Immanuel Kant, in his Critique of Pure Reason, offered some of the most influential criticisms of the classical proofs of God's existence. As we shall see, in many respects Kant builds his critique of dogmatic theology on the earlier arguments of Hume, whom Kant credited with "waking" him from his "dogmatic slumber". The most original aspect of Kant's criticism of the cosmological argument is his attempt to establish a linkage between it and the so-called ontological argument. Consequently, we must begin by examining the ontological proof and Kant's critique of it.
The ontological argument was first developed by Anselm of Canterbury, a 12th century cleric and philosopher. Philosophers and scholars have found several distinct threads in Anselm's work. A simplified version of one of Anselm's proofs was very popular in the early modern period (17th and 18th centuries), when it was used by Descartes, Spinoza and Leibniz, to mention a few. The simplified argument goes something like this:
Consider the following analogy. To ask how many planets there are, and to ask what is the square of three, are to ask two separate questions, even if the answers to the two questions are identical. Even though the answer is the number nine in both cases, we are approaching the number nine from two different perspectives, looking at in under two different guises (its number-of-planets guise, and its square-of-three guise). Similarly, even if God's essence and existence are the same thing, for us there are two independent questions: what is God's essence, and does God exist? The simplified ontological argument tries improperly to conflate these two.
Kant's second critique of the ontological proof is somewhat more original, but I think that Aquinas would agree with it as well. Kant argues that existence is not a property of the existing thing, and so existence cannot be included in the essence or nature of a thing. Kant makes a very compelling case for this. Suppose that existence were a property. Suppose I describe a possible thing, like a possible paper-machie sculpture. I describe its shape, size and composition in absolutely complete detail. Now I ask, does a sculpture of exactly this kind really exist? Suppose you answer that sculpture S fits my description to a "t". If existence were a property, there would be a property that S possesses that was not included in my description, namely, existence. This means that it is always impossible to give a complete description of a merely possible thing, since the description would always be missing one property that any actual thing must possess.
It seems clear that Kant is right. Existence is not a property of a thing that exists. Instead, existence is an act or eventuality by which the essence or nature of a thing is realized in the world. It is never a part of that nature, or else the property of existence (the property of being realized by an act of existence) would have to be realized by a further act of existence, which is absurd, leading to an infinite regress of acts of existence.
This Kantian thesis (that existence is not a property) may appear to be inconsistent with al-Farabi's claim that God's essence is identical to His existence, but this is not so. Al-Farabi claimed that God's existence (that is the act by which God exists, and not the property of existence-in-general) is identical to (not included in) God's essence. Descartes and Leibniz transformed al-Farabi's tenet into something comprehensible, but obviously impossible, namely, that the property of existence is contained in God's essence.
It is very clear that Kant believed that the cosmological argument depends on the ontological argument. Kant was confident that, since the ontological argument can be shown to fail, the same fate must then befall the cosmological argument. However, it is much less clear why Kant believed this dependence to hold. It seems that he has several, independent lines of argument in mind.
The key connection between the two arguments is the idea of necessary existence. Since God's existence is included in His nature, according to the first premise of the ontological argument, it is clear that we can infer, not only that God exists, but that He exists necessarily. Similarly, the primary conclusion of the cosmological argument is that something exists necessarily, which necessary being we can identify with God. Thus, the same concept appears in both arguments. However, this fact alone is not enough to show that the cosmological argument depends on the ontological argument.
One possible argument for this dependency that seems to be implicit in Kant's discussion is this: the cosmological argument presupposes that necessary existence is at least possible, since if it is not possible, it cannot be actual. If necessary existence is possible, then there would be a possible nature that includes existence as one of its essential features. This would make existence a property, contrary to what Kant established in his discussion of the ontological argument.
There are at least two things wrong with this argument. First, the cosmological argument does not depend on first assuming that necessary existence is possible. Instead, the argument tries to show that necessary existence is actual, from which we can infer that it must be possible. Aquinas admits that we cannot, by examining the idea of necessary existence, or the idea of a being whose existence and essence are identical, establish to ourselves that such a being is possible. Instead, by tracing the series of causes back to their primordial limit, we are forced to postulate that such a being exists.
Second, the idea of necessary existence is not the same thing as the idea of a being whose nature or essence includes existence. A being exists necessarily if it is impossible for that being not to exist. This need not involve the inclusion of a property of existence in the nature of the thing in question. Thus, one can accept that there is a necessarily existent being without accepting either premise of the simplified ontological proof.
There is a second reason that Kant gives for thinking that the cosmological proof depends on the ontological argument. This argument depends on looking at the closure of the cosmological proof. A defender of the cosmological argument must offer some reason for thinking that the necessary being is God, that is, for thinking that any necessary being must be a supreme being or a being of maximum reality (to use Kant's terminology). Further, the defender of the cosmological proof must contend that there can be only one supreme being, if she is to reach the conclusion of monotheism. Now, Kant wants to argue that the defender of the cosmological argument must admit that any supreme being must be a necessary being. Suppose that it were possible for something to be a supreme being but not a necessary being. Since the cosmological argument is supposed to prove that there is a necessary being, a being that exists in every possible situation, then in this imagined situation we have both a necessary being and a supreme being that is not necessary. Since every necessary being is a supreme being, this would mean that we had two supreme beings, which is supposed to be impossible
Consequently, to be a necessary being and to be a supreme being are necessarily equivalent: anything of one sort must also be of the other. This means that the defender of the cosmological argument must admit that the very idea of a supreme being includes that of being a necessary being. Since whatever exists necessarily exists in fact, this means that the idea of a supreme being includes the property of existence. Thus, the defender of the cosmological argument is forced to accept the premises of the ontological proof.
However, this argument of Kant's involves confusing existence and necessary existence. Kant is quite right in asserting that existence is not a property. However, he has not shown that necessary existence (or its contrary, contingent existence) is not a property. To say that something exists necessarily is to qualify its nature or essence. An existing 100 dollars is the same thing as 100 dollars: a necessarily existing 100 dollars is another thing altogether.
In the twentieth century, a number of philosophers (including Norman Malcolm, Charles Hartshorne, and Alvin Plantinga) have revived a second version of Anselm's proof, called the modal version of the ontological argument. This proof goes like this:
A defender of the cosmological argument is committed to accepting that all of the premises (and, of course, the conclusion as well) are true (with the possible exception of the S5 axiom). However, this does not mean that the defender of the cosmological argument must accept that the modal-ontological proof is a good argument. For example, Aquinas would argue that we cannot know premise 1 to be true (that God is possible) until after we have established (through the cosmological argument) that God actually exists. Consequently, the modal-ontological argument should not persuade anyone of God's existence, since we should accept the first premise only after we have accepted (on independent grounds) the truth of the conclusion.
Kant argues that the idea of necessary existence is incomprehensible. We cannot prove that such a thing is possible, since we have no effective criterion of possibility to apply. However, this is irrelevant to the cosmological argument, since as we have seen, it does not depend on first establishing the possibility of necessary existence.
Kant argues that we have no criterion of existence except that of connection to actual sensation. Since God cannot perceived, the category of existence cannot apply to him. This argument seems to prove too much. If it were sound, we could never know that any observable thing exists, which would make much of contemporary science (particle physics, physical chemistry, astronomy, cognitive psychology) impossible.
Kant insists on identifying necessity with the constraints of our concepts. Like Hume, he identifies impossibility with inconceivability. According to Kant, if this were not so, we would have no criterion (no effective test) of impossibility. Knowledge of impossibility would be thus be impossible.
This position is self-defeating -- see my section on Kant in "A New Look...", section 8.7. We have to ask Hume and Kant, is the principle that all necessity is based on inconceivability supposed to be necessary or contingent? If it is contingent, where is the empirical evidence that it is true? If it is necessary, how can they explain the fact that the opposite view (one that distinguishes between impossibility and inconceivability) is itself conceivable?
Kant argues as follows. Since necessary existence is incomprehensible, we have no knowledge of the properties of a necessary being. For all we know, a finite being could be necessary. ( p. 498)
However, even if we grant that we can have no knowledge of the positive properties of a necessary being, it does not follow that we cannot know what properties are incompatible with necessary being. We do know many finite beings, including physical facts. We know these to be contingent -- hence, we can know that the necessary being is not physical, does not have a beginning, is not composed of separable parts, etc. Nothing Kant has said challenges these negative facts. In addition, we know that the first cause is the cause of all contingent facts, and this knowledge may also give us a limited degree of positive knowledge of the first cause. This additional knowledge is not derived by examining our idea of necessary existence, but rather by examining the effects of the first cause.
Throughout the Critique of Pure Reason, Kant argues for a distinction between the noumenal and the phenomenal world. The noumenal world is the world of reality, the world of things as they are "in themselves". The phenomenal world is the world of appearance, the world of things as they are "for us". All of our knowledge is limited to the phenomenal world. This world is constructed by the human mind, working on the raw materials of sensation. The orders of time and space, the structure of causation and of enduring objects, all of these are put in the phenomenal world by our own cognitive faculties (the "understanding"). This theory is called by Kant "transcendental idealism"> Kant has a number of arguments for transcendental idealism:
However, it seems that Kant cannot consistently limit the realm of the phenomenal world to things that we can actually observe. In many cases, we must trace the causes of observable events to the activities of unobservable forces and processes. These scientifically accessible unobservables are as much a part of the phenomenal world as are the ordinary objects of human sense perception. The first cause is reached by an extrapolation of our ordinary scientific practice. Hence, the first cause seems to be an essential part of the phenomenal world, even if it is unobservable.
Kant is one of the first philosophers to offer an explanation of why we mistakenly think we know that God exists. In this respect, Kant is a precursor of such thinkers as Feuerbach, Nietzsche, Freud, and Durkheim, all of whom offered psychological or sociological explanations for the mistaken beliefs at the root of religion.
Kant describes the inference to God's existence as a "natural illusion of reason". He explains this illusion in two stages. First, in thinking about the world, we always begin with an archetypal idea of a complete, perfect being. We then whittle this archetype down to size by adding various negations or limitations. Kant think that this process explains our belief in the law of excluded middle: for every thing x and every propery y, either x has y or x does not have y. This logical axiom is true because we always begin, in thought, with an ideal object that possesses all positive properties, and form our ideas of particular things by a process of negation. Belief in God begins by mistakenly reifing this archetypal idea, treating it as a real, particular thing -- the thing of maximum reality.
Kant's second stage depends on his theory of transcendental idealism. Kant believes that is the human mind that actively organizes sense experience into the orderly and regular world of appearance. We recognize, in a foggy and confused way, that intelligence is somehow involved in the construction of our world. Instead of realizing that we ourselves are this intelligence, we mistakenly personify the most-perfect being, the being of maximum reality, ending up with a belief in a personal God.
If it were true that we have no knowledge of God, then of course it would be useful to find an explanation of the fact that so many people believe that God exists. However, if we really do have knowledge of God's existence, then no such deflationary explanation is needed, or even germane.
Kant believed that the correct use of the idea of a First Cause is a heuristic or regulative use. Instead of positing the actual existence of such a first cause, we should instead assume, but only as a working hypothesis, that there is an underlying unity to the phenomenal world. Moreover, we must never be content with any stage of our reconstruction of the causes of things: we must always push the causal inquiry still further.