Bill Forgie, fellow worker in the vineyards of existence, sent me this paper which will be appearing in the Journal of the History of Philosophy. Since this is right up my alley, and contradicts long-held tenets of mine, I cannot let it pass unexamined. My comments, indicated by 'BV' -- which does not abbreviate BloViator -- will appear in blue.
Professor Forgie makes two main points in this paper. The first is that Pierre Gassendi (1592 - 1655) and Immanuel Kant (1724 - 1804) are referring to two different things when they use the word 'existence' (or their Latin and German equivalents: esse, existentia, actualitas, Existenz, Dasein, Sein, Wirklichkeit.) Gassendi is talking about a 'property' (in a broad sense) or determination of individuals, whereas Kant is talking about a property of concepts, the property of being instantiated. The second main point is the familiar one that Kant anticipates Frege's theory of existence as a second-level concept.
Gassendi and Kant on Existence
In rejecting Descartes’ ontological proof for the existence of God, Gassendi maintained that existence is not a property and Kant said that it is not a "real predicate." It is commonly supposed that both are making the same claim. Some have even thought that they advance essentially the same argument for that same claim. I believe none of this is correct. Gassendi and Kant offer different arguments. And they are arguing for different conclusions. These differences stem from a more fundamental one: they mean different things by existence.
To my knowledge, Gassendi is the first to have based a rejection of the ontological argument on the claim that existence is not a property.
BV: Correct. Cf Dieter Henrich, Der Ontologische Gottesbeweis, p. 76.
Discussing Descartes’ fifth Meditation proof, Gassendi calls attention to Descartes’ claim that existence is inseparable from the essence of God just as having its three internal angles equal to two right angles is inseparable from the essence of a triangle. Gassendi claims that there is a peculiar asymmetry in such a comparison:
It is quite all right for you to compare essence with essence, but instead of
going on to compare existence with existence or a property with a property, you
compare existence with a property. To compare existence with property is not
to compare property with property.
But Gassendi does not directly tell us why he believes existence is not itself a property. He does, however, argue for a related claim, namely that existence is not a perfection, and it is clear that the sort of argument he uses for this latter claim can also be used, with suitable modifications, for the view that existence is not a property. Here is his argument that existence is not a perfection:
... existence is not a perfection either in God or in anything else; it is that
without which no perfections can be present. For surely, what does not exist has
no perfections or imperfections, and what does exist and has several perfections
does not have existence as one of its individual perfections; rather, its
existence is that in virtue of which both the thing itself and its perfections
are existent, and that without which we cannot say that the thing possesses the
perfections or that the perfections are possessed by it. Hence we do not say
that existence 'exists in a thing' in the way perfections do; and if a thing
lacks existence, we do not say it is imperfect, or deprived of a perfection, but
say instead that it is nothing at all.
If we substitute ‘property’ for ‘perfection’ in this passage, we can extract the following argument that existence is not a property: (a) If existence were a property, something lacking existence would be lacking a property; (b) But existence is necessary for anything to have or lack any properties; (c) So, something lacking existence cannot lack any properties; (d) Therefore, existence itself is not a property.
BV: I would put Gassendi's argument as follows. 1) If existence is a property of individuals, then x exists in virtue of x's instantiating of existence. But, 2) existence is a necessary condition of an individual's having any property whatsoever: only if an individual exists can it support properties if properties are thought of along the lines of Aristotelian accidents), or stand in an instantiation relation to properties (if properties are thought of along more Platonic lines). Therefore, 3) if existence is a property of individuals, then existence is both a property of individuals, and something more fundamental that an individual must have if it is to instantiate any properties at all, including the putative property of existence. But 4) the consequent of (3) is self-contradictory: existence cannot be both a condition of an individual's having any properties whatsoever and one of the properties had. Therefore, 5) the antecedent of (1) is false: existence is not a property.
For a more detailed, and more rigorous reconstruction and evaluation of Gassendi's argument, see my A Paradigm Theory of Existence: Onto-Theology Vindicated (Kluwer 2002), pp. 53-55.
For Hugh McCann's review of this book, see here.
It is not my aim to discuss the merits of this argument. But one issue needs to be mentioned. Suppose we agree that existence is a necessary condition for having or lacking properties. Why should we conclude that existence is therefore not itself a property? Why should we not conclude, instead, that it is a universal property, a property of everything (or, what is the same, of everything having any properties at all)? This is exactly the view argued more than forty years ago in an elegant paper by Nakhnikian and Salmon. On this latter view, one could accept (b) in Gassendi’s argument while rejecting (a) - holding that ‘exists’ does not admit of internal negation and that therefore nothing ‘lacks’ existence in the sense in which a soldier, together with all his other properties, might nevertheless lack the attribute of courage.
BV: Well, if Gassendi's argument is reconstructed in my way, then I think it is clear why the Nakhnikian and Salmon view is to be rejected. If existence is a property of everything, then, for any x, x exists in virtue of instantiating existence. But an individual cannot instantiate a property unless it already (logically speaking) exists: an individual's existing is a logically prior condition of its instantiating any property. This is why existence cannot be a property of individuals. As Gassendi sees, existence is too ontologically basic to be a property. Gassendi refuted N & S 300 years before their article was published.
Or, if you insist that there is a first-level property of existence, then it is not a property in virtue of whose instantiation any individual exists. Existence construed as such a property is not the genuine article. As I recall the N & S paper, they end up saying that x exists =df x is self-identical. But surely nothing exists in virtue of its being self-identical!
Fortunately we do not need to decide between Gassendi and Nakhnikian and Salmon. But the dispute does draw attention to a feature of Gassendi’s view that needs to be stressed. Even after denying that existence is a perfection or property, Gassendi is not shy about talking of existence as something that is had by things (objects). In the passage quoted above, e.g., he speaks of "its existence." And in surrounding passages he speaks - as though they were on the same level at least - of both the existence and the essence of Plato, God, a triangle, a sloping mountain and a winged horse. Gassendi apparently thinks of existence as something that is possessed by, or applies to objects, even though it is not a property of those objects. It is something that operates at first level, though not as a property. Let us call this a first-level pseudo-property. So the dispute between Gassendi, on the one hand, and Nakhnikian and Salmon on the other, is a dispute about whether something which both sides agree operates at first level is a property or only a pseudo-property. Existence, on both views, is something possessed by objects; and, again on both views, an object must have existence in order to have any properties at all. The only real issue seems to be whether something that is a necessary condition for having properties is itself a property.
BV: You need to tell us exactly what you mean by a property. Suppose P is a property =df P is possibly such that it is instantiated. Then existence cannot be a property. See A Paradigm Theory of Existence, Ch. 2. I would insist, however, that existence can belong to individuals without being a property of them.
Suppose P is a property of x =df P is an accident that inheres in substance x. It is spectacularly clear that existence cannot be a property of an individual on this definition of 'property.'
Suppose P is a property of x =df P is an ontological constituent of x. Then it is also clear, as I argue in PTE, Ch. 2, that existence cannot be a property of individuals. The existence of a thing cannot be one of its properties; it is more like the togetherness of all its properties.
In his famous paper defending Anselm’s ontological proof, Norman Malcolm, believing he is "merely restating an observation that Kant made in attacking the notion that ‘existence’ or ‘being’ is not a ‘real predicate’", argues as follows:
Suppose that two royal councilors, A and B, were asked to draw up separately
descriptions of the most perfect chancellor they could conceive, and that the
descriptions they produced were identical except that A included existence in
his list of attributes of a perfect chancellor and B did not. ... One and the
same person could satisfy both descriptions. More to the point, any person who
satisfied A’s description would necessarily satisfy B’s description and vice
versa. This is to say that A and B did not produce descriptions that differed in
any way but rather one and the same description of necessary and desirable
qualities in a chancellor. A only made a show of putting down a desirable
quality that B had failed to include.
Malcolm speaks here of "desirable" qualities, because he is specifically addressing Descartes’ claim that existence is a perfection, but it is clear that his argument is meant to apply to any two lists of qualities, desirable or not, which are the same except that one contains existence and the other does not. His claim is that anything satisfying the list without existence will necessarily satisfy the list that includes existence. This view contains two important features. First, existence is evidently something that operates at first level; persons, for example, will satisfy, i.e., will have all the elements included in, the list containing existence. And second, having existence is necessary for a thing to have any properties at all. Malcolm says he is restating the view of Kant. But Malcolm’s Kant sounds just like Gassendi. So it is not surprising that Malcolm adds the following to his purported exposition of Kant:
It is worth noting that Gassendi anticipated Kant’s criticism when he said,
against Descartes: Existence is a perfection neither in God nor in anything
else; it is rather that 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 a
perfection), so much as to be nothing.
Malcolm is a clear case of one who thinks Gassendi and Kant agree in proclaiming the doctrine that existence is not a property. But Malcolm goes further. Not only do Gassendi and Kant put forward the same doctrine, they do so, he maintains, for the same reasons. Other commentators do not mention Gassendi, but attribute a view to Kant that is in fact Gassendi’s. Jonathan Barnes, for example, stipulating ‘x is an existent F’ to be an abbreviation of ‘x both is F and exists’, focuses on the following claim: For any property F and any object x: x is F if and only if x is an existent F and maintains that this claim "represents the substance of Kant’s claim that being is not a real predicate." Like Malcolm, then, Barnes believes the heart of the Kantian claim is the view that existence is necessary for having properties.
Alvin Plantinga, like Barnes, does not mention Gassendi. But also like Barnes, he attributes a Gassendi-like view to Kant. Plantinga employs the notion of what we have called a "complete" concept of an individual. He calls it the "whole concept" of an individual. Let C1 be the whole concept of the Taj Mahal. Let C3 include everything that C1 contains except existence. Plantinga calls C3 the whole concept of the Taj Mahal "diminished with respect to existence." Now Plantinga suggest that Kant’s point is as follows:
... there are no possible circumstances in which C3 but not C1 has application;
it is a necessary truth that if C3 is exemplified, so is C1. Since the converse
is also true, C1 and C3 are, we might say, equivalent concepts; in annexing
existence to C3 we don’t really get a different concept. ... the whole
concept of a thing diminished with respect to existence is equivalent to the
undiminished whole concept of that thing.
Plantinga’s Taj Mahal example makes the same point as Malcolm’s example of the perfect chancellor. Anything satisfying a description, or instantiating a concept, without existence will necessarily satisfy or instantiate the corresponding one including existence. Anything having any properties at all must exist. Existence is necessary for having properties. Once again, Kant is represented as arguing as Gassendi did.
BV: The Gassendi-Kant point is this: the existence of an individual, that which makes it be there in the first place, and so equips it to instantiate concepts, is not includable in any concept. Given that properties are includable in concepts, existence is not a property. If, however, one perversely tries to import existence into a concept, like your Santa Barbara colleague Nathan (not to be confused with Wesley) Salmon, then the existence so imported is not the genuine article, but what I call pseudo-existence.
Is Kant just an echo of Gassendi?
BV: Of course, Kant on existence is not just an echo of Gassendi on existence, but I would urge that they have the same insight although they articulate it in different ways against the backdrop of different assumptions and concerns.
The famous passages in the first Critique about existence and the ontological argument are difficult to understand and have been variously interpreted. I believe the arguments presented there are relatively compressed variations of the argument Kant expressed more clearly and more fully in the 1763 pre-critical essay, "Der einzig mögliche Beweisgrund zu einer Demonstration des Daseins Gottes" (hereafter "Beweisgrund"). Consider the following passage:
Existence is not a predicate or determination of anything whatsoever. ... Take
BV: Surely this passage is consistent with the interpretation whereby Kant is assimilated to Gassendi on the score of existence.
any subject you like, for example Julius Caesar. Combine in it all its
conceivable predicates (not excepting those of time and place). You will then
see that, with all these determinations, it may or may not exist. The being
which gave existence to the world and to this hero was able to recognize all
these predicates - not a single one excluded - and could still regard him as a
merely possible thing which, save for His decree, did not exist. Who can deny
that millions of things that really do not exist are, with all the predicates
they would contain if they existed, merely possible; that in the conception
which the highest being has of them, not one of these predicates is lacking,
although existence is not among them. For He knows them only as possible things.
Therefore, it cannot occur that if they exist they contain one more predicate;
for in the possibility of a thing according to its complete determination, no
predicate whatsoever can be missing. And if it had so pleased God to create
another series of things, another world, then it would have existed with all the
determinations, and nothing more, which He discerned in it, although it is only
Kant often speaks of an individual’s concept of a particular thing. Here, for example, he speaks of the concepts God has of "millions" of things that do not exist. He also speaks of the predicates that are "in" God’s concept of a thing. And he invites us to "combine in [a thing]" various predicates. Let us assume that either of these idioms can be used when we want to talk about the predicates one conceives a thing to have. Suppose that, prior to creation, God has a concept of Caesar and that he conceives Caesar to be white and to be intelligent. That is, God's concept of Caesar is a concept of a being having the predicates "being white" and "being intelligent." We can then say that those two predicates are "in" God's concept of Caesar or that they are two of the predicates which God has "combined" in Caesar when conceiving him.
While speaking of the "millions" of things of which God has concepts, but which do not exist, Kant says that every predicate each such individual would have if it were to exist is in God's concept of it. Let us say that any such concept, whether God's or anyone else's, is "complete." If the thing of which we have such a concept actually exists then, of course, every predicate that thing actually has is in our concept of it.
BV: So far, so good. But your talk of "our concept" prompts me to say that no finite mind such as ours could have a complete concept of anything. A complete concept would include not only all monadic determinations, but all relational ones as well. It follows that to have a complete concept of an individual one would have to have a complete concept of a possible world.
Kant tells us that God (the "highest being") has concepts of "millions of things" which do not exist. Each of God's concepts is complete ("... in the conception which the highest being has of them, not one of these predicates [i.e., those they would have if they existed] is lacking. …"). I take Kant's mention of God here to be a rhetorical device used to make the point that it is possible to have a complete concept of something that is merely possible.
Now Kant tells us that, although God's concepts are complete, existence is not one of the predicates included in any of those concepts ("... existence is not among them. ...").The reason existence is not included is that God knows these things only as possible things ("For He knows them only as possible things."). The suggestion appears to be that if existence were included in God's concepts, the objects of those concepts would be actual things, and God would know them as actual things.
BV: I would put it this way. Not even God can import actual existence into a concept. A complete concept is saturated -- this is a chemical metaphor -- in the sense that adding another conceptual determination to it would 'precipitate' -- sticking with the chemical metaphor -- a contradiction. For God to create is for God to actualize. But God creates Julius Caesar not by adding a further determination to his concept -- which is already completely determined, and so cannot be further determined -- but by willing the existence of something that falls under the concept. You could say he brings it about that the concept is instantiated, but not in the sense that be brings it about that the first-level concept instantiates a second-level concept, but in the sense that he creates an existing individual. The existing of an individual is its being divinely posited (gesetzt). Thus existence is neither a property of an individual, nor a property of a property (or concept) of an individual.
Thus Kant appears here to be making the claim that existence cannot be in a concept of a merely possible being. Now if God's concepts of these merely possible beings are complete, and if existence cannot be in a concept of a merely possible being, it follows that if one of those beings were later to exist, existence would not be one of its predicates ("Therefore, it cannot occur that if they exist they contain one more predicate..."). We can set out this argument as follows:
(1) It is possible to have a complete concept of a merely possible being, N;
(2) Existence cannot be included in a concept of a merely possible being;
(3) Therefore, if N were to exist, existence would not be one of its predicates.
This argument looks nothing like Gassendi’s. If Kant and Gassendi are arguing for the same conclusion, they are doing so in quite different ways.
But deeper differences emerge when we consider step (2) of the "Beweisgrund" argument. Why can’t existence be included in the concept of a merely possible being? If we approach Kant’s argument thinking of what Gassendi is talking about when he discusses existence, this will seem a most implausible claim. On our present understanding of the notion of a predicate's being "in a concept," to say that "being white," for example, is in God's pre-creation concept of Caesar is to say that God conceives Caesar to be white.
BV: This is not an exact way of putting it. Before God creates Caesar there is no Caesar to conceive as white or anything else.
Let us say that there exists an instance of one's concept of N if, and only if, there exists something having each of the predicates one conceives N to have. Then if "being white" is in God's concept of Caesar, being white is a necessary condition anything must meet if it is to be an instance of God's concept. Now there is no problem in supposing that "being white" can, in this sense, be in a concept of a merely possible being. For to suppose that being white is a necessary condition for anything's being an instance of God's concept of Caesar is not to suppose that there is an instance of that concept. For it might be that there exists nothing meeting that necessary condition. And even if there do exist things satisfying that condition, it might be that none of them satisfies all of the other necessary conditions for being an instance of God's concept. But in the sense we are currently considering, there is also no problem in supposing that existence - what Gassendi calls existence ("G-existence" for short ) - can be in a concept of a merely possible being.
BV: Now you have lost me entirely. G-existence is that which an individual must have if it is to instantiate any properties. Clearly, G-existence cannot be contained in any concept: no concept is such that, if you analyze it far enough, you will extract G-existence. Othewise, one could prove that the existent golden mountain exists simply by analyzing the concept existent golden mountain.
Suppose that in God's concept of Caesar are the predicates P1, P2, ... Pn and existence.
BV: I find this supposition to be absurd. Genuine, pound-the-table existence simply cannot be analytically contained in any concept. If you insist that existence can be contained in a concept, then I will charge you with having changed the subject. The subject is genuine existence, the existence that makes the difference between the possible (even when completely determined) and the actual, the existence that a thing must have if it is to have properties.
The reason Gassendi rejects Descartes' Meditation V ontological argument, which is an argument aus lauter Begriffen in Kant's phrase, is precisely because existence cannot be contained in any concept, not even the God-concept.
This commits us only to supposing that there are several necessary conditions any being must meet in order to be an instance of that concept - it must exist and it must have properties P1, P2, ... Pn.
BV: Now it seems you are equivocating on 'existence.' When you say, "it must exist..." you are speaking of G-existence, the genuine article. But in the preceding sentence you were speaking about the existence that can be imported into a concept, which is NOT the genuine article. There is a sense in which the existent golden mountain exists: it C-exists, i.e, it exists conceptually. But that's not to say that it G-exists, that it exists in reality.
Now although existence is in God's concept it might be that nothing meeting that condition meets all the other conditions necessary for being an instance of that concept.
BV: Sorry, but this sentence makes no sense. Let us suppose that C-existence is in God's concept of x. How could anything that is possible fail to meet that condition? C-existence is a condition of everything actual and possible. Since everything actual and possible meets the condition of C-existence, it follows that it makes no sense to suppose that nothing meeting the C-existence condition meets all the other conditions.
It might be, in other words, that nothing which exists has the predicates P1, P2 ... Pn, i.e., that God's concept has no instance. Indeed, it is easy to see that if a concept containing P1, P2, … Pn but not including existence, has no instance, then the related concept containing P1, P2, … Pn, including existence, also has no instance. The claim that there exists no instance of the first of these concepts is logically equivalent to the claim that there exists no instance of the second.
It seems, then, that even if G-existence were included in God’s concept of Caesar, Caesar might still be (and God might still know Caesar as) a merely possible object.
BV: But again, G-existence cannot be included in any concept. If, per impossibile, it were imported into a concept, it would not be G-existence, but C-existence.
I believe I have here simply been adapting an insight earlier used by Aquinas and Caterus in attacking versions of the ontological argument - viz., that no matter what predicates one conceives N to have (and so no matter what predicates are "in" one’s concept of N, in the sense we have been using), it is an independent, and further, question whether one’s concept of N has an instance, i.e., whether N is an actual or only a possible object. The Aquinas/Caterus insight appears to cast doubt on step (2) of the "Beweisgrund" argument - at least if Kant is talking about G-existence.
BV: Why do you say this? What you are calling the Aquinas/Caterus insight is precisely what (2) expresses, namely, that existence cannot be contained in any concept!
But the Aquinas/Caterus point is an insight employed by Kant himself. And indeed it is this insight that underlies and explains - rather than casts doubt on - Kant’s claim that existence cannot be in a concept of a merely possible being.
BV: This is certainly correct. But note that existence cannot be in the concept of an actual being either. Concepts are neutral with respect to existence. It doesn't matter whether they are incomplete or complete, or of actual or merely possible beings.
Once this is apparent, we will see that Kant is not talking about G-existence at all.
BV: What?!? The preceding two sentences were crystal clear. But how did you arrive at this sentence? Of course Kant is talking about G-existence, and what he is saying about it is that it cannot be included in any concept. I hate to repeat myself, but G-existence is the existence that makes an individual exist rather than being nothing at all. It belongs to individuals, but not in the manner of a property. It is too basic to be a property since it is what makes possible a thing's having of properties. Here is a crude analogy. For pins to be stuck into a pin cushion, there must exist a pin cushion. Suppose someone says that the existence of a given pin cushion is one of the pins stuck into it. That is patently absurd.
Consider, for example, the following passages from the first Critique.
In whatever manner the understanding may have arrived at a concept, the
existence of its object is never, by any process of analysis, discoverable
within it; for the knowledge of the existence of the object consists precisely
in the fact that the object is posited in itself, beyond the [mere] thought of
it. Whatever, therefore, and however much our concept of an object may
contain, we must go outside it, if we are to ascribe existence to the object.
Even when the concept of a thing is quite complete, I can still inquire whether
this object is merely possible, or is also actual ... (B266)
Kant puts this claim in other ways. He says that no "mark" of the existence of the object
can be found in the concept of it:
In the mere concept of a thing no mark of its existence is to be found. For
though it may be so complete that nothing which is required for thinking the
thing with all its inner determinations is lacking to it, yet existence has
nothing to do with all this, but only with the question whether such a thing be
so given us that the perception of it can, if need be, precede the concept.
And he also puts the claim by saying that existence is never one of the predicates included in a concept of a thing. In "Beweisgrund," for example, he warns that one should not:
... attempt to derive existence from merely possible concepts, as one is
accustomed to doing when attempting to prove absolutely necessary existence. For
one would then be looking in vain among the predicates of such a possible being
- existence cannot be found among them. (76)
BV: All of these passages, which I have read hundreds of times over the past 35 years, beautifully support my interpretation. Indeed, from them I got my interpretation.
Kant is claiming in various ways that nothing included in a concept in the way we have been considering so far - i.e., none of the features we or God conceive the object of the concept to have, and so none of the conditions anything must satisfy in order to be an instance of that concept - tells us that that concept is instantiated. When Kant claims that existence cannot be in the concept of a merely possible being, he appears to mean not that we cannot include G-existence among the necessary conditions anything must meet in order to be an instance of the concept - a claim which seems clearly false (and would have to be false if Gassendi is correct) - but simply that nothing which we do include in a concept in that way indicates whether its object is actual or merely possible, i.e. indicates whether or not that concept is instantiated.
BV: Part of what you are saying is that if Gassendi is correct, then we can include G-existence in concepts. Why? G-existence is that which an individual must possess if it is is instantiate properties. Now concepts and properties are 1-1. I mean, every concept is a concept of a property, and every property is conceptualizable, even if not actually conceptualized. Now since existence cannot be a property, as Gassendi proved, existence cannot be a concept, or included in a concept. (Everything included in a concept is itself a concept: inclusion is a relation defined over concepts.)
On this understanding Kant’s claim that existence cannot be included in the concept of a merely possible being will be trivially true. For if something included in God’s concept of Caesar in our earlier sense indicated that that concept was instantiated, then Caesar would actually exist and not be a merely possible being.
BV: There is something wrong here. Existence cannot be included in the concept of any being, actual or possible. But this is scarcely trivial. After all, Descartes thought that by analyzing the concept of God one could arrive at the insight that God exists.
Kant is not discussing G-existence.
BV: I say that is exactly what he is discussing. What he is saying about it is that it cannot be included in, or extracted by analysis from, any concept.
What he is calling "existence" seems to be the property "being instantiated" or "having instances".
BV: Are you saying that Kant changed the subject? Or that he is equivocating on 'existence'?
Today we would call this a second-level property. And we would say that it is a property of concepts and not properties we would expect to find in concepts of objects (though we might find them in concepts of concepts). And so, of course, Kant’s "existence" will not be a property of objects; it will not operate on the right level for that. But Kant’s argument cannot be as slick, or as quick, as this. For he does not talk of "levels" of properties, and his argument is made completely independently of any reference to levels of predication. He simply notes that no matter how complete a concept of an object might be, that is (epistemically) compatible with its being a concept of a merely possible being. In other words, whether the concept is one of an actual or merely possible being is not itself something included in the concept. He puts this point by saying existence is not included in the concept of a merely possible being. And this is all he thinks he needs, once we grant the claim that a complete concept of a being is possible, to conclude that existence - what Kant refers to as existence, at any rate (call it "K-existence") - is not a property of those objects that do exist.
BV: It is important to realize that if existence is not a property of individuals, it does not straightaway follow that it is a property of properties or of concepts. For it may well be that existence, though not a property of individuals, still belongs to them as it would not belong to them if Frege and Russell are right.
The impression that Kant is not talking about G-existence, indeed that he is not talking about anything operating at first level at all but instead discussing something that (as we would say) operates at second level, is reinforced by the following passages from "Beweisgrund":
If I say, 'God is an existing thing,' it appears that I express the relation of
BV: But don't you see that this presupposes first-level existence? An existing thing is an individual that exists.
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
... [existence] appears in common usage as a predicate, not so much as a
predicate of the thing itself, as it is of the thought one has of it. E.g.
existence belongs to the sea-unicorn but not to the land-unicorn. This is to say
nothing more than: the conception of the sea-unicorn is a concept of experience,
that is, the conception of an existing thing . . . Not: regular hexagons exist
in nature, but: the predicates which are thought together in a hexagon belong to
certain things in nature. (76-77)
In the first of these passages Kant points out that when we say 'God is an existing thing,' or just 'God exists,' the grammatical structure of the sentences we utter makes it look as though we are ascribing the predicate, "being an existing thing," or the predicate of existence, to God. But this appearance, he claims, is misleading. What we are actually doing is ascribing the predicate, "belonging to an existing thing," to the predicates of divinity; we are not saying that God has a first-level predicate of existence, but rather that the (first-level) predicates of omnipotence, omniscience, etc., belong to an existing thing. In short, we are not ascribing a first-level predicate to God, but rather ascribing the second-level predicate, "belonging to an existing thing," to some first-level predicates.
We find a similar theme in the second of the two passages just quoted. With his hexagon example Kant seems to be making the claim that when we say 'regular hexagons exist in nature' or just 'regular hexagons exist,' we are not ascribing a first-level predicate of existence, or the predicate "existing in nature" to hexagons, but instead are ascribing a second-level predicate to those predicates included in our concept of a hexagon. We are saying that those predicates "belong to certain things in nature." In the sea-unicorn example Kant appears to be making a slightly different claim. When we say 'sea-unicorns exist' we are not ascribing a predicate of existence to sea-unicorns, but are instead ascribing the second-level predicate, "being a concept of experience," or "being a concept which applies to an existing thing," to the concept of a sea-unicorn.
Although they differ in detail, Kant's three examples appear to make the same general point, namely, that in claiming that something exists we are not ascribing a first-level predicate of existence to some thing but are rather ascribing a second-level predicate to something, whether it be a concept or a collection of predicates. And it is this second-level property that Kant has in mind when he talks of existence. For the several examples are intended to illustrate the claim with which the second passage begins, the claim that existence is a predicate alright, not of things but rather of the thoughts we have of things. (His examples indicate he could say, instead of "thoughts," "concepts," or "a collection of predicates.")
The difference between G-existence and K-existence can be brought out in the following way. Notice how Kant at one time characterizes the second-level predicate at issue. He says it is the predicate of "belonging to an existing thing". But the reference to an "existing thing" appears to invoke G-existence. We might just as well describe the second-level predicate as that of "belonging to things which exist, i.e., to things which have the first-level property, or pseudo-property, of existence". K-existence is a second-level predicate, and it is something that can be explicated using Gassendi’s first-level notion of existence.
Kant’s claims about existence should not be likened to Gassendi’s. If we are looking for an historical parallel, we need to look ahead to Frege. In the Foundations of Arithmetic, Frege writes:
… the proposition that there exists no rectangular equilateral triangle does
state a property of the concept "rectangular equilateral triangle;" it assigns
to it the number nought. In this respect existence is analogous to number.
Affirmation of existence is in fact nothing but denial of the number nought.
Because existence is a property of concepts the ontological argument for the
existence of God breaks down.
According to Frege when we make an affirmative existence assertion we deny of a concept the number nought. Similarly, to make a negative existence assertion is to ascribe to a concept the number nought. It is clear from other passages that to assign or deny the number nought to a concept is to say, respectively, that nothing, or something falls under that concept:
… the number 0 belongs to a concept, if the proposition that a does not fall under that concept is true universally, whatever a may be. In "On Concept and Object," Frege makes a similar point. To make an affirmative existence assertion is to say of a concept that it is not empty:
I have called existence a property of a concept. . . . In the sentence ‘there is
at least one square root of 4’, we have an assertion not about (say) the
definite number 2, nor about -2, but about a concept, square root of 2, viz.,
that it is not empty.
Like Kant, Frege maintains that when we make existence assertions we are really ascribing or denying a second-level property to a concept. The second-level property in question, i.e., the property "not being empty," is evidently what Frege means by existence. For, again like Kant, Frege takes his examples about existence assertions to illustrate his claim that existence is itself a property of concepts, i.e, a second-level property.
I have been attempting to distance Kant’s views on existence from those of Gasssendi. But the following objection might be made to what I have said so far. Suppose we agree that Gassendi is talking about G-existence and that Kant, at least in the passages we have examined, is talking about something quite different, K-existence. But what Kant says about K-existence is perfectly compatible with what Gassendi says about G-existence. One could maintain that K-existence is a second-level predicate, not belonging to objects, and also agree that G-existence is something operating at first-level, that it is necessary for an object to have or lack any properties at all and thus is itself not a property of those objects that have it. So just because there are passages in which Kant makes points different from those we have attributed to Gassendi, we cannot conclude that Gassendi’s claims are not to be found in Kant. And indeed there is a famous passage in the first Critique in which Kant might naturally be read as discussing G-existence and making Gassendi’s claims about it. So the distance between Kant and Gassendi may not that great, and those commentators who find Gassendi’s views in Kant may be right after all. They may have missed Kant’s discussion of K-existence, but perhaps they are correct in finding in Kant echoes of Gassendi’s discussion of G-existence. Perhaps the truth is that Kant is concerned both with K-existence and G-existence, claiming that neither is a predicate of objects. I believe the distance is greater than this objection suggests. Kant surely recognizes G-existence. And, as we saw earlier, he appears to invoke the notion of G-existence in his "Beweisgrund" discussion of what we are doing when we make existence assertions. Nevertheless, the evidence that Kant makes Gassendi’s claims about G-existence is weak and can be explained more plausibly in a different way. The first Critique passage in question is the following:
…the illusion which is caused by the confusion of a logical with a real
predicate (that is, with a predicate which determines a thing) is almost beyond
correction. Anything we please can be made to serve as a logical predicate; the
subject can even be predicated of itself; for logic abstracts from all content.
But a determining predicate is a predicate which is added to the concept of the
subject and enlarges it. Consequently it must not be already contained in the
concept. ‘Being’ is obviously not a real predicate; that is, it is not a
concept of something which could be added to the concept of a thing. It is
merely the positing of a thing, or of certain determinations, as existing in
It is possible to give this passage a Gassendian reading as follows. Kant tells us that a real or determining predicate is one which is added to a concept in such a way as to "enlarge" that concept. Perhaps we can suppose that something enlarges a concept if it creates a non-vacuous further ("additional") condition anything must meet to be an instance of that concept. Kant goes on to say that existence, or being, is not something which even "could be" added to a concept (and so, a fortiori, not something which even could enlarge a concept) of a thing. Why is this so? The only condition Kant explicitly mentions here that would disqualify something from being a real predicate, something which enlarges the concept of a subject, is that it is already contained in the concept. (This is relevant, presumably, because then the attempt to add that feature to a concept will not create a further condition for instantiation - it will merely duplicate, as it were, a condition already present.) But why would Kant suppose existence is already in any concept? One plausible answer to this question is that he believes existence is a necessary condition for a thing to have or lack any properties. Thus existence is a necessary condition for anything to instantiate any concept and in that sense is already in any concept.
BV: You've lost me entirely. Yes, existence is a necessary condition for an individual to instantiate a first-level concept. But this does not entail that existence is in any concept, neither for Gassendi nor for Kant. Gassendi's point was that existence cannot be a property of individuals. Concepts and properties are 1-1 as previously explained. So, there is no concept of existence that could be included in any concept.
Part of the problem may be that you muddied the waters right at the outset by bringing in the N & S article and suggesting that there were two ways of reading Gassendi. That was a mistake right there. As I argued above, for Gassendi, existence cannot be a concept of everything, whence it follows that existence cannot be a concept included in every concept.
The rigmarole you are presently engaged in is extremely confusing, and probably not necessary for the points you want to make.
Attempting to add existence to a concept will not create a new condition for instantiation. It will therefore not enlarge any concept and is thus not a real predicate. This is essentially Gassendi’s argument: because existence is necessary for a thing to have or lack any property, it is not itself a property. There is another way to read this passage, however, one on which it comports better with the other passages from Kant’s writings that we have considered. The clue comes from the end of the passage. Kant says existence is "the positing of a thing, or of certain determinations, as existing in themselves." This remark should be heard in conjunction with what Kant says later on in the same paragraph:
If, now, we take the subject (God) with all its predicates … and say "God is,’
or ‘There is a God,’ we … only posit the subject in itself with all its
predicates, and indeed posit it as being an object that stands in relation to my
Both passages are plausibly seen as making the point, originating in "Beweisgrund," that when we say that a thing exists we are asserting of (e.g.) a collection of predicates that those predicates are exemplified by, or belong to, an existing thing, i.e., that a certain concept is instantiated. And we should recall that Kant makes that point about what we are doing in making assertions of existence as a way of illustrating his claim that existence is a second-level predicate, a predicate of a concept or of a collection of predicates.
BV: It should be clear that the existence of an individual cannot possibly be a second-level property. Being instantiated is a second-level property. Apparently, you are not an eliminativist about the existence of individuals as Frege is. You presuppose that individuals exist, and you interpret Kant is this way. So this is a respect in which Kant does NOT anticipate Frege.
This provides a nice way of explaining why Kant says existence is not something that can be added to the concept of a thing: it cannot be added to the concept of a thing, not because it is already in any concept of a thing, but because it is not a predicate that belongs to things (it is not first-level), but rather to concepts of things. (Recall the "Beweisgrund" remark that existence is used "not so much as a predicate of the thing itself, as it is of the thought one has of it.")
BV: There are three ways it could be true that "existence is not something that can be added to the concept of a thing," namely, (a) existence is 'automatically' included in any concept of anything; (b) existence is not a first-level concept; (c) existence is not a concept at all: it is that which an individual must have if it is to fall under any (first-level) concepts, and instantiate any properties.
On my interpretation of Kant, he is not saying that existence does not belong to individuals, but that it does not belong to individuals in the manner of a property. My point could be put this way: existence is 'first-level' without being a first-level property.
You cannot assimilate Kant to Frege because for Frege it is meaningless to predicate existence of individuals, whereas for Kant it is not meaningless. Both Kant and Frege reject the OA, but for different reasons. For Frege, singular existentials such as 'God exists" are meaningless. But Kant never says or even implies that. Indeed he implies the opposite inasmuch as existence is one of the categories of the understanding: existence is a category applicable to phenomenal individuals. Kant rejects the OA on the ground that existence is not a quidditative property; it cannot enter into a description of what a thing is. Equivalently, no concept, not even the God-concept, is such as to include existence. If a concept includes existence -- in Frege's jargon has existence as a Merkmal as opposed to an Eigenschaft -- then existence is a concept. But if existence is a concept, then existence is a quidditative property of individuals. But existence is not a quidditative property of individuals.
Someone wanting to find Gassendi’s point in Kant might look at Kant’s words, "it is not a concept of something which could be added to the concept of a thing" and focus
on "added to." But this sentence fits much better into Kant’s overall discussion of existence if we focus on "of a thing." If one looks at the B626 passage in isolation, the Gassendian reading will not seem unnatural. But when we realize that, starting with "Beweisgrund," Kant has not focused on G-existence but instead on something that operates at second-level, the Gassendian reading of this single, isolated passage, can be resisted.
In the history of the discussion of whether existence is a predicate, two different things have been referred to as existence. Consequently at least two different claims have been made with the words "existence is not a predicate (property)." Gassendi is talking about something that operates at first-level. We can debate whether or not G-existence is necessary for a thing to have (or lack) any predicates at all.
BV: That's the issue that divides the Meinongians from the anti-Meinongians. I'm an anti-Meinongian: if x has properties or stands in relations, then x exists. Contrapositively, if x does not exist, then x cannot have properties or stand in relations. For me, these are both necessary truths, where the necessity in question is 'broadly logical,' i.e., metaphysical, not narrowly logical.
And if we decide it is necessary for a thing to have or lack any predicates, we can debate further whether that means it is itself not a predicate of those things that have it. But Kant is talking about something different. He is discussing what we can variously describe as the predicate "being instantiated," or "not being empty," or "belonging to an existing thing." K-existence is something that operates at second-level. It can be explicated in terms of the first-level notion Gassendi is talking about.
BV: What does this explication look like? Presumably you mean that first-level concept C is instantiated iff existing individuals fall under C. But then you are presupposing first-level existence. That is a reason why Kant does NOT anticipate the Frege-Russell view.
Kant’s originality in this area is not sufficiently appreciated. He is not echoing claims made over a century earlier by Gassendi. Instead, he is the first to articulate a view more commonly thought to have originated over a century later with Frege.
BV: With all due respect -- and I apologize for any harshness in the tone of the above remarks -- I think you are making the mistake of thinking that Kant has one single unified doctrine of existence that can be isolated and set forth. I say there is none, not on this topic, nor on many other Kantian topics. There is no denying that there is a definite Fregean flavor in that Beweisgrund passage you cited. But it is clear to me that Kant's concern is with singular existence, the existence of individuals, and that he is not an eliminativist about singular existence as are Frege and Russell.
I have more to say, but I will leave it for our future discussions, which I look forward to.
J. William Forgie
University of California