for his doctrine of the limits of reason. I believe

that argument leads to an affirmation of the existence of God, or at any rate, that of the

only God to which philosophical argument can lead us, the God of the philosophers.

Here’s how the argument proceeds. In the case of a non-contingent, or necessary

being, the ground of its being has to be contained within its own nature in such a manner

13
Kant makes this point a central premise in his Beweisgrund proof. Where material possibility is
concerned, possibility presupposes prior actuality; see the Beweisgrund, op. cit., 71-73.
that it could never come into or go out of existence, even in principle. For example, if the

classical ontological argument is sound, then there is a being for which it is inconceivable

that it not exist, simply as a matter of conceptual analysis. 14 In a like manner, if there is a

being in which essence and existence are identical, then it will the essence of that being

to exist, in which case its existence will once again be accounted for by that fact. 15 Again,

we may be able to conceive of a being such that the mere logical possibility of its

existence entails its necessary, and hence its actual existence. 16 In any of these ways,

then, the existence of a necessary being would be grounded in the notion of its essence or

nature, and thus capable of being exhibited by conceptual analysis, even if the proof for

the existence of such a being, considered as such, may require a premise known a

posteriori.

In the case of a contingent being, one by nature indifferent to existence, the

ground of such a being must always lie outside its nature, and be such that it obtains prior

to and independently of that which depends upon it. This will hold regardless of whether

we are talking about a temporal or a non-temporal ground. Nor does it seem possible, on

reflection, that we could account for the materially possibility, and thus the existence, of

any contingent being, solely in terms of the grounds provided by other contingent beings.

Since each contingent being requires that the grounds of its material possibility be

realized prior to its existence, such a series cannot end in a first member, since in that

case some contingent thing would have come to exist in circumstances in which the

14
Kant, of course, rejects the classical ontological argument, but I find no reason to accept his arguments.
See “Kant’s Critique of the Ontological Argument: FAIL” also on this website.
15
I argued this in Duncan (2008) op. cit., chapters 7-10, without using the ontological argument as a stand-
alone argument for God’s existence.
16
This is Plantinga’s ontological argument, the father of many versions of the argument.
materially necessary grounds have not obtained and which are thus sufficient to prevent it

from coming into existence.

At this point, of course, David Hume will object. How can we exclude the logical

possibility that something comes into existence in W from nothing (i.e. from no set of

independently constituted antecedent conditions specifiable in principle) for no cause or

reason whatsoever? It seems perfectly conceivable, i.e. coherently imaginable, that such a

state-of-affairs could obtain. I simply close my eyes and imagine something, e.g. an

aardvark, just popping into existence out of thin air into an empty region of space.

Further, the arguments against this supposition based on the principle of sufficient reason

are all either circular or question begging. There is thus no incoherence in supposing that

it is both possible that something exist in W and yet that nothing exist in W, and indeed,

that nothing ever exist in W.

In Hume’s time, the Principle of Sufficient Reason (PSR) was treated, by Clarke,

Leibniz, and even (perhaps especially) Kant as a self-evident, a priori principle known by

pure reason, even if indemonstrable. We cannot blame Hume for having attacked it this

form, defying anyone to demonstrate that its denial was somehow self-contradictory.

However, even if Hume is correct in his critique of the arguments for the PSR received

by his contemporaries, we need only note in this regard that, as we have seen, mere lack

of contradiction, i.e. formal possibility, is not sufficient for logical possibility in the full

and proper sense. More is needed to explain why one contingent thing exists and another,

equally coherently conceivable, fails to exist in any given possible world. Indeed, given

the definition of contingent being presented above, there is an evident contradiction in the

idea that a contingent being could just pop into existence for no cause or reason
whatsoever. A contingent being is one that is by nature indifferent to existence, and thus

can only exist when some set of independently constituted antecedent conditions

specifiable in principle obtain, such that if those conditions exist or obtain then that being

will exist, and if they fail to exist that thing’s existence will be excluded or prevented by

that very fact. What Hume envisages is that something that can only exist on the

supposition that prior conditions necessary and sufficient for its existence obtain comes to

exist anyway despite the fact that those conditions have not been met, itself a sufficient

condition for the prevention of that thing’s existence. That is a contradiction, surely.

Hume will naturally deny this, claiming that the account of contingent being that I

have given begs the question somehow against his supposition. While I admit that it does

have the logical consequence that Hume’s scenario is not logically possible after all, I do

not see how this is sufficient to entail that I am begging the question against Hume.

Perhaps Hume would want to evade this by proposing some alternate analysis of what a

contingent being is that blocks my argument. For example, he might say that, on his

view, a contingent being is simply one that exists at one time and not at another – after

all, he might say, this is a logical consequence of your “definition” of contingent being –

and then defy me to prove that such a being, so defined, cannot pop into existence from

nothing for no cause or reason whatsoever. However, I am not sure that Hume really

deserves a response on this score.

Suppose that someone claims it is logically possible that the interior angles of a

triangle not add up to 180 degrees. I respond by giving him Euclid’s proof for this claim.

In response, my opponent proposes that I am begging the question against him. Of

course, if we are going to define a triangle as plane figure with three sides, then there is a
contradiction in denying the conclusion of Euclid’s proof. However, he says, “I don’t

define “triangle’ in that way. Instead, I define “triangle” as “a plane figure with at least

one side” – after all, that is a logical consequence of your definition of triangle, right?”

He then defies me to prove that the interior angles of a triangle, so defined, necessarily

add up to 180 degrees. The fact that I cannot do this does not make me any less confident

that what Euclid says is correct. The same holds in the case of the PSR, which is a

principle that applies, not specifically to logical possibility, but instead directly to

material possibility. At any rate, to define a contingent being as one that exists at one

time and not another is hardly adequate, since it overlooks the possibility that something

could be contingent yet exist at all times, something that Hume himself affirms and

exploits in his writings. If that is so, then there is more to the concept of contingent being

that what my hypothetical Hume is willing to concede in this hypothetical context.

Further reflection on this point, I believe, will lead “Hume” (or his modern counterparts)

to admit the justice of the analysis I have presented above.

Nor is Hume’s thought experiment sufficient to prove even the conceivability of

the claim that something could come into existence out of nothing for no cause or reason

whatsoever. As Jonathan Edwards would have pointed out (and more recently Bede

Rundle has pointed out) it is impossible to conceive of a state of affairs in which nothing

exists, if what this means is to imagine such a scenario. 17 The best we can do, as Hume

does, is to imagine something popping into existence in a previously empty space. Even

this thought experiment, then, presupposes something given prior to and independently of

that which is supposed to pop into existence out of nothing. The same holds for modern

17
Jonathan Edwards, “Of Being,” Jonathan Edwards Selections, ed. by Clarence Faust and Thomas H.
Johnson, New York, Hill and Wang, 1962, 18-23, and Bede Rundle, “Why is there Something rather than
Nothing?”, Oxford, Clarendon Press, 2004, 111, 116-117.
cosmologies that claim to show how the universe comes into existence out of “nothing.”

Proponents of these views never really mean this – there always turns out to be

something (e.g. a “quantum vacuum”) pre-existing the universe that is causally necessary

for producing it in accordance the operation of some indeterministic physical law, which

incidentally also pre-exists the universe as a necessary condition for its possibility. In

Hume’s case, his claim to be able to imagine something popping into existence from

nothing for no cause or reason whatsoever is mere stipulation. Given what we imagine in

this scenario, we can at best infer that it is apparently coherently imaginable that some

material thing pops into existence without any observable cause or reason of the same (or

the usual) sort. That is far from proving that it popped into existence from literally

nothing for no cause or reason whatsoever, or that in imaging this I have succeeded in so

imagining this.18 There is, in fact, not enough data given in the scenario to assure us that

it is even conceivable for something to come into existence without a cause. So far as the

imaginative episode itself is concerned, we of course know that the cause of the

imaginative popping of the mental image into Hume’s consciousness is Hume’s own will,

which intentionally summons the image into the intentional field of his imagination. As

such, even his own supposed demonstration of the possibility of something popping into

existence from nothing for no cause or reason whatever belies what he supposes it to

show. Hume will undoubtedly claim that I am begging the question against him, but I

prefer to think that I am simply holding him to the impossibly high standards of

demonstration that he requires of those who hold views contrary to his own. The claim

that some contingent thing could pop into existence out of nothing for no cause or reason

18
See Virgil C. Aldrich, “Image Mongering and Image Management,” (1962) Philosophy and
Phenomenological Research 23 (1):51-61
whatever seems to me obviously false and self-contradictory, something to be accepted

only on the most compelling arguments, which Hume has failed to give and, in the nature

of the case, could not possibly give. I have no qualms in rejecting his suppositions out of

hand.

Nor will it do, as Hume does elsewhere, to postulate that there is no first member

to the series, but that each member of the series of contingent beings is the product of an

actually infinite set of prior contingent beings. I have presented the argument for this

claim at length in another place; let me just summarize it here.19

Pick any event in the series – call that event E. According to this suggestion, an

actually infinite number of antecedents, each of which was causally necessary for its

existence, preceded E. Since E is a contingent being, and did not always exist, there was a

time prior to its existence at which it did not exist; call this time T. At T, some prior event

upon which E depended for its existence was occurring; call this event E-n. If E-n is to

causally produce E, then it must be possible to reduce the number of events linking E-n to

E to zero by progressively producing them one at a time, with the result that E is finally

produced and that will be the case only if the number of such events is a finite number.

There is a problem with this. E-n is stipulated to be part of an actually infinite series of

such events occurring prior to E. At whatever point in the series we have reached at T,

only a subset of that actually infinite series will have been realized in actually. According

to the standard account of infinity, however, no matter how many members we remove

from an actually infinite set, or how many sub-tasks we complete in a supertask requiring

that we perform an actually infinite set of sub-tasks in order to complete it, an actually

infinite number of members remain, or remain to be completed. In that case, there will be
19
See “Why there can’t be a Self-Explanatory Series of Infinite Past Events” also on this website.
no less members in the proposed series left to be realized at T than there were before T,

so that an actually infinite number of such events will still remain to be produced prior to

the production of E. Because of this, E-n will be no closer to realizing E than was any

earlier member of that series and the envisaged act of production will not be possible

even in principle. E, then, cannot be produced in this way and thus cannot exist in the

envisaged scenario. We can iterate this same argument for each and every member of the

series. This means that the same will apply to every member of that supposed actually

infinite series, with the result that none of the members of that series could exist, since

the materially necessary conditions for their existence could not possibly obtain. Since

none of the members of that set can exist in the envisaged circumstances, neither can the

series as a whole, unless that series has an external cause, not a member of the series, that

produces it as a completed whole in a single, timeless act of creation.

However, since we know with certainty that contingent beings exist, since each of

us is one such, we know that the materially necessary conditions for the existence of such

beings have in fact been met. If they cannot be met on the supposition that they have only

contingent grounds, then we must have recourse to the existence of a necessary being,

which we have already proven to exist on independent grounds, as the First Cause of all

things. Only such a being could provide the ultimate ground for the existence of all

contingent beings.

To this it may be objected that we have not proved that there is only one such

being; all this in good time, when we get back to Kant’s exposition in the Beweisgrund.

For now, we need only note that this necessary being as First Cause is as yet merely a

theoretical entity about which we know nothing. In this context, considerations of

parsimony and simplicity dictate that we not multiply theoretical entities beyond

necessity. We shall then proceed on the assumption that there is only one such being,

with the proviso that if this should prove an inadequate hypothesis, we are prepared to

revise this judgment. With this proviso in mind, let us now conclude the argument.

Fourth and Final Stage of the Argument

21. Only an ens realissimum could be ground of all material possibility.

22. There can be at most one such being.
23. Therefore, an ens realissimum exists.
24. Thus, the God of the philosopher exists as the ens realissimum.

Our long excursus has ended and we are now ready to rejoin the main line of Kant’s

argument in the Beweisgrund. Leibniz had defined God as the ens realissimum – the

“most real being.” Kant’s 1762 proof is tailored to this conception of God, identifying

God as the ultimate ground of all possibility, by which I here take him to mean material

possibility. In this case, we are relying on another version of the principle of sufficient

reason, one to which Descartes also subscribes and upon which he bases his primary

argument for God’s existence: there cannot be more reality in the effect than there is in

its cause. Something cannot come from nothing, but that is precisely what would be the

case if there were more reality, i.e. perfection, in the effect than in its cause. On that

supposition, there could be no explanation for the overplus of being. So if we are to

follow our hypothesis that there is but one necessary being serving as the ultimate ground

for all contingent beings (such as I know myself to be), we must hold that this being

possesses all the perfections that creatures possess either formally or eminently. If the

ground of all possibility possesses a perfection formally, it does so by instantiating the

highest degree of a perfection that comes in degrees and can be exemplified to a greater

or lesser degree among its possessors. If the ground of all possibility possesses a
perfection eminently, then it does so by possessing some greater perfection in virtue of

which that being is capable of causing something to have that perfection.20

The ground of all possibility, then, possesses all perfections in a manner

appropriate to its nature and we should attribute them to that being on the same grounds

that we attribute features to any theoretical entity. According to Kant, the necessary being

conceived of in this way is necessarily singular.21 Considered merely as the God of the

philosophers, and thus as a theoretical entity, we can defend Kant’s claim in the

following manner. If there were two such beings, one would either have to be dependent

on the other or independent of it. In the first case, its material possibility would have to

be explained in terms of a ground more ultimate than itself and thus is would not be the

ultimate ground of whatever it, in turn, makes materially possible and thus would not be

the necessary being we have been describing. On the second option, either one would

have to be more powerful than the other or they would have to be equally powerful. If

one is more powerful than the other, then once again the weaker one would not be the

ultimate ground of whatever it makes materially possible, since its operation could be

overridden by the other, more powerful one and is thus the ground of those possibilities

only by the sufferance of the other. If those two beings were equally powerful, then either

they could interfere with each other’s operation or they could not. In the first case, neither

would be the ultimate ground of anything material possible through their operation, since

each would depend on the sufferance/noninterference of the other in order to actualize

those possibilities. However, if we suppose that these beings are so equally balanced in
20
As Aquinas says, God is formally wise and eminently a stone. As to formal perfection, Kant clearly
endorses the traditional, neo-Platonic and later, Christian metaphysics of participation in his discussion of
the categories of quality in the Table of Categories (see Kemp Smith, op. cit. 113-15), a passage usually
passed over in silence by commentators. See also the discussion in Wood (1978), op. cit., 25-34.
21
See the Beweisgrund, 79 and 81 (page 80 contains the German original of which 79 and 81 are the
translation.) However, Kant’s argument is, as usual, unpersuasive as given.
power that neither was capable of interfering with the operation of the other in either its

exercise or its outcome, we would have no reason for regarding them as two from the

theoretical point of view. In that case, considerations of parsimony and simplicity

behoove us to posit only one such being, a necessary being that is the ground of all

possibility. Since we have seen no reason why we ought not to do this, nor derive any

advantage from doing otherwise, we do best to simply posit a single necessary being

serving as the ultimate ground of all material possibility. That, however, is precisely to

posit that being as the ens realissimum, containing formally or eminently all the

perfections we see instantiated in or exemplified by creatures. This, according to Kant, is

the God of the philosophers.

We can thus conclude the proof as follows. Since I exist, then at least one material

possibility has been realized or actualized in the actual world in the form of something

that really or actually exists in that world. Since the real existence of anything requires

that all of the materially necessary conditions for its existence obtain, that something

really or actually exists in the actual world entails that those materially necessary

conditions exist, occur, or obtain in that world as well. However, as we have seen, the

existence of any such thing requires the operation of an ens realissimum serving as the

ultimate ground of all material possibility, which in turn can only be a necessary being.

Therefore, from the existence of anything at all we can infer that such a being exists. This

ens realissimum is the God of the philosophers. Thus, if anything exists, then God (so

described) exists. So, after all, there is a God.

This is the conclusion to which Kant was aiming in 1762 and at which he would

have arrived had the baleful influence of Hume not led Kant to initiate his quixotic and
wrong-headed “Copernican Revolution” in philosophy. Even so, Kant was convinced that

this was the sole basis for a proof of God’s existence. Like many philosophers, he

contends that he, and he alone, has proven this. On my view, Kant was surely wrong

about this. There are many possible grounds for belief in God, and numerous different

demonstrations of the existence of the God of the philosophers that can be established on

their basis. However, that does not prevent Kant’s argument from being one among the

many possible arguments for God’s existence, or without value simply because it is only

one among many. I recommend this argument to the study of other philosophers, for it is

a rich source of reflection on the issues it raises, and bids fair to shed further light on the

question concerning just what we are asserting when we say “God exists.”