Kant on the Ontological Proof
Kant’s refutation of the ontological proof of the existence of god is almost as familiar to Philosophy 101 grads as the Cartesian Cogito. "Existence is not a predicate" and "One hundred real thalers are not worth one penny more than one hundred imaginary thalers" have long earned their place in the philosophical Hall of Fame. Nevertheless, despite the fact that Kant’s argument expresses a strong intuition that only a very few, perhaps rather foolish philosophers have been tempted to discount, the specifics of his refutation may not withstand the kind of close examination we have given the ontological proof itself.
I. In a deviation from the ontological proof as we have come to know it from Descartes and Anselm, Kant does not begin with the concept of a greatest or best entity. Rather he skips directly to talk about an absolutely necessary entity (absolutnotwendiges Wesen – presumably by "Wesen," Kant means "entity." "Essence" in any sense different from "entity" would be inappropriate). What he skips is how we are supposed to arrive at the notion of an absolutely necessary entity, by what train of reasoning we come up with this fairly strange concept. The starting point for most philosophers who have endorsed one or another version of the ontological proof is to conceive of some entity that is the greatest or the best or has all the perfections, and to thereby concede that existing is one of this entity’s perfections or good qualities. It is noteworthy that the Descartes and Anselm versions of the ontological proof do not employ the concept of necessity. They rely only on whether a greatest or best or infinite and therefore existing entity can be conceived (cogitare). Kant does not distinguish this classical proof from the undoubtedly more baroque Leibniz version according to which, if a necessary entity is possible, then it is necessary and so it exists. In fact, introducing the concepts of possibility and necessity into the discussion introduces a new (and, shall we say, unnecessary?) set of problems into the course of the proof. Consider the concept of a necessary entity. We are tempted to ask, "Necessary for what" If I, for one, were asked to think about a necessary entity, I would think of something I needed in order to perform some task or to fulfill some project. A hammer is (more or less) necessary to pound in a nail. But this is not the sense of "necessary" Kant has in mind. The absolutely necessary entity is not necessary for anything or at least not for anything specific. As the object of the ontological proof, it is necessary because the proposition asserting that it does not exist would be false on logical grounds alone. Supposedly that proposition would contradict some logical law. But in that case, "necessary" in the phrase "absolutely necessary entity" and "necessary" in the phrase "tools necessary to complete the job" don't look like the same term at all. They don't seem to share much more than their spelling. (By the way, this entity is not just necessary, it is absolutely necessary. What is the difference between a necessary entity and an absolutely necessary entity? Neither Kant nor Leibniz are very helpful on the matter.)
The term "necessary" in "absolutely necessary entity" does appear to be related in meaning to "necessarily true" as signifying a property of propositions. The idea is that some entities are necessary because they are the subject (in a loose sense of "subject": "talked about," "referred to") of necessarily true propositions. But that sense of "necessary" is unclear. For one thing, the logically tautological proposition,
(1) Hammer A is hammer A,
where "hammer A" refers to a single object and where "Hammer A" and "hammer A" refer to the same object, does not entail that hammer A is necessary for any given project such as pounding in nails (and it isn't: Hammer B will do). So the sense of "necessary" as a property of objects referred to in necessarily true propositions is not the above-referenced sense of "necessary" as "necessary for...." Leibniz's comments appear to qualify necessary entities that are referred to in necessarily true propositions as necessarily existing entities. That is simply false as Kant's first objection to the ontological proof shows. Kant cogently remarks that necessity, as a property of propositions (or judgments, Urteile), does not entail that the objects referred to in the propositions exist. The necessary proposition,
(2) A triangle has three angles,
does not entail that triangles exist. In the same way statements about the present day King of France do not imply that the present day King of France exists.
But there are more problems than that. Outside of the familiar uses of "necessity" in the sense that something is necessary to achieve a certain purpose or it is necessary because the king commands that it be so, necessity does not sit very well as a property of objects even if it were to have a well defined role as a property of propositions.
Consider possibility, which is also sometimes defined as a property of propositions. And in this case there may be a relation between possible propositions and the objects (or some of the objects) referred to in those propositions. For the sake of clarity, let us say that "possible proposition" really means "possibly true proposition." Then the relation would be: if a proposition is possible, it corresponds to a possibly obtaining state of affairs. Some entities named or described in those propositions could be termed "possible entities," which is shorthand for "possibly existing entities." When we call an entity "possible," in this usage, we really mean that it is a possibly existing entity. Galatea doesn't exist but she could exist. As a property of propositions, necessity can be defined as the contradictory of possibility (in a modal system, for example), but the objects referred to in a necessary (i.e. necessarily true) proposition are not thereby necessarily existing, i.e. necessary entities. "Snow is white if and only if snow is white" is a necessarily true proposition, but, even though any state of affairs, if such there be, corresponding to the proposition necessarily obtains, snow is not a necessary or necessarily existing entity. "Galatea is a babe if and only Galatea is a babe," does not entail that Galatea is a necessary entity.
There is a problem as to whether there are indeed such things as necessary, i.e. necessarily existing, entities at all. One might say that the concept of a necessary entity in anything but the natural language sense of usefulness or command is meaningless. At the very least one cannot argue that there are such things as necessary entities on the assumption that there are validly necessarily true propositions. With an eye toward Leibniz we might say that necessary entities are impossible.
Certain modern schoolmen (Calling them "logicians" would give them too much credit) construe necessary entities in terms of modal logical systems of possibility and necessity mixed in with a kind of homemade semantics of those systems. They fail. In the first place, while they recognize that a concept of a necessarily existing entity depends on an understanding of necessity that is broader than the necessity of logical theorems, they fail to define such an understanding. Secondly, in their search for a necessarily existing entity, i.e. an entity that exists in every possible world, they use a concept of existence that is nothing more than "not impossible." That is, an object exists in a possible world as long as its existence is not impossible (Because of the lack of a coherent understanding of non-logical necessity, the concept of an impossible object, outside of an object the assertion of whose existence would entail a logical contradiction, also remains undefined). But this weak, almost paraplegic, concept of existence is surely unworthy of so grand a personage as god.
None of this vitiates Kant’s argument. It just shows how introducing necessity and possibility into the ontological argument just forces extra angels onto that pinhead.
II. Kant marshals his famous thaler example to support his statement that existence is not a real predicate. Yet, the example is not entirely felicitous. There is indeed a difference between one hundred real and one hundred imaginary thalers. You can (or could) spend a hundred real thalers. And, assuming, there is a one-to-one correspondence between real money and printed bills (as there may have been in the eighteenth century), existing could be just one among many properties you can take away from the hundred thalers without changing their value. One hundred red thalers and one hundred green thalers would have the same value as long as red and green ink were permitted currency colors. But both "red" and "green" are predicates. Or else, one hundred thalers touched by Frederick the Great and one hundred thalers not touched by Frederick the Great have the same value. Of course, the thalers with the different properties are not identical. But the real and imaginary thalers are not identical either.
But in what sense do we really qualify one hundred thalers when we say they are real? What Kant should say is that "One hundred thalers exists" is not a well-formed sentence. It is not meaningful in itself. If someone were to utter the sentence we would be inclined to say, "What do you mean" In fact, it is more correct to say not that "exists" is not a predicate, but that it is an adverb-like particle that qualifies at least some predicates such as to further differentiate the objects to which those predicates properly apply. It is like the adverbial particle "super-" in the sense that "Harry is super-efficient" or "Steam power is superannuated." To just say "Harry is super-somethng" is to invite the same question, "What do you mean" Or else, "exists" is a shorthand expression for a more complete or more meaningful expression. In this sense "The hundred thalers in the Baron's ledger exists" is a shorthand for "The hundred thalers in the Baron's ledger was printed by the Prussian Central Bank and authorized by the Prussian government as legal tender, and its provenance can be traced from the Baron's ledger or the Baron's safe to an authorizing and printing act on the part of the Central Bank." To say that the one hundred thalers in the Baron's ledger does not exist is to say that all or part of the preceding sentence is false. The first embedded clause in the latter sentence is useful in legal proceedings if one suspects fraud or counterfeit on the part of the Baron.
Kant says something like "exists" is an adverb-like particle that qualifies other predicates that belong to a certain class such that they can be qualified by "exists," when he defines existence as a property of objects of sense: "Whatever and however much our concept of an object contains, still we must proceed from that concept in order to attribute existence (Existenz). For objects of the senses this occurs through the relation (Zusammenhang) with some one of my perceptions according to empirical laws; for objects of pure thought, however, we have no way of knowing their existence (Dasein)….(p. 535)"
The problem with the ontological proof is that it uses an adverb-like qualifier as if it were a completely meaningful predicative expression. It is like saying "God is super something" or, since a like sentence is in fact also part of the ontological proof, it is like saying "God is -ly" in English. The reason "God exists" looks well-formed to some people is that they confuse two different ways in which "god exists" is used. Their mental picture when they say "God exists" is one of spatio-temporal location with mass and form - the old man sitting in the clouds. The actual ontological proof eschews that meaning of "God exists" as insufficient (and subject to empirical falsification). The ontological proof assumes that "god exists" is meaningful but unqualifiable by any such mental picture or explanatory predicate. It is on a par with saying that "God is -ly" is completely meaningful in itself and a well-formed sentence on a par with "God is boringly obtuse."
Predicates from natural language contain so many ambiguities and layers of meaning that their predicative function needs a great deal of clarification before it can be used in a philosophical argument. As regards existence, the natural language term, "exists," is always a shorthand whose expanded meaning is usually fairly clear from the context. Any use of "exists" that refuses to provide some exposition of what it is a shorthand for, is incomprehensible and meaningless. And that is what the ontological proof does. Perforce it uses "exists" (expecially when the phrase "necessarily existing entity" is in play) in a way that is so broad and general that the term itself is deprived of meaning. It is like insisting that one hundred imaginary thalers exist or that a griberschlag exists tout court without qualifying what we mean. As Bacon observed, " a poor and unskilful code of words incredibly obstructs the understanding."
III. Kant’s refutation is tightly connected with his faculty theory. It appears to rise or fall with the fortunes of the faculty theory. And indeed empiricist-style faculty theories have suffered much from doubts about the notion of pure sensations as mental entities and about how truly scientific such theories may be.
However, none of the deficiencies of faculty theory directly impugn the validity of Kant’s refutation of the ontological proof. For we have shown that the ontological proof falls due to the incoherence of the concepts it employs, concepts like greatest, infinite, existence and necessary. Showing that incoherence has nothing to do with Kant’s faculty theory. Likewise, Kant’s understanding of "existence" in the case of objects of sense, namely "being the object of a possible perception," can be replaced by the more epistemologically neutral "possessing mass and extension at points in time and space." (The notion of non-existent time and space throws a wrench in the dispute that is beyond the scope of this little essay.)
Despite the avatars of the faculty theory, it would be wise to reflect a bit on the subtleties of Kant’s gloss on "existence" with respect to sensible objects as "being the object of a possible experience." For example, let us transpose the one hundred thalers to a somewhat more modern currency system where they appear, on the one hand, only as debits and credits in various bank ledgers and are never transformed into printed bills for the entire time that the thaler is in use as currency, and, on the other, one hundred thalers that were ensconced in a safe deposit vault as soon as they were printed. Both exist. The imaginary thalers in this case could be the result of a fraudulent credit in someone’s ledger. This last example may throw some light on what physicists mean when they say something exists if it is measurable. The entity does not have to occupy a point in space and time with mass. But in some way it must be theoretically related to entities that do occupy points in space and time with mass. The theoretical relation could be by means of a purely physical theory or it may be by means of observed human or animal behavior. Behavior may be considered experientiable even though it cannot clearly be reduced to some physical theory or other.
Kant was fully aware that to be an object of possible experience meant not only to be an object of a possible perception but also to be the object of "…conclusions that connect it to perception….(p.535)" The idea of possible (mögliche) experience, however, is a problem. It threatens to undermine the carefully drawn distinction between objects in the real world and pure inventions of reason or the imagination. The challenge for a Kantian, and it is not an easy one, is to tie our understanding of real objects existing in this world to knowledge and scientific observation without relying on the troublesome and suspect concept of possibility.