Kant on the Notion of Being İlhan İnan Bogazici University, Department of Philosophy In his Critique of Pure Reason Kant attempts to refute Descartes' Ontological Argument for the existence of God by claiming that "'being' is not a real predicate". In this essay I wish to explore what Kant means by this, what arguments he gives in support of it, and whether by this claim Kant anticipates the Frege/Russell view that existence is a secondorder concept. 1 I An important heritage that remains from The Linguistic Turn is the Existential Quantifier. Frege and Russell independently came to the same conclusion that existence is not an ordinary concept that ascribes a property to a particular object. Frege analyzes existence as a second level concept ; a proposition in the form "F exists", according to Frege, should be analyzed as "There is at least one object that falls under the concept F". Similarly a negative existential claim in the form "F does not exist" turns out to subscribe to the concept of F being empty. 2 In a similar vein, Russell takes existence to be a property of a "propositional function 3 which comes close to, and perhaps is the same as Frege's "concept" (which Frege defines as functions from objects to truth values). For both philosophers an existential claim makes no reference to an object, which allows them to solve the time honored Riddle of non-being. For instance if the sentence "God exists" makes reference to God, then the same should be expected of its negation "God does not exist", which in effect would lead to a contradiction. Therefore in both of those sentences, there is no reference to God, on the Frege/Russell view, but only a reference to the concept God; the first says the concept is not empty and the latter says that it is. Today perhaps there is no philosopher who denies that this is at least one good use of the existence predicate.
A commonly held view among historians is that Frege and Russell were the pioneers of the view that existence is a second-order concept ascribing a property to a concept rather than an object. Did Kant anticipate this view in his Critique of Pure Reason which was published more than a century before Frege and Russell had published their work? 4 Let us then proceed to the section of the Critique where Kant takes up this issue, which predominantly is concentrated in A599-B627 in the Ideal of Pure Reason. Here Kant's main purpose is not to explore what existence is but rather to demonstrate what goes wrong in the ontological arguments given for the existence of God. Though Kant only makes reference to Descartes, it seems clear that he has in mind the earlier versions due to Anselm and others. The Cartesian version of the argument (given in the fifth Meditation) briefly runs as follows: "God by definition has all perfections; existence is perfection, therefore God exists". Such arguments, according to Kant, wrongly presuppose that " being is a real predicate". Before we go into Kant's arguments, it is vital that we try to clarify the crucial notions of "being" [Sein] and "predicate" [praedikat]. II In the English translations of the Critique, the term "Sein" is, in general, translated as "being" especially in the section we are considering. It seems clear that the term "being" as it occurs in certain utterances of English means "object" or "thing", as for instance in the sentence "Humans are rational beings". If this is the sense we attach to the term in Kant's major claim "'being' is not a real predicate", we would get "An object is not a real predicate" which would be far from what Kant meant by this claim. Rather "Sein" should be taken, in this context as "existence", which I believe is the common interpretation. So in what follows I will put aside the term "being", and use "existence" in the discussion. On the other hand what is more problematic is the term "predicate" [praedikat] which I believe is used ambiguously by Kant. In certain contexts Kant seems to be using the term
to mean property or attribute, yet in other contexts he uses the term to refer to a syntactic/logical part of a sentence. For instance Kant makes a distinction between "real" and "logical" predicates, where he defines the former as a predicate that "does not enlarge the concept of the subject", and the latter as one that does. Now it seems to me that we cannot take the term "predicate" to mean property here, or else we would end up saying that there are two types of properties, real ones and logical ones. This would be inconsistent with Kant's claim that the same predicate can occur as a logical one in one sentential context and a real one in another. He says: Anything one likes can serve as a logical predicate, even the subject can be predicated of itself. Clearly this indicates that whether a predicate is a logical one or a real one is to be determined in terms of the sentential context in which it occurs. It would be wrong, for instance, to ask "is 'red' a logical or a real predicate?" independent of a sentential context. In the sentence "apples are red" it appears as a real predicate, whereas in the sentence "all red things are red" its latter occurrence is a logical one. Nonetheless redness is a property after all. In all of this discussion then, the term "predicate" should be taken not to mean property, but rather a syntactic unit of a sentence, namely the part that remains when the subject term and the copula are deleted from that sentence. On the other hand there are other passages where it could make a lot more sense to take the term "predicate" to mean property. So in what follows, I will use the term "predicate" in the syntactic sense, and in those sections in which I interpret Kant to mean property by this term, I will prefer to simply use the term "property". Here are then two possible disambiguations of Kant's major claim that "existence is not a real predicate". Existence is not a real property. "Existence" is not a real predicate. The former claim makes no mention of terms of a language, whereas the latter is a metalinguistic claim about the English language. Now which one did Kant have in mind? Was he trying to make a syntactic/semantic claim about the use of "Sein" in the German language, or was he making a substantial philosophical and therefore a metaphysical claim? I am not sure as to how to answer this question, and it seems that Kant's text is of
not great help. However intuition has it that it is the former that would seem to reflect the philosophical significance of the issue. So that is the interpretation I will prefer. III The idea that what goes wrong in Descartes Ontological Argument is a misuse of the concept of existence was in fact stated by Gassendi before Kant: Existence is a perfection neither in God nor anything else; it is rather in the absence of which there is no perfection...hence neither is existence held to exist in a thing in the way that perfections do, nor if the thing lacks existence is it said to be imperfect (or deprived of perfection), so much as to be nothing. (Haldane and Ross, 1931:186) Similarly Hume also had noted that existence is not an ordinary concept that applies to an object: The idea of existence...is the very same with what we conceive to be existent. To reflect on anything simply, and to reflect on it as existent, are nothing different from each other. That idea [i.e. of existence], when conjoined with the idea of any object, makes no addition to it. (Hume, 1960: 60-67) Both philosophers seem to have anticipated Kant's thesis. In fact as we shall see one of the arguments Kant gives in support of his thesis (which I call the Epistemological Argument ) is quite close to and perhaps the same as Hume s. However as we go through his other arguments it will become clear that Kant goes well beyond both Hume and Gassendi in his discussion of the matter. I will now try to isolate three separate but inter-related arguments that I believe are inherent in Kant s discussion of his main thesis that existence is not a real property. As it is well known, the discussion is packed into a passage between paragraphs A 592/B 619 and A 603/ B 631, especially A 599/ B 627 in the Ideal of Pure Reason. None of them are explicitly stated in the text, at least not in the way I will reconstruct them. As I do so, I will provide my interpretation of certain key terms used by Kant in his text, though, some of them will no doubt be controversial. At most what I could claim is that
these are the most plausible ways I could think of to reconstruct these arguments in the spirit of Kant s philosophy. 5 a. The Epistemological Argument The first argument one could extract from Kant s discussion of the topic is epistemological in its nature. Though I think that it is one of the main arguments that Kant has in mind in support of his claim, he never explicitly states it. The argument runs like this: One extends his knowledge of a certain object by learning that it has a property that he did not know earlier. Learning that a certain object exists does not extend one s knowledge of that object. Therefore, existence is not a real property. Though I think that this is one of the central arguments Kant had at the back of his mind, his discussion does not contain any reference to extension of one s knowledge or some such epistemic notion. Rather his discussion seems to be, at least on the surface, more semantic regarding the deep grammar of the term existence. The crucial paragraph A599/B 627 starts off with the following line: Being is obviously not a real predicate, i.e. a concept of something that could add to the concept of a thing, and then a few sentences later he says that when Now if I take the subject (God) and say God is, then I add no new predicate to the concept of God In the next paragraph we have the same idea expressed a bit differently: Thus when I think a thing, through whichever and however many predicates I like (even in its thoroughgoing determination 6 ), not the least bit gets added to the thing when I posit in addition that this thing is. (A600/B628) And again a few sentences later he says: Even if I think in a thing every reality except one, then the missing reality does not get added when I say the thing exists... (A600/B628) In all these passages Kant makes use of a key term: adding to the concept of a thing. Similarly in various other passages Kant frequently uses the notion of the
expansion of a concept as well. Now did Kant sincerely subscribe to the view that concepts are these unusual entities that can expand and shrink like balloons? I believe not, for such terms ought to be taken in a metaphorical sense that refer to an mental epistemic process. For instance when Kant says that in a synthetic judgment in the subject/predicate form, the concept of the predicate enlarges the concept of the subject, it is clear that he wishes to refer to a mental phenomenon. So the enlargement of a concept is not an event that takes place in the abstract world of concepts, but rather it is what the mind does in making a synthetic judgement, namely it is through the enlargement of the concept of the subject that we extend our knowledge of the object (referred to by the subject term). Now a predicate used in a judgment that has this function is called by Kant a real predicate, the determination, as well as a determining predicate. 7 These phrases may at times not be used by Kant as synonyms, but as far as our topic is concerned they seem to add up to the same thing. In fact Kant does seem to be using them interchangeably in A596/B 624: the illusion consisting in the confusion of a logical predicate with a real one (i.e. the determination of a thing)..." and at the end of the same paragraph he says: the determination is a predicate, which goes beyond the concept of the subject and enlarges it. Now a real or determining predicate on Kant s view is a predicate that enlarges, or adds something to, the concept of the subject. In that sense a real predicate is, or denotes, a genuine property. A predicate that does not do this is called a (merely) logical predicate. From this we should not conclude that the class of predicates nicely divide up into real ones and merely logical ones, for as I said earlier Kant points out any predicate may be used as a logical one in a sentence. Having said this however, existence can never be used as a real predicate for Kant, as I read him. The main claim then is that existence is not a predicate that enlarges our concept of the subject, and I take that to imply that our knowledge of a thing does not increase when we learn that it exists. Here we run into a difficulty though. If existence is not a real predicate, then we should expect it to be a logical one given that a third category is not suggested by Kant. So assuming that existence is merely a logical predicate, and that an existential judgment does not extend our knowledge of the object in question, we may be forced to claim that all existential judgments are analytic. Note that the other example given by Kant of an
occurrence of a logical predicate is a sentence in which the subject is predicated of itself, a typical analytic judgment. Clearly Kant holds that existential statements are synthetic. In fact before the main discussion starts, Kant introduces this as the main puzzle to be solved. I ask you: is the proposition, This or that thing exists, an analytic or a synthetic proposition? If it is the former, then with existence you add nothing to your thought of the thing if you concede, on the contrary, as in all fairness you must, that every existential proposition is synthetic, then how would you assert that the predicate of existence may not be cancelled without a contradiction? (A598) So what Kant has to show is that unlike other predicates that are used in the logical mode, existence is a peculiar logical predicate such that when we use it in a statement we get a synthetic judgment that is in fact informative. The Epistemic Argument then should be taken not to imply that existential judgments can not extend our knowledge, but rather the extension of our knowledge is not about the object but rather something else. In other words learning that a exists does not extend our knowledge of a but it does extend our knowledge nonetheless. 8 In any case what exactly is Kant s solution to this puzzle, and even if he has one is not clear. Though as I will suggest there are certain vague passages in which Kant hints at a solution that is in the spirit of the Frege/Russell view. But before we get to that we should first look at the other two arguments Kant gives for his main thesis. b. The Conceptual Argument The second argument is purely semantic in its nature, and here we are asked to compare what may seem to be different concepts to notice that they are all one and the same. Again this argument is not stated explicitly either, and here is how it may be reconstructed: Take any concept of an object, say God, and then add existence to it, i.e. existent- God. You will notice that the two concepts will not differ in any respect. Again
take the initial concept, God for instance, and this time subtract existence form it, i.e. non-existent God. Again you will notice that you have not ended up with a different concept. If existence had been a real property, these concepts would have been different, and given that they are not, existence is not a real property. Now an extraction of this argument is not as easy as the first one, for it involves his notoriously difficult passage concerning his example of the hundred dollars, which includes the surprising claim that a hundred actual dollars do not contain the least bit more than a hundred possible ones. Interestingly in this crucial sentence, from which I have derived this argument, the notion of existence does not appear. Rather there is the distinction between the actual and the possible. Yet in other passages we get the distinction between the real and the merely possible. It seems that the notions of the real and the actual are used interchangeably; but what is more problematic is whether they are also used synonymously with the notion of the existent. Now elsewhere Kant does distinguish these terms, especially within his discussion of how ideas may be schematized. 9 Judging that something exists does not imply that the concept of the subject can be schematized, as in the case of God, but judging that it is actual does have that implication. This need not concern us here, for a hundred dollars is no doubt schematizable. In any case, under this interpretation whatever is real and actual is existent, but not vice versa. So at least we could take the above sentence concerning the hundred dollars, even if it is not synonymous with it, to imply the following: the concepts of existent-hundred dollars and non-existent hundred dollars have the same content.(here I have taken possible to be a shorthand for the merely possible, which is supported by the discussion in the text.) This of course is an implausible claim at first blush, for it seems clear that the two concepts cannot be substitutable for one another (salva veritate) in every sentential context. Kant is aware of the fact that his remark may be misunderstood, as he later says in the same paragraph that in my financial condition there is more with a hundred actual dollars than with the mere concept of them (i.e. their possibility.) Here Kant is distinguishing between the concept and the object that falls under the concept, and saying that what improves his financial condition is not the concept but rather the object (the actual money) that falls under the concept. If so how
can we make sense of the initial remark then? How could merely possible dollars have the same content as actual hundred dollars? On the surface these simply are inconsistent claims. There is a more charitable reading though: The concepts of existent-hundred dollars, and non-existent hundred dollars both include the concept of hundred dollars, nothing more and nothing less. The reason why our financial condition is improved by hundred real dollars rather than possible ones, is not because the two concepts are different, but rather in one case the money is actual (object) in the other case it isn t (the possible, i.e. the mere concept.) 10 On the surface this argument is not at all convincing, for anyone who has a prior conviction that existence is in fact a real property would not be persuaded by this argument. All that Kant says here is that the concepts of existent-hundred dollars and non-existent hundred dollars both contain the property of hundred dollars which his opponent need not disagree with. Clearly just because the two concepts have one common element does not in any way show that they are identical. Philosophers who hold that there are non-existent objects could react by saying that a hundred actual dollars have the property of existence, whereas hundred possible ones do not. 11 As we shall see now the third argument is given in support of the same claim by this time going further than the second one. So I think that the Conceptual Argument is an argument in transition, and is not intended to be conclusive. It is in the third argument that Kant offers a new piece of reason to convince us that these concepts are the same. Note that this argument is vital for Kant s thesis: given that existence is not a real property that enlarges or shrinks the concept of the subject, God, existent-god, nonexistent God should all be the one and the same concept, and obviously this is valid for all concepts not just for the concept of God. c. Different-objects Argument Here then is my reconstruction of the third argument: If the concept of F and existent-f had been different, they would have been the concepts of different things. But then when I entertain the concept of F in my
mind it would be impossible to find its object; for any such object would not fall under the concept F, but rather the concept of existent-f. Given that entertaining a concept and then finding its object is possible, the concept of F and existent-f must be identical. If so, then existence does not add anything to a concept, and thus it is not a real property. Again Kant makes use of the same example about the hundred dollars in this argument as well. We are to assume, for a reductio, that hundred actual dollars and hundred possible dollars are different concepts and then conclude that "...in case the former contained more than the latter, my concept would not express the entire object and thus would not be a suitable concept of it." (A599/B627) In the next paragraph the same idea is put differently: "For otherwise what would exist would not be the same as what I had thought in my concept, but more than that, and I could not say that the very object of my concept exits." (A600/B628) Here I take it that the concept of hundred actual dollars even if not synonymous with it to imply or contain the concept of existent hundred dollars. On the other hand the term possible seems to be used by Kant in some epistemic sense, which is that if a concept is possible then its object can be given in intuition, in Kant's terms. This I believe has the implication that merely entertaining a possible concept in the mind, has the implication that the entertainer of the concept does not know the object that falls under it. Let us now run the argument by making use of a simple example. Suppose I am curious about whether there is anyone in the audience who has never read Kant. In order to be curious about such an issue I need to form the concept of people in the audience who have never read Kant. To find out the answer, I could go around asking each person in the audience whether he or she has ever read Kant. Now as a result of my inquiry it would seem that I would find out the extension set of the concept. But that extension set would be that of the concept of existing people in the audience who have never read Kant as well. But if we assume that existence is a real property, then the two concepts would not be identical. If so, then their extensions should also differ. Therefore the extension I find would not be the extension of my original concept.
Now a crucial premise of the argument is the following: If two concepts are not identical, then their extensions would also not be identical. This, at first blush seems obviously false: Suppose, for instance, I take the concept of the youngest person in the audience and add the property of being male to it and get the concept of the youngest male in the audience; it may turn out that the two concepts are co-extensional if it happens to be the case that the youngest person in the audience is a male. So it seems that Kant wrongly presupposes here that if two concepts are different than their extensions should also differ. There is however a more charitable reading of the argument which need not presuppose this fallacious principle. Going back to the earlier example, doing an inquiry with respect to the concept of people in the audience who have never read Kant would be exactly the same as what we would do with respect to the concept we get by deleting existence form it. After all it sounds ridiculous to go up to someone in the audience and ask "do you exist, and if so have you read Kant?" Things would have been different if instead of adding existence I had added, say the property of being male to the first concept. Even if it turns out that the only people who have never read Kant in the audience are all male, making my two concepts co-extensional, the inquiry I would have to make would differ. So there is something peculiar about existence, namely, when it is added to a concept, there is a sense in which it does not expand the original concept. Therefore we are to conclude that possible hundred dollars is the same concept as existing hundred dollars, which in effect shows that existence is not a real property. IV There are of course various ways in which one may object to these arguments. It is highly unlikely that anyone who holds that there are merely possible objects would be convinced by them. There is however a greater problem for Kant. If existence is not a real property, then it does not expand or add anything to the concept of the subject. But this is an essential feature of analytic judgments. In the Introduction to the Critique he explicitly says that "analytical judgments...do not add anything to the concept of the subject". We would then have to conclude that existential judgments are also analytic, but Kant
obviously would not be happy with this implication for he holds that "as in all fairness you must" concede that every existential proposition is synthetic. (A598/B627) So what then accounts for the synthetic character of existential judgments? The reason why the concept of the predicate does not expand the concept of the subject in an analytic judgment is because the latter is already included in the former. For instance our concept of body already has in it the concept of being extended, which is what makes our judgment that all bodies are extended analytic. Now this does not hold for existential judgments. The judgment that trees exist is not true in virtue of the fact that the concept of tree already includes the concept of existence. So there is this kind of asymmetry between analytic and existential judgments; though they share the feature that the concept of the predicate does not expand the concept of the subject, only in analytic judgments will it be correct to say that the concept of the predicate is already included in the concept of the subject. Kant then has to give a different kind of account concerning the synthetic character of existential judgments. 12 But interestingly Kant does not even address the issue, so we need to read between the lines. Here is a passage in which there is a hint towards a solution: Now if I take the subject (God) together with all his predicates...and say God is, or there is a God, then I add no new predicate to the concept of God, but only posit the subject in itself with all its predicates, and indeed posit the object in relation to my concept. (A599/B627) 13 This is as close as Kant gets to the Frege/Russell view. Had he taken a further step, he could have said that existence is not a property of objects but rather a property of concepts, and conclude that when we judge that God exist, we judge that the concept of God has an object. This would have provided him the means to give an account of the synthetic character and thus the informativeness of existential judgments. However it seems clear to me that Kant never reached this conclusion explicitly. This is made evident by a footnote added to the second edition where he comments on Descartes' Cogito:...here existence is not yet a category, which is not related to an indeterminately given object, but rather to an object of which one has a concept, and about which one wants to know whether or not it is posited outside this concept. (B423)
Interestingly, after a lengthy discussion attempting to show that existence is not a real property that applies to objects, Kant seems to be saying here that existence does apply to objects in this peculiar way. If for instance I am curious about whether God exists, following this line of reasoning, I should say that what I wish to know is whether God (the object) is posited outside my concept of God. But that would make my curiosity de re, implying that existence does apply to objects. This confusion, I believe, stems from the fact that Kant never reached the idea that there may be certain predicates, and hence certain properties that apply to concepts rather than objects. Nonetheless it seems that he did come very close to the Frege/Russell view; but had he explicitly endorsed it, his slogan would not have been "'being' is not a real predicate", but rather "'being' is a special kind of predicate that applies to concepts and not objects." 1 This essay is an expanded version of a talk I gave at the International Kant Symposium organized by the Philosophy Department of Mugla University marking the two hundredth anniversary of Kant's death. I would like to thank the audience for their comments, especially Paul Guyer and Manfred Baum. I have greatly benefited from my discussions I had with Lucas Thorpe on an earlier draft. 2 See Frege (1960), especially 46ff. Frege's analysis of existence seems to involve circularity. When we analyze "F exists" as, "There is something that falls under the concept F", we appeal to existence in the analysis. In order to overcome this apparent circularity, one could say that the proper Fregean analysis of "F exists" ought to be "F is not an empty concept", where we take the concept of emptiness as being primitive, or at least one that is not analyzed further into something involving existence. 3 See Russell (1914). It seems that Russell takes the discussion further than Frege and claims that in the sentence "God exists" the term "God" cannot be a proper name. This is in fact a logical implication of Frege's view as well, though to my knowledge it was never stated explicitly by Frege. 4 A common view among historians is that it was Frege and Russell who first came up with the idea that existence is a quantifier that is a second-order concept. As I will argue, the origins of this view can be found in certain parts of the Critique of Pure Reason, whose significance I believe has been greatly ignored. An important exception is Forgie (2000), who argues that Kant clearly defended the view that existence is a second-order concept. As I will try to show, on my view Kant never clearly stated this. In his discussion Forgie takes the textual evidence for this claim mainly from the Beweisgrund (1763) which was published before the Critique, though in my talk I only took up the relevant parts of the Critique. Nonetheless it also seems to me that Forgie reads too much into what Kant says in the Beweisgrund as well; or else Kant would not have stated his main thesis as "'being' is not a real predicate". I will come back to this at the end of my essay. 5 I received no substantial reaction to my re-construction of these arguments form Kant scholars such as P. Guyer, M. Baum, and L. Thorpe. I take this to be a good sign that my interpretation is in the right track. 6 I take the "full determination" of an object to mean the concept of the object that includes all the properties of that object. 7 The determination of an object is achieved by collecting the properties of the object in a concept of it. That is why a logical predicate does not add anything to the determination of the object. 8 Perhaps Kant could have appealed to the de re/de dicto distinction that was available to him at the time, and claim that existential judgments are not de re but rather de dicto. 9 I owe this point to Lucas Thorpe. 13
10 After reading an earlier draft Uygar Abaci put this rather neatly: "With real hundred dollars you can buy a lot of stuff, but with merely possible ones you can only do philosophy." 11 I have in mind especially A. Meinong (1914), T. Parsons (1979), and N. Salmon (1987). Despite the fact that these philosophers differ sharply in what they say about existence, it seems they share the common view that there are merely possible, therefore non-existent objects. So it follows from their view that at least one understanding of the existence makes it a first order property that could be attributed to objects. However unlike Meinong, both Parsons and Salmon also allow for another notion of "existence" as expressing a second-order concept. On the other hand Parsons and Salmon disagree on how to individuate possible entities. 12 Everitt (2004) after clearly stating the problem claims that Kant falls into a contradiction and has no solution to offer. 13 We come across a similar passage in the "Beweisgrund": "If I say, 'God is an existing thing', it appears that I express the relation of a predicate to a subject. But there is an incorrectness in this expression. Expressed exactly, it should say: something existing is God, that is, those predicates that we designate collectively by the expression 'God' belong to an existing thing." As it can be noticed both here and in the quote in the main text, the circularity problem (see note 2) crops up, that is Kant has to appeal to the notion of "existence" in analyzing it. References: Kant, I., (1999) (first published in 1781) Critique of Pure Reason, Editors: P. Guyer, A.W. Wood, Cambridge University Press. Forgie, W., (2000) Kant and Frege: Existence as a Second-Level Property, Kantien Studien, Band 91. Frege, G., (1962) The Foundation of Arithmetic, tr. J.L. Austin, New York: Harper. Haldane & Ross, (1931) The Philosophical Writings of Descartes, London: Cambridge. Everitt, N. (2004) The Non-existence of God, London: Cambridge. Hume, D. (1960) (first published in 1888) A Treatise Concerning Human Nature, ed. L.A. Selby-Bigge, Oxford: Clarendon Press. Meinong, A. (1960) (first published in 1914) The Theory of Objects, Realism and the Background of Phenomenology, ed. R. Chisholm, New York: The Free Press. Russell, B., (1918) The Philosophy of Logical Atomism, The Monist. Parsons, T. (1979) Referring to Nonexistent Objects, Theory and Decision s.95-110, Kluwer. Salmon, N. (1987) Existence, Philosophical Perspectives 1. 14