On “The Ontological Argument”: A Response To Makin (1991)
Graham Oppy
[This article was originally published as “Makin On The Ontological Argument”, Philosophy, 66, 255, January 1991, pp.106-114]
In “The Ontological Argument”, (Philosophy 63, 1988, pp.83-91) Stephen Makin offers a defence of what he calls “Anselm’s Ontological Argument”. I am not much interested in the question whether the argument which Makin defends can properly be attributed to St. Anselm, though I suspect that there is considerable room for disagreement on this score; rather, I want to suggest that the argument which Makin offers is quite clearly invalid (and hence unsound)–and I also want to suggest that it is very plausible to suppose that any version of “the ontological argument” is vitiated by the same fallacy in which Makin’s argument is entrapped.
At the core of Makin’s paper is the defence of the following principle:
(Principle A): Iff F is a necessarily exemplified concept, and G is not, then F’s are a greater kind of thing than G’s.
In particular, Makin defends the claims: (i) that it is possible to make sense of the notion of a necessarily exemplified concept; (ii) that it is possible to support Principle A–taken on the strong reading according to which necessarily exemplified concepts must be exemplified by at least one being which exists in all possible worlds–by referring to the sorts of attitudes which it would be appropriate to have only towards necessarily existent entities; and (iii) that it is possible to defend Principle A even on the weak reading–according to which necessarily exemplified concepts must be exemplified in every possible world by at least one being (though not necessarily the same being in different worlds)–by referring to the same sorts of attitudes which are referred to in (ii). (I should point out that I have made one modification to Makin’s principle; for, in his original paper, Principle A is a conditional whereas, in my version, it is a biconditional. This modification is needed, for otherwise–as Andrew Gleason pointed out to me–Makin’s argument is guilty of the fallacy of “affirming the antecedent”. However, I am fairly confident that Makin intended that Principle A should be given the biconditional understanding–and, moreover, that is exactly how I understood Principle A when I read his paper.)
I do not propose to take issue with Makin’s defence of Principle A. It is clear that Principle A plays the same role in Makin’s argument that the claim that “existence is a perfection” (or perhaps “necessary existence is a perfection”) plays in the ontological arguments of St. Anselm and Descartes–and, moreover, it is clear (to me, at least!) that Makin’s principle is no more and no less plausible than these earlier, and much venerated, principles. I do not think that any of these claims is particularly plausible–since I do not see that there is any special virtue which attaches to entities which exist in our world (as opposed to entities which exist in other worlds) merely in virtue of the fact that those entities exist in our world; and nor do I see that there is any special virtue which attaches to entities which exist in all worlds (as opposed to entities which exist in only some worlds) merely in virtue of the fact that those entities exist in all worlds–but I do not propose to argue for this opinion here. Rather, I want to argue that the argument which Makin gives after he has–in his view–established the truth of Principle A is invalid (and hence unsound).
(Perhaps I should add that the part of Makin’s argument which I have considered thus far has been criticised before–cf. P.J. McGrath “The Ontological Argument Revisited” Philosophy 63, 1988, pp.529-533. However, I do not think that any of the objections which McGrath gives are convincing.
McGrath’s first objection is that Makin fails to show that talk of “necessarily exemplified concepts” makes sense. However, as McGrath’s subsequent discussion shows, what McGrath means by saying that a concept “makes sense” is that the concept is “exemplified in at least one possible world”. (Thus: the notion of “the greatest prime number” makes no sense.) I see no reason to follow McGrath in making this assimilation of the notions of “exemplified in at least one possible world” and “makes sense” if one is also to suppose, as McGrath seems to do, that the underlying modal logic is S5.
McGrath’s second objection is that Makin fails to show that his own talk of “necessary beings” makes sense. This objection can also be met by insisting that the discussion of ontological arguments needs to be carried out in the context of a modal logic which allows that accessibility relations between worlds are not–e.g.–symmetric (so that one can say that it is possible for a state of affairs to be necessary and yet for it not to be the case that that state of affairs actually occurs).
McGrath’s last objection is that “to claim that a necessary being is eo ipso greater than a contingent being seems totally counter-intuitive, for there appears to be no good reason for thinking that necessary existence must outweigh all the other great-making characteristics which a being might possess” (p.532). However, Makin’s suggestion is that, if two beings have exactly the same properties in our world except for the fact that one is necessarily existent and the other is not (and any other differences that this difference entails), then the one which is necessarily existent is greater. Given the correct understanding of the underlying modal logic, there is no problem in understanding this claim–and, moreover, it seems to be just the principle which Makin needs for his argument. Of course, there is still a further question whether this principle is true–but, as I said before, I do not intend to try to defend my intuition that this principle is false here.)
Here is what Makin says:
I do not see how, if Principle A is granted, the conclusion of the argument can be resisted. After all, if the concept S (something-than-which-nothing-greater-can-be-conceived) is of a kind of thing such that no greater kind of thing can be conceived, and if the fact that F is a necessarily exemplified concept shows that F’s are a greater kind of thing than otherwise (which is what we get from Principle A), then the concept of a kind of thing such that no greater kind of thing can be conceived must be a necessarily exemplified concept. (p.90)
In order to see that there is something wrong with this argument, we need only to follow the lead of a recent disciple of the monk Gaunilo (Michael Tooley, in “Plantinga’s Defence Of The Ontological Argument, Mind 90, 1981, pp.422-427), by arguing in the following way: Consider the concepts S1(a-solvent-than-which-no-greater-solvent-can-be-conceived) and S2 (an-insoluble-substance-than-which-no-greater-insoluble-substance-can-be-conceived). If S1 and S2 are concepts of kinds of things that which no greater things of that kind can be conceived, and if it is true that the fact that F is a necessarily exemplified concept shows that F’s are a greater kind of thing that otherwise (which is what we get from principle A), then S1 and S2 must be necessarily exemplified concepts. But surely (i) a solvent than which no greater can be conceived will dissolve everything; (ii) an insoluble substance than which no greater can be conceived will dissolve in nothing; (iii) if S1 and S2 are necessarily exemplified, then there is a solvent which can dissolve everything and an insoluble substance which will dissolve in nothing; and (iv) it is impossible for there to be a solvent which can dissolve everything and an insoluble substance which will dissolve in nothing. So, the very sort of reasoning which Makin uses leads us into contradiction–and hence should be repudiated.
However, even if it is granted that this objection to Makin’s argument is sound, there is still a puzzle, namely: Where exactly does his argument go wrong? This is the question which I wish to take up in the present paper.
I
The key to isolating the fallacy in which Makin’s argument is involved is to draw a distinction between two different senses in which it can be correct to say that a concept C is a “concept of something”, and which I shall call “the two senses of concept-of-ness”. (Throughout this discussion, I shall ignore the distinction between individual concepts and general concepts, and speak as if it were the case that all concepts are individual concepts. There is an obvious way of extending the distinction which I draw to cover the case of general concepts as well; however, this extension is not relevant to the use which I wish to make of the distinction.)
In the first of these senses, which I shall call “the extensional sense of concept-of-ness”, a concept C is a “concept of something” iff the actual extension of the concept is non-empty. In other words, a concept C is a concept of something in the extensional sense of concept-of-ness provided that there is something in the actual world which falls under the concept C. So, in this sense, the concept “The current President of The United States” is, in the extensional sense of concept-of-ness, a concept of George Bush. And, on the other hand, in this sense, the concept “The least rapidly converging sequence of natural numbers” is not, in the extensional sense of concept-of-ness, a concept of anything–for there is no least rapidly converging sequence of natural numbers.
In the second of these senses, which I shall call “the intensional sense of concept-of-ness”, a concept C is a “concept of something” iff the concept C is a concept of a certain intensionally characterisable sort. In this intensional sense of concept-of-ness, the concept “The current President of The United States” is a concept of the current President of The United States precisely because it is a current-President-of-The-United-States-type of concept. Similarly, in this intensional sense of concept-of-ness, the concept “The least rapidly converging sequence of natural numbers” is a concept of the least rapidly converging sequence of natural numbers precisely because it is a least-rapidly-converging-sequence-of-natural-numbers-type of concept.
For the purposes of the current paper, the most important point to note about the distinction between these two sorts of concept-of-ness is that, in the case of the extensional sense of concept-of-ness, whether a concept is of anything is an external or extrinsic question: it is a question of whether there is anything which stands in the right sort of relation (the “falls under” relation or the “satisfies” relation) to the concept in question. However, in the case of the intensional sense of concept-of-ness, whether a concept is of anything is an internal or intrinsic question: it is a question of whether or not the concept belongs to a non-relationally characterisable kind. (In this latter sense, it seems that every concept will be of something, since it seems that every concept will belong to a least one non-relationally characterisable kind.) No doubt more could be said in order to clarify the ground of the distinction which I have drawn; however, it is I think sufficiently clear to enable us to proceed.
Let us return to Makin’s argument. Roughly, Makin’s argument is as follows:
Consider the concept S = the concept: something than which no greater can be conceived. By Principle A, if S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived. But S is exactly a concept of something than which no greater can be conceived. That is, it follows from Principle A that the concept S is necessarily exemplified. And so it follows, quite plainly, that there is something than which no greater can be conceived.
Given the distinction between the two senses of concept-of-ness described above, the fallacy in this argument is evident. On the one hand, the claim that “By Principle A, if S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived” is naturally understood in terms of the extensional sense of concept-of-ness: if there is not some actually existing object which exists in all worlds (or which is of a kind of thing which is instantiated in all worlds) and which “falls under” or “satisfies” the concept something-than-which-no-greater-can-be-conceived, then S is neither exemplified nor necessarily exemplified–and so S is not a concept of something than which no greater can be conceived in the extensional sense of concept-of-ness. But, on the other hand, the claim that “S is exactly a concept of something than which no greater can be conceived” is naturally understood in terms of the intensional sense of concept-of-ness: the concept S is a something-than-which-no-greater-can-be-conceived-type of concept (under the intrinsic or non-relational method of characterising types of concepts). And if these two claims are understood in the ways which I have suggested is most natural, then Makin’s conclusion simply does not follow–for there is an obvious equivocation on the sense of concept-of-ness which is required in order to reach the conclusion that “It follows from Principle A that the concept S is necessarily exemplified”.
Now, in the preceding paragraph, I spoke throughout about “the most natural way” of construing certain claims which are essential to Makin’s argument. Consequently, it might be wondered whether it is possible to impose a uniform construal of the notion of concept-of-ness throughout the argument (thus avoiding the problem of equivocation which I have suggested undermines “the most natural way” of understanding Makin’s argument) and yet to end up with a sound argument. Well, there are two possibilities: either we construe concept-of-ness in an extensional sense throughout or else we construe concept-of-ness in an intensional sense throughout.
If we construe concept-of-ness in an extensional sense throughout, then we must construe the claim that “S is exactly a concept of something than which no greater can be conceived” in terms of the extensional sense of concept-of-ness. But, in that case, we are assuming that there actually is something than which no greater can be conceived–i.e. we are asserting, without any supporting argument, that there is an entity in the actual world which “falls under” or “satisfies” the concept S–and so we are simply begging the question. (Of course, given this assumption, it is perfectly correct–albeit trivial–to apply Principle A in order to reach the conclusion that the (postulated) entity in question is either “necessarily existent” or else “of a kind which is necessarily instantiated”; but what we want to know is whether the (postulated) entity actually exists, i.e. whether there is such a “necessarily existent” entity.)
On the other hand, if we construe concept-of-ness in an intensional sense throughout, then we must construe the claim that “By Principle A, if S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived” in terms of the intensional sense of concept-of-ness. There are now two possible outcomes, depending upon two different construals which can be given to Principle A.
On the one hand, if we construe Principle A in an extensional sense–according to which it tells us that entities which actually instantiate concepts which are of necessary existents (or necessarily instantiated kinds) in the extensional sense of concept-of-ness are greater than entities which do not instantiate concepts of this sort–then the claim that “If S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived” is obviously unsupportable, except in a question-begging way: for S is a concept of something than which no greater can be conceived in the intensional sense of concept-of-ness merely because it is a something-than-which-no-greater-can-be-conceived-type of concept (under the intrinsic or non-relational method of characterising types of concepts); and yet the question of whether S is a necessarily exemplified concept turns on the question of whether S is appropriately related to entities which exist in the actual world.
However, on the other hand, if we construe Principle A in an intensional sense–according to which it tells us that concepts which are of necessary existents (or necessarily instantiated kinds) in the intensional sense of concept-of-ness are greater than concepts which are not of this sort–then, while the claim that “If S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived” is now true (since a something-than-which-no-greater-can-be-conceived-type of concept is a necessarily-exemplified-type of concept, under the intrinsic or non-relational method of characterising types of concepts), we are now faced with the problem that the conclusion of Makin’s argument no longer follows from the previous claim. For, from the fact that a concept is a necessarily-exemplified-type of concept (under the intrinsic or non-relational method of characterising types of concepts), nothing at all follows about whether there is an entity in the actual world which “falls under” or “satisfies” that concept. (Recall the earlier example of the perfect solvent and the perfectly insoluble substance. Both of these are instances of the necessarily-exemplified-type of concept–but we know that they can’t both be instantiated in the actual world.)
Thus, if we construe the claim that “By Principle A, if S is not a necessarily exemplified concept, then S is not a concept of something than which no greater can be conceived” in terms of the intensional sense of concept-of-ness, then, on either possible construal of Principle A, Makin’s argument does not go through. But we saw earlier that, if we construe the claim that “S is exactly a concept of something than which no greater can be conceived” in terms of the extensional sense of concept-of-ness, then it is also the case that Makin’s argument does not go through. Moreover, in order to avoid equivocation on the sense of concept-of-ness which is required by the argument, Makin must construe at least one of these claims in such a way that his argument does not go through. So, I conclude that his argument does not go through. (Furthermore, I conclude that, on the most plausible construal of the argument, the argument involves an equivocation, and hence is informally invalid.)
II
It seems to me that the distinction between the extensional and intensional senses of concept-of-ness can be used to show that no ontological argument for the existence of God (or for the existence of something than which no greater can be conceived) can possibly be made to work. However, before I turn to my argument for this conclusion, I want to introduce a heuristic device which may help to clarify thought on these matters.
So far, I have followed Makin in speaking about “concepts”. However, it seems to me that there is nothing which is lost if one chooses instead to speak about “descriptions”. Moreover, if one likes, one can think of “concepts” in terms of “descriptional sentence tokens” which are quite literally located inside the heads of cognitive subjects. Whether this is psychologically and/or scientifically plausible is unimportant; what matters is that this heuristic picture does not misrepresent any of the important features of the debate about the validity of ontological arguments.
Now, consider a person who has a token of the description “something than which no greater can be conceived” inside her head. From our point of view, there seems to be no reason to be tempted by any sort of ontological arguments which begin from the existence of this token of the description in her head; after all, the fact that she has a certain sort of non-relationally characterisable type of sentence in her head seems to have no logical bearing on the question whether there is actually an entity outside her head which is “satisfactionally” related to that description-token. Moreover, the situation does not seem to be substantially different when seen from her own point of view: the fact that she has a certain sort of non-relationally characterisable type of sentence in her head surely has no logical bearing on the question whether there is actually an entity outside her head which is “satisfactionally” related to that description-token. Of course, it may be that there is something out there which satisfies the description: but the point is that there is surely no way in which one can argue from the mere existence of the description-token to the existence of that external entity.
Perhaps there will be some who do not share the intuition which I expressed at the end of the last paragraph. After all, the hope of those who have defended ontological arguments is that there is some way of arguing from the non-relational properties of a description-token to a relational property of that description-token. But I fail to see any reason at all to maintain this hope. For consider the following thought experiment: It certainly seems to be possible to imagine a world in which the only thing that exists is a head with a token of the description “something than which no greater can be conceived” inside it. But, in this world, it would simply be a mistake to argue from the non-relational properties of the description-token to the conclusion that there is something in the world which satisfies the description–for there is nothing in the world which satisfies the description. However, since such a world is a logical possibility, it is equally a mistake to make this argument in our world.
Now, to this, it may be objected that I have begged the question against the existence of any necessary beings. However, this objection misses the point: for, since there is prima facie good reason to suppose that such a world is logically possible, the defender of necessary beings requires some argument to show that there can be no such world. But, in view of the considerations raised in the first section of this paper, it seems clear that no ontological argument could supply this sort of support.
Finally, I can now explain why I think that no ontological argument can be made to work. The essential strategy of ontological arguments is to begin by noting that one can construct a certain sort of mental description-token–e.g. “something-than-which-none-greater-can-be-conceived” — and then to suppose that the denial that there is anything in the world which satisfies or falls under that description-token requires one to construct a further mental description-token–e.g. “something-than-which-none-greater-can-be-conceived-and-which-does-not-exist”–which one must then suppose is satisfied by something in the world, and yet which cannot possibly be satisfied by anything in the world. However, to argue in this way is to be guilty of commiting the fallacy which I discussed in the first part of this paper–for the move to the claim that the description-token “something than-which-none-greater-can-be-conceived-and-which-does-not-exist” must be satisfied by something in the world involves an illicit move from the intensional sense of concept-of-ness to the extensional sense of concept-of-ness. Moreover, this explanation of the failure of ontological arguments does justice to the intuition which is shared by many opponents of ontological arguments, viz: that it is evidently impossible to build a bridge between the non-relational properties of an entity and the relational properties of that entity. And, finally, it should be noted that, in the particular case mentioned here, the denial that there is actually anything which satisfies the description-token “something-than-which-no-greater-can-be-conceived” only requires the construction of a mental sentence of the form “There is no thing-than-which-no-greater-can-be-conceived” or “No thing-than-which-no-greater-can-be-conceived exists”. It is obvious that these mental sentences can quite happily co-exist with mental sentences of the form “I can form the mental description: something-than-which-no-greater-can-be-conceived”.
“On ‘The Ontological Argument’: A Response To Makin” is copyright © 1993 by Graham Oppy. All rights reserved.
The electronic version is copyright © 1998 by Internet Infidels with the written permission of Graham Oppy. All rights reserved.