The following is a discussion of certain kinds of arguments for the existence of God. The distinctive feature of the arguments - at least according to the traditional Kantian method of classification - is that they proceed from premises which at least some defenders of the arguments allege can all be known a priori. Consequently, it would be most appropriate to call these arguments ‘a priori arguments for the existence of God”. However, following Kant, it has been established practice to call these kinds of arguments “ontological arguments,” and I see no urgent reason to depart from this tradition. Many things have well-established but inappropriate names - for example, the Holy Roman Empire, which, as Voltaire pointed out, was neither holy, nor Roman, nor an empire.

I shall divide the arguments that I consider into six classes, namely, (i) definitional arguments, whose premises invoke certain kinds of definitions; (ii) conceptual arguments, whose premises advert to the possession of certain kinds of concepts or ideas; (iii) modal arguments, whose premises advert to certain possibilities; (iv) Meinongian arguments, whose premises invoke a distinction between different categories of existence; (v) experiential arguments,whose premises include the assumption that the concept of God is only available to those who have had veridical experiences of God; and (vi) “Hegelian” arguments, which, at least in my view, bear some relation to the philosophy of Hegel.

The best-known - most often cited - objection to ontological arguments is encapsulated in the Kantian slogan ‘Existence is not a predicate’. I have already argued that Kant himself did not provide an adequate explanation - justification - of the critique that he supposed to be embodied in this claim. In this chapter, I propose to argue that the many subsequent attempts to explain the content of the claim have fared no better. I shall also suggest, albeit tentatively, that there is a straightforward sense in which existence is a predicate - and that this sense is all that defenders of ontological arguments need to invoke in order to defend themselves against the neo-Kantian onslaught.

REAL PREDICATES

As Barnes (1972) notes, the point of the slogan ‘Existence is not a predicate’ is not claim that finite parts of the verb ‘to exist’ do not function as grammatical predicates in sentences that are in subject predicate form. After all, that claim would seem to be falsified immediately by the existence of sentences of the form ‘a exists’ where a belongs to the category of singular terms, namely, proper names, definite descriptions, demonstratives, indexicals, pronominal compound expressions. Barnes suggests that this argument should be treated with care, since grammarians have no agreed canons for the division of sentences into subject and predicate.

Consider the following arguments, which I take to be paradigmatic cases of definitional ontological arguments:

1. God is a being which has every perfection. (Definition)

2. Existence is a perfection. (Premise)

3. (Hence) God exists. (From 1,2)

1. God is an existent supremely perfect being. (Definition)

2. (Hence) God exists. (From 1)

It seems clear that there is no good reason for an agnostic to be persuaded by these arguments. Why not? First, it should be noted that these arguments can be paralleled to their detriment. Corresponding to the first argument, there are many arguments of the following form:

1. God* is a being that has all perfections except for moral perfections - and the moral attributes of God* are as follows:.… (Definition)

2. Existence is a perfection. (Premise).

3. (Hence) God* exists. (From 1,2)

And corresponding to the second argument, there are many arguments of the following form:

1. m is an existent Ϝ. (Definition)

2. (Hence) m exists. (From 1)

Thus, parallel to the second argument, there are arguments that purport to establish the existence of existent unicorns, existent round squares,and so on. And parallel to the first argument, there are arguments that purport to establish the existence of a vast range of distinct all-but-perfect beings. The existence of these parallels strongly Ontological arguments and belief in God suggests that there is something wrong with the thought that these arguments ought to be recognized by agnostics as sound proofs of the existence of God.

A simple version of an experiential ontological argument is the following:

1. The word ‘God’ is not susceptible of an explicit definition but is a term whose meaning can only be had on the basis of religious experience. (Premise)

2. A body of experience adequate as a basis for an understanding of this term must also be adequate as an evidential basis for assent to the proposition that God exists. (Premise)

3. (Hence) A denial of God's existence is indicative of a failure to grasp the meaning of the word ‘God’. (From 1,2)

Rescher (1959a) notes that there seem to be words of which it is plausible to claim that their meaning can only be grasped on the basis of experience. In particular, this seems true of words that designate experiences or aspects of experience - words for perceptual contents (colors, odors), words for sensational contents (aches, pains), words for feelings (alacrity, lethargy), and words for emotions (anger, delight). But, given that there are some words of which it is true to say that their meaning can only be grasped on the basis of experience, why should it be thought unreasonable to hold that the meaning of the word ‘God’ can only be had on the basis of religious experience?

There are a number of observations that need to be made.

As things stand, we don't really have here an a priori argument for the existence of God.

This book is part of a larger work, which seeks to expound and defend an agnostic stance on theistic matters. The larger work, most of which remains to be written, is intended to have four parts:

1. i. an examination of conceptions of deities, including a discussion of the consistency and mutual compatibility of allegedly divine attributes, and an examination of the structure and function of religious vocabulary;

2. ii. an examination of traditional arguments for and against the existence of various deities, including: ontological arguments, cosmological arguments, teleological arguments, arguments from evil, moral arguments, arguments from revelation, arguments from authority, arguments from religious experience, and arguments from miracles;

3. iii. an examination of formulations of agnosticism, including a comparative evaluation of formulations of theism and atheism; and

4. iv. an investigation of the epistemological merits of agnosticism, including a somewhat qualified defense.

Obviously, the present work is one subpart of part (ii). Not surprisingly, the main thesis that is defended in it is that ontological arguments do not provide an agnostic with any good reason to change her view - that is, to give up her agnosticism. One of the projected themes of the larger work is that the most fruitful approach to arguments in philosophy of religion is to ask: Do these arguments embody reasons for agnostics - atheists, theists - to change their views? That is, do these arguments embody considerations that reasonable and reflective agnostics - atheists, theists - must recognize as providing motivating reasons for them to change their views?

I am not confident that what I have to say in this section is an accurate representation of the thought of Hegel, or of any subsequent Hegelians. However, what I provide is a characterization of certain kinds of a priori arguments for the existence of God that might perhaps be arrived at on the basis of a reading of some of the work of, say, R. G. Collingwood or E. E. Harris. There are three kinds of arguments: (i) an argument that, albeit tendentiously, can be traced back to Kant's Critique of Pure Reason; (ii) an argument that can be traced back to some ideas in Plato; and (iii) an argument that can be traced back to some ideas in Husserl and other continental phenomenologists.

THE NEO-KANTIAN ARGUMENT

In this section, I discuss an argument that can be traced back to Kant. This argument is a kind of transcendental deduction of the existence of God. In presenting this argument, I draw on Harris (1972X1977), though I doubt that he would approve of the use that I have made of his work.

In The Critique of Pure Reason, Kant provides the materials for an a priori proof of the necessity of belief in the existence of God. Very roughly, this argument goes as follows:

1. Perception of objects is possible only as a result of an a priori synthesis under the categories and the pure forms of intuition. (Premise)

2. […]

At the end of the preceding chapter, I concluded that there is a global objection to ontological arguments - that is, an objection that suggests in advance that no ontological argument can be dialectically effective. It is important to emphasize that it is not part of this objection to claim that no ontological argument can be sound. For all that I have said, it may be that there are sound versions of all ontological arguments.

Other philosophers have thought that it is possible to show in a single stroke, once and for all, either (i) that no ontological arguments can be sound or else (ii) that no ontological arguments can be thought to be sound on a priori grounds. In this chapter, I shall consider several such attempts. Of course, I have already argued that the Humean, Kantian, and logical positivist attempts to carry out this project fail. I shall not repeat those arguments in this chapter. Moreover, I shall hold over discussion of the claim that existence is not a predicate until the following chapter.

THE MISSING EXPLANATION ARGUMENT

The following line of argument is adapted from a more general argument in Johnston (1992). It is intended to establish that there could not be an ontological argument that provides anyone with a good reason to believe that God exists - that is, there could not be an ontological argument that could reasonably be taken to be sound on a priori grounds.

Throughout these chapters, I have often adverted to arguments that parody the ontological arguments under consideration. In this section, I wish to collect together the different sorts of parodies that might be made of ontological arguments and to consider the uses to which they might be put. I shall divide the initial discussion according to the different conceptions of God that are invoked in the arguments.

BEINGS OF KIND K THAN WHICH NO GREATER BEINGS OF KIND K CAN BE CONCEIVED

We have already seen that Gaunilo maintained that St. Anselm's argument could be paralleled with an argument that purports to establish the existence of an island than which no greater island can be conceived. Moreover, there seems to be nothing special about Gaunilo's choice of ‘island’ in this example: If he is right, it seems that St. Anselm's argument can be paralleled by an argument that purports to establish the existence of a being of kind K than which no greater being of kind K can be conceived, for any kind of object K.

Some kinds K of objects make for more plausible parallels to St. Anselm's ontological arguments than do others. In particular, given that it is a premise of the argument that a being of kind K than which no greater being of kind K can be conceived and that actually exists is greater than a being of kind K than which no greater being of kind K can be conceived but that does not actually exist, there are kinds that make for prima facie implausible parallels.

I begin with a synoptic history of ontological arguments. In the remaining part of this chapter, I shall discuss the works of some of the more prominent historical figures - St. Anselm, Descartes, Leibniz, Hume, and Kant - in more detail. Then, in the following chapters, I shall provide analyses of different kinds of ontological arguments, and show how these analyses relate to the historically important arguments.

HISTORICAL SYNOPSIS

Ontological arguments have been “found” in ancient Greek philosophy, in St. Augustine and other early Christians, and in the work of Jewish and Islamic philosophers prior to the eleventh century. However, it can be reasonably contended that the first clear statement of an ontological argument is that of St. Anselm, archbishop of Canterbury, in the eleventh century. St. Anselm's argument was much discussed throughout the succeeding centuries. Many prominent medieval thinkers accepted St. Anselm's argument – or modified versions of it – including Duns Scotus and St. Bonaventure. However, other significant medieval thinkers, including St. Thomas Aquinas and William of Occam, rejected the argument. One of the most important objections to St. Anselm's argument was provided by one of his contemporaries, the monk Gaunilo.

Ontological arguments received a fresh defense in the work of Descartes.

Despite the negative conclusions that can be drawn about the utility of ontological arguments as instruments of persuasion, some theists and atheists continue to see value in some types of ontological arguments. In particular, some theists hold that ontological arguments can be used to demonstrate certain facts about the rationality of their beliefs: namely, either (i) that theism is rational, (ii) that nontheism is unintelligible, or (iii) that theism is absolutely epistemically secure. Similar positions can be occupied by atheists. In this chapter, I shall concentrate on defenses that have been offered for the theistic views (i)-(iii); exactly parallel considerations will apply to the atheistic counterparts, but will go unmentioned.

There is one unproblematic possible use for ontological arguments that can be conceded from the outset. It might be that some ontological arguments - for example, ontological arguments involving necessity - can have a use in the systematic exposition of theistic positions. There can be no objection to the use of dialectically inefficacious proofs in the exposition of a view - for the aims of exposition and dialectic are very different. However, once this has been conceded, it should be noted that it seems very implausible to think that it is actually the case that the arguments discussed in the earlier part of this book are suitable candidates for such roles. A straightforward exposition of theism surely ought to begin with the assertion that God exists - rather than, for example, the assertion that it is possible that God exists - since this provides a more immediate picture of the commitments of the view.

There are four kinds of modal arguments that I shall consider in this section. The first kind turns on the use that can be made of the ‘actually’ operator; the second kind turns on the use that can be made of the ‘necessity’ operator; the third kind turns on a weak version of the principle of sufficient reason; and the fourth kind turns on the properties of incomprehensible beings.

MODAL ARGUMENTS INVOLVING ACTUALITY

A simple version of a modal ontological argument that involves the ‘actually’ operator is the following:

1. It is possible that it actually is the case that God exists. (Premise)

2. (Hence) God exists. (From 1)

There are modal logics in which this is a valid argument. In particular, if a “nonshifty” actually operator - that is, an actually operator that is such that even when it is embedded in the scope of other operators, it always refers back to the world of its utterance - is added to the modal logic S5, then the argument is valid.

However, if the actually operator is given a nonshifty reading, then there is no reason for an agnostic to assent to the premise. After all, on this reading, the possibility judgment in question just amounts to a judgment that God exists. Any reasonable agnostic will see equal merit in the following argument:

1. It is possible that it is actually the case that God does not exist. (Premise)

2. (Hence) God does not exist. (From 1)

Given the preceding system of classifying ontological arguments, and given the further assumption that there are no ontological arguments that do not belong to at least one of the categories in the system, we can provide a critique of the historically important ontological arguments that does not depend upon controversial points of interpretation of those arguments. The point of the present chapter is to show how this can be done. Some people will find the application of the foregoing absolutely straightforward; nonetheless, it may be worthwhile to provide a detailed summary of the preceding results.

THE ARGUMENT FROM PROSLOGION 2

There are a number of somewhat plausible interpretations of the argument in Proslogion 2. If we represent the argument in a logically inperspicacious way, then it has just two premises and a conclusion:

1. A being than which no greater can be conceived exists in the understanding. (Premise)

2. If a being than which no greater can be conceived does exist in the understanding but does not exist in reality, then a being than which no greater can be conceived that does exist both in the understanding and in reality is greater than a being than which no greater can be conceived. (Premise)

3. (Hence) A being than which no greater can be conceived does exist in reality. (From 1,2)

We can interpret this argument in three different ways: (i) as a conceptual argument, (ii) as a modal argument involving actuality, and (iii) as a Meinongian argument.

On the basis of the discussion in Chapters 2-7,1 conclude that there are no ontological arguments that provide me with a good reason to believe that God - a being than which no greater can be conceived, a most perfect being - exists. Furthermore, though I have not emphasized this point, I also conclude that there are no ontological arguments that provide me with a good reason to believe that God - a being than which no greater can be conceived, a most perfect being - does not exist. The ‘ontological disproofs’ offered as parallels to the ‘ontological proofs’ are no more worthy of my acceptance, given that I start from a position of agnosticism.

More generally, on the basis of the argument in Chapter 8, I conclude that there are perfectly general grounds on which I can dismiss the possibility of a dialectically effective ontological argument. Only those who make the relevant presuppositions will suppose that some ontological arguments are sound; but there is nothing in ontological arguments that establishes a case for those presuppositions from the standpoint of those who do not share them.

I also note that there may be various other grounds on which each ontological argument can be dismissed.

There is no easily stated version of the type of ontological argument that I call ‘Meinongian’. What these arguments have in common is that they rely upon, or assume, a theory of objects that, in relevant respects, is similar to the theory of objects defended by Meinong. Hence, these arguments cannot be understood in isolation from the theory of objects - and so I begin with a sketch of the relevant features of that kind of theory.

The theory begins with the assumption that there are properties. This assumption is intended to be understood in a way that does not exclude the possibility of nominalism - that is, it leaves open the possibility that talk of properties ought to be reconstrued as talk of predicates. The theory also assumes that predicate expressions in natural language uniquely express properties - so that, for example, each of the following expresses a different property: ‘is round’, ‘is round and square’, ‘is square and round’, ‘is taller than Dudley Moore’, ‘prefers the squalid life of a stockbroker to the noble life of a politician’.

Against this background, objects are specified by, or perhaps identified with, unordered collections of properties. Any distinct collection of properties specifies a distinct object. Thus, for example, the property ‘is round and square’ specifies a distinct object from the property ‘is square and round’.