Arguments for the Existence of God

Arguments for the Existence of God
C. D. Broad

Published in the Journal of Theological Studies 40 (1939): 16-30; 156-67. Reprinted in C. D. Broad, Religion, Philosophy and Psychical Research (Routledge & Keagan Paul, 1953).

      From the time of Plato at latest learned men in Europe have excogitated a considerable number of arguments which have the following feature in common, viz. that they would be described as 'professing to prove the existence of God'. I purposely delimit the class of arguments which I have in mind by this purely verbal description because the word 'God' has been used in so many different senses by different thinkers. It is doubtful whether there is anything common and peculiar to all these arguments except that the conclusions of all of them are stated in the verbal form 'God exists' or in a translation of this into Greek or Latin or some other tongue.
      The object of the present paper is to classify these arguments and then to discuss in some detail a selected few of them. The selection will be made in respect of philosophical interest or historical importance, and the arguments chosen will be discussed with a view to determining what precisely they would prove, if they were valid, and whether they are in fact valid. An argument may fail to prove its conclusion either through its premisses being doubtful or through its structure being logically defective. Nevertheless, the persons who employed it may have had something of importance at the back of their minds, and the criticism which shows the argument to be invalid may incidentally separate this grain of wheat from the chaff which has surrounded it.

Classification of the Arguments
      I begin with a very external and superficial division of these theistic arguments, viz. into those which are closely bound up with the peculiar doctrines of some particular philosophic system and those which are not. An example of the former is Berkeley's argument that God must exist in order to cause those bundles and sequences of correlated sensations which plain men mistakenly believe to be manifestations of material things. I shall neglect these highly special arguments and confine myself henceforth to more general ones which have been very widely used and accepted.
      We may divide up the latter arguments, to begin with, in accordance with the nature of their premisses This gives the following three classes:

Arguments whose premisses include neither specifically ethical nor specifically religious propositions;
Arguments whose premisses include specifically ethical but not specifically religious propositions;
Arguments whose premisses include specifically religious propositions.

      The first class may be subdivided into
(1.1) arguments which do not, and
(1.2) those which do, use an existential premiss, i.e. a proposition of the form So-and-so exists.
There is one and only one argument which does not use an existential premiss. This is the famous Ontological Argument, invented by St. Anselm of Canterbury. Arguments which do use an existential premiss always employ also some form of the notion of causation. For the only way in which one can infer that X exists from the fact that Y exists is by showing that the existence of X is a necessary condition for the existence of Y. These latter arguments fall into two sub-classes, according to the nature of the existential premiss that they use. These are
(1.21) arguments which use only the highly indeterminate premiss Something or other exists, and
(1.22) those which use a more determinate premiss of the form Something having such and such a nature exists.
It is plain that a premiss of the latter kind is needed if one is to prove the existence of anything with assignable non-formal properties, i e. properties beside those of existing, being a substance, being a cause, and so on. There is one argument in each of these classes. That which starts from the indeterminate existential premiss is called the Cosmological Argument. It goes back at least to Aristotle, and it is accepted by St. Thomas, Descartes, Leibniz, Locke, and many other philosophers The argument which starts from determinate existential premisses about the nature and inter-relations of actual things is called the Argument from Design or the Physico-theological Argument. These are the three arguments of what Kant calls 'Speculative Theology'.
      I do not know of any systematic way of subdividing the arguments with ethical but without religious premisses or the arguments with specifically religious premisses. So we will leave our classes (2) and (3) undivided.
      Before leaving the subject of classification I wish to call attention to the following question. Let us suppose that several of these arguments, e.g., the Cosmological Argument, the Argument from Design, and the Argument from Religious Experience, turned out to be valid, in the sense that each of them established the existence of something which, in one sense or another of the word 'God', could be called 'God'. What would be the relation between the conclusions of these various valid arguments? I think that it is commonly and uncritically assumed that they would all establish the existence of the same something; and that, from the evidential standpoint, they would be like so many strings, each attached to a different hook, and all co-operating to support a single weight. It is assumed that the differences in the conclusions would reduce to the fact that some supply us with more determinate information than do others about the common object whose existence they all conspire to establish, or that one reveals one aspect and others reveal other aspects of this common object. Now this assumption may be correct, but we have no right whatever to make it uncritically. In view of the extreme ambiguity of the word 'God' and the extreme variety of the premisses and the modes of reasoning in the several types of theistic argument, it would seem to me that there is a strong presumption against any such identification. If, e.g., two such utterly different arguments as the Cosmological Argument and the Argument From Design both establish the existence of something that can be called 'God', it seems most likely that they establish the existence of two different 'Gods', one a ground and not a designer, and the other a designer and not a ground, of the rest of the universe. Anyone who claims to identify the two should be expected to bring forward strong positive evidence for doing so. Unless such an identification can be justified, the various arguments cannot be regarded as corroborating each other. They will be like so many different strings, each acting as the sole support of a different weight.
      I will now consider some of the arguments in detail, and I will begin with:
      (1.1) The Ontological Argument. This argument presupposes the notion of degrees of 'reality' or 'perfection'. This notion is never clearly defined, but it seems to amount roughly to the following. A would be said to have 'more reality' or 'a higher degree of perfection' than B. if either of the two following conditions were fulfilled.

A has all the positive powers and qualities which B has and, in addition, it has some which B lacks. (When this condition is fulfilled we will say that A is 'extensively superior to B'.)
A is either extensively equal or extensively superior to B; some of the positive qualities or powers which are common to both are present in A with a higher degree of intensity than in B; and none of them are present in B with a higher degree of intensity than in A. (When this condition is fulfilled we will say that A is 'intensively superior to B'.)

      Now the first thing to notice is that these two criteria do not allow us, even in theory, to arrange everything in a single scale of perfection. Plainly the following cases are logically possible.

It might be that A has some powers or qualities which B lacks, and that B has some which A lacks. Cats, e.g., can climb trees, whilst dogs cannot; but dogs can track by scent, whilst cats cannot. In that case A is neither extensively superior, nor equal, nor inferior, to B. Now the criterion for intensive superiority presupposes extensive equality or superiority between the terms to be compared. Therefore, in the case supposed, there can be no comparison between A and B in respect of either extensive or intensive perfection.
A might be extensively superior to B and intensively inferior
A and B might be extensively equal. But some of their common powers or qualities might be present in A with greater intensity than in B, whilst others of them might be present in B with greater intensity than in A. Let us suppose, e.g., that the minds of any two human beings are extensively equal. How are we to compare, in respect of intensive perfection, a mathematical genius of very slight musical capacity with a musical genius of very slight mathematical capacity?

      These considerations are highly relevant to the Ontological Argument; for it uses the phrase 'most perfect being', and it presupposes that this is not meaningless verbiage like the phrase 'greatest integer'. In accounts of the Ontological Argument one finds the phrase 'most perfect being' translated in two different ways, one comparative and the other positive. The comparative interpretation makes it equivalent to the phrase 'a being such that nothing more perfect than it is logically possible'. The positive interpretation makes it equivalent to the phrase 'a being which has all positive powers and qualities to the highest possible degree'. Now, as Leibniz noted, it becomes very important at this point to consider whether all positive characteristics are mutually compatible, i.e. whether it is possible for them all to co-inhere in a common subject. Let us consider how this affects the two interpretations of the phrase 'most perfect being'.
      (i) Evidently, unless all positive characteristics are mutually compatible, the positive interpretation becomes meaningless verbiage. Suppose, e.g., that it is impossible for an extended substance to be conscious and impossible for a conscious substance to be extended, then it is impossible that there should be a substance which has all the positive properties that there are. The phrase 'a being which has all positive powers and qualities' would be meaningless verbiage like the phrase 'a surface which is red and blue all over at the same time.
      (ii) How would the comparative interpretation of the phrase 'most perfect being' fare on the same supposition, viz. that not all positive properties are compatible with each other? Let us suppose, e.g., that there were just three positive properties X, Y, and Z; that any two of them are compatible with each other; but that the presence of any two excludes the remaining one. Then there would be three possible beings, viz. one which combines X and Y. one which combines Y and Z, and one which combines Z and X, each of which would be such that nothing extensively superior to it is logically possible. For the only kind of being which would be extensively superior to any of these would be one which had all three properties, X, Y. and Z; and, by hypothesis, this combination is logically impossible. Moreover, these three beings, each of which answers to the comparative definition of a 'most perfect being' so far as concerns extensive perfection, would be incomparable with each other in this respect. For, if you take any two of them, e.g., XY and YZ, each has a positive property which the other lacks. Now the Ontological Argument talks, not merely of 'most perfect beings, but of 'the MOST PERFECT BEING'. It is now plain that, unless all positive properties be compatible with each other, this phrase is just meaningless verbiage like the phrase 'the greatest integer'.
      (iii) Let us now make the opposite supposition, viz. that all positive properties are mutually compatible. Then it is easy to see that nothing could answer to the comparative definition of 'most perfect being' unless it answered to the positive definition of that phrase. For consider any substance which had some but not all of the positive properties. Since all positive properties are now assumed to be compatible with each other, it is logically possible that there should be a substance which should have all the properties which the one under consideration has, together with the remaining ones which it lacks. This would be extensively superior to the one under consideration, and therefore the latter would not answer to the comparative definition of a 'most perfect being'.
      (iv) I have now shown

that the phrase 'the most perfect being' is meaningless unless all positive properties be compatible with each other; and
that, if they be all mutually compatible, nothing could answer to the comparative interpretation of the phrase unless it answered to the positive interpretation thereof.
The next point to notice is that, even if all positive properties be mutually compatible, the phrase 'most perfect being' may still be meaningless verbiage. For we have now to attend to that part of the positive interpretation of the phrase which we have hitherto ignored, viz. that each positive property is to be present in the highest possible degree. Now this will be meaningless verbiage unless there is some intrinsic maximum or upper limit to the possible intensity of every positive property which is capable of degrees. With some magnitudes this condition is fulfilled. It is, e.g., logically impossible that any proper fraction should exceed the ratio 1/1; and again, on a certain definition of 'angle', it is logically impossible for any angle to exceed four right angles. But it seems quite clear that there are other positive properties, such as length or temperature or pain, to which there is no intrinsic maximum or upper limit of degree.
      For these reasons it seems to me fairly certain that the Ontological Argument is wrecked before ever it leaves port. However, we will waive these objections and consider the argument itself. I will try to state it as plausibly as I can. It might be put as follows:
'Anything that lacked existence would lack a positive property which it might conceivably have had. Nothing which lacked a positive property which it might conceivably have had would be a most perfect being; for it is logically possible that there should be something superior to it, viz. a being which resembled it in all other respects but had the additional property of existence. Therefore no most perfect being would lack existence. Therefore all most perfect beings exist.'

      Let us now consider this argument. It has two steps, viz. a syllogism followed by an immediate inference. There is nothing wrong with the syllogism in respect of its verbal form. It is verbally of the form 'Anything that lacked P would lack M. Nothing that lacked M would be S. Therefore no S would lack P.' This breaks none of the rules; it is in fact a slightly disguised form of the valid fourth figure syllogism Camenes. The second step looks like a generally accepted form of immediate inference, viz. Obversion. But at this point there is a serious risk of a fallacy. The verbal form 'All S is P' is ambiguous. It may mean simply 'If anything were S it would be P', or, what is equivalent, 'Anything that was S would be P'. Interpreted in this way, it leaves the question whether anything in fact is S quite open. We will call this the 'conditional' interpretation. On the other hand, it is much more often taken to mean 'There are some S's and none of them lack P'. This may be called the 'instantial' interpretation. Now it is a general principle of logic that it is always illegitimate to draw an instantial conclusion from premisses which are wholly conditional. Let us now apply these principles to the second step of the argument.
      The two premisses of the syllogism are purely conditional. Therefore the conclusion must be interpreted purely conditionally if the syllogism is to be valid. So the conclusion of the syllogism must be taken to be 'If anything were a most perfect being it would not lack existence'. Now all that can be legitimately inferred from this by obversion is the conditional proposition 'If anything were a most perfect being it would exist'. If you interpret the sentence 'All most perfect beings exist' in this way, the conclusion follows from the premisses but is completely trivial and useless. If, on the other hand, you interpret it instantially, i.e. take it to mean There are most perfect beings and none of them lack existence', there are two fatal criticisms to be made.

You are attempting to draw an instantial conclusion from purely conditional premisses and therefore are committing a logical fallacy.
The sentence as a whole is pleonastic. It is idle to add 'none of them lack existence' to 'there are so-and-so's', whether the so-and-so's be most perfect beings or potatoes or dragons.

      Let us now consider the syllogism itself. As I have said, it is correct in verbal form. Nevertheless, as I shall now proceed to show it is radically vicious. Its defect is, not that its premisses are false, but that they are meaningless. They are sentences which seem, from their verbal form, to express propositions; but in fact they express nothing whatever. The argument presupposes that existence is a quality or power, like extension or consciousness or life; it assumes that there is sense in talking of a comparison between a non-existent term and an existent term; and it produces the impression that this is like comparing two existing terms, e.g., a corpse and a living organism, one of which lacks life and the other of which has it.
      Now all this is nonsensical verbiage. It is intelligible to make a categorical comparison between two actual existents, e.g., Hitler and Stalin, in respect of their qualities and powers. It is intelligible to take a description of a merely possible existent, e.g., a creature with a horse's body and a man's head, and to make a conditional comparison with an actual existent. It is, e.g., intelligible to say 'If a centaur existed (or, if there were a centaur), it would be swifter than any actual man and more rational than any actual horse'. Lastly, it is intelligible to take descriptions of two merely possible existents, and to make a doubly conditional comparison. It is, e.g., intelligible to say 'If centaurs existed and unicorns existed (or, if there were centaurs and unicorns), the former would be superior (or inferior) to the latter in such and such respects'. Now the Ontological Argument professes to make a categorical comparison between a non-existent and an existent in respect of the presence or absence of existence. The objection is twofold. (i) No comparison can be made between a non-existent term and anything else except on the hypothesis that it exists. And (ii) on this hypothesis it is meaningless to compare it with anything in respect of the presence or absence of existence.
      It is evident, then, that the Ontological Argument must be rejected. Probably most people feel that there is something wrong with it; but the important and interesting and not too easy task is to put one's finger on the precise points at which it goes wrong. When a fallacious argument has seemed cogent to many people of the highest intelligence, such as St. Anselm, Descartes, and Leibniz, it is desirable to supplement the refutation of it by an attempt to explain the causes of its plausibility. I believe that there are two causes, in the present case; and I will now proceed to exhibit them.
      (i) The first and most important cause of the illusion is the fact that existential propositions and characterizing propositions are expressed by sentences which have the same grammatical form. Thus, e.g. existential propositions are expressed by such sentences as 'S exists' or 'S is real', while characterizing propositions are expressed by such grammatically similar sentences as 'S eats' or 'S is red'. This linguistic fact tempts people to assume uncritically that existential propositions are logically of the same form as characterizing propositions. This uncritical assumption makes the Ontological Argument seem plausible. But it is certainly false, as can easily be shown. The demonstration of this fact may be put as follows.
      Let us begin with the two negative propositions Cats do not bark and Dragons do not exist. It is obvious that the first is about cats. But, if the second be true, it is certain that it cannot be about dragons; for there will be no such things as dragons for it to be about. The first might be expressed, on the conditional interpretation, by the sentence 'If there were any cats, none of them would bark'. On the instantial interpretation it might be expressed by the sentence 'There are cats, and none of them bark'. Suppose you try to express the negative existential proposition in the same way. On the first alternative it would be expressed by the sentence 'If there were any dragons, none of them would exist'. On the second alternative it would be expressed by the sentence 'There are dragons, and none of them exist'. Both these sentences are self-contradictory and meaningless. So, if you try to analyse negative existential propositions in the same way as negative characterizing propositions, you will find that they are all self-contradictory. But it is plain that Dragons do not exist is not self-contradictory. It is not only logically possible but is almost certainly true.
      Now consider the two affirmative propositions Cats scratch and Cats exist. On the conditional interpretation the former would be expressed by the sentence 'If there were any cats, none of them would fail to scratch'. On the instantial interpretation it would be expressed by the sentence 'There are cats, and none of them fail to scratch' Suppose you try to express the affirmative existential proposition in the same way. On the first alternative it would be expressed by the sentence 'If there were any cats, none of them would fail to exist'. On the second alternative it would be expressed by the sentence 'There are cats, and none of them fail to exist'. Now both these sentences are mere platitudes. So, if you try to analyse affirmative existential propositions in the same way as affirmative characterizing propositions, you will find that they are all platitudes. But it is plain that Cats exist is not a mere platitude. It is a substantial proposition which might very well be doubted by a person who had never seen a cat. So it is certain that existential propositions need a different kind of analysis.
      The right analysis, as is now well known, is somewhat as follows. These propositions are not about cats or dragons. i.e. about things which have the cat-characteristics or the dragon-characteristics. They are about these characteristics themselves. What they assert is that these characteristics do apply to something or that they do not apply to anything, as the case may be. 'Cats exist' is equivalent to 'The defining characteristics of the word "cat" apply to something. Again 'Dragons do not exist' is equivalent to 'The defining characteristics of the word "dragon" do not apply to anything'. Suppose, e.g., that a 'dragon' is defined as a reptile which flies and breathes fire. Then the statement that dragons do not exist is equivalent to the statement that nothing combines the three properties of being a reptile, of flying, and of breathing fire. Such statements are neither tautologies nor contradictions.
      It only remains to apply this analysis to statements about the existence or non-existence of a most perfect being. To say that a most perfect being exists is equivalent to saying that something has all positive characteristics to the highest possible degree. For reasons which I have given, it seems likely that this is not only false but also self-contradictory and nonsensical. To say that a most perfect being does not exist is equivalent to saying that nothing has all positive characteristics to the highest possible degree. For the same reasons it seems likely that this is not only true but a truism.
      (ii) I strongly suspect that another linguistic fact about the use of the word 'exist' has helped to make the Ontological Argument seem evident truth instead of meaningless nonsense. It is not uncommon to say, of a person or animal who has died, that he has 'ceased to exist'. Now in this case there is something visible and tangible left, viz. the corpse, which can be compared with the person or animal as he was before he died. Moreover, it is obvious that a living organism is more perfect than a corpse. This leads people to think of existence as a positive characteristic which can be added to or subtracted from a thing, and whose presence makes a thing more perfect than it would have been without it. But, in the sense of 'existence' required for the Ontological Argument, a corpse exists as much as a living organism. So this linguistic fact does nothing to justify the speculations which it encourages.
      (1.2) The Cosmological Argument. This argument goes back, historically, to a physical argument of Aristotle's about motion. Aristotle's attempt to prove that there must be an unmoved cause of motion is of considerable interest, but, for the present purpose, it seems more profitable to consider the argument in a less specialized form. It may be put as follows.
      It starts with the premiss that there are particular things, persons, events, etc. Each of us, e.g., can take himself as an indubitable instance of a particular person and can take any one of his present experiences as an indubitable instance of a particular event. Now any thing or person begins to exist at a particular time and place, lasts for a longer or shorter period, and then ceases to exist. Similarly, any event in the history of a thing or person begins at a certain time. Now the coming into existence of a thing or person of such and such a kind at a certain time and place is felt to need explanation. Similarly, the occurrences at a certain date in the his tory of a thing or person, of a change of such and such a kind is felt to need explanation. The first move is to try to explain it by reference to previously existing things or persons (such as parents) and by reference to earlier events. We will call this 'explanation in terms of ordinary causation'. Now this kind of explanation is, in one respect, never completely satisfactory. This is for two reasons. The first is that such explanations always involve a reference to general laws as well as to particular things, persons, and events. Now the general laws are themselves just brute facts, with no trace of self-evidence or intrinsic necessity about them. The second and more obvious reason is the following. The earlier things, persons, and events, to which you are referred by explanation in terms of ordinary causation, stand in precisely the same need of explanation as the thing or person or event which you set out to explain. It is obvious from the nature of the case that no extension of this kind of explanation to remoter and remoter depths of past time has the slightest tendency to remove this defect.
      Before continuing the argument I would point out that nothing that has been said casts any doubt on the theoretical interest or the practical importance of explanation in terms of ordinary causation. When we 'explain' in this way we are learning more and more about the inter-connexions of things and events in time and space. Moreover, by learning these facts, we are enabled to acquire more extensive control over nature, to make new kinds of substances, and to modify the course of future events.
      We can now go on with the argument. It is alleged that we can conclude, from the negative facts already stated, that there must be a substance which is neither a part of nature nor nature as a collective whole. And we can conclude that there is another kind of dependence, which is not the ordinary dependence of a later state of affairs on an earlier one in accordance with de facto rules of sequence. The existence of this non-natural substance must be intrinsically necessary. And the existence of all natural events and substances must be dependent upon the existence of this non-natural substance by this non-natural kind of dependence.
      Let us now consider whether this argument is valid. It may be divided into two parts, negative and positive. At the transition from the negative to the positive part there is a suppressed premise. My criticism will be as follows.

I accept the negative part of the argument.
The suppressed premiss, which forms the transition from the negative to the positive part, seems to me to be false. Therefore I see no reason to accept the conclusion.
I suspect that the conclusion is not only unproven but is either false or meaningless. I will now develop these statements.

      (i) What kind of explanations do completely satisfy the human intellect? The human intellect is completely satisfied with a proposition when either (a) the proposition is seen to be intrinsically necessary by direct inspection of its terms, or (b) it is seen to follow by steps, each of which is seen to be intrinsically necessary, from premisses which are all seen to be intrinsically necessary. This kind of complete intellectual satisfaction is reached in pure mathematics and hardly anywhere else. Now it is quite certain that no explanation in terms of ordinary causation is capable of giving this kind of satisfaction to the intellect. For no causal law has any trace of self-evidence, and no premiss to the effect that such and such things existed or that such and such events happened in the past has any trace of self-evidence. The causal explanations of science are useful for predicting and controlling the future, for reconstructing the past, and for learning about what is remote in distance or minute in size. But they provide no explanation of anything in the sense in which the proof of a proposition in pure mathematics does provide a completely satisfactory explanation of the mathematical fact asserted by that proposition.
      Now it is logically possible that complete intellectual satisfaction should be obtained about natural events and substances if and only if the following conditions were fulfilled.

If there were one or more existential propositions which are intrinsically necessary, like mathematical axioms. And
if all other true existential propositions followed with strict logical necessity from these, combined, perhaps, with certain intrinsically necessary universal premisses.
Suppose that these conditions were fulfilled; and suppose, further, that there were a man who actually knew these intrinsically necessary premisses and actually saw in detail that they entail, e.g., the existence at a certain time and place of a person answering to the description of the historical Julius Caesar. Then he would actually enjoy complete intellectual satisfaction about the existence of Julius Caesar.
      I therefore accept so much as follows of the Cosmological Argument. I admit that no explanation in terms of ordinary causation is capable of giving that kind of intellectual satisfaction about natural things and persons and events which is obtainable about purely mathematical facts. And I admit that, if the universe is such that this kind of intellectual satisfaction is theoretically obtainable about nature, then its structure must be very much as philosophic Theism says that it is.
      (ii) The Cosmological Argument claims to prove a categorical proposition, viz. that the universe has this structure. In order to do so it must add a categorical premiss to the hypothetical proposition which I have just admitted. It is plain that this categorical premiss is the proposition that the universe is such that this kind of intellectual satisfaction about natural things, persons, and events is, at least in theory, obtainable. This, then, is the suppressed premiss of the argument. Is there any reason to accept it?
      We must not unfairly exaggerate what it claims. It is not asserted that any human being ever will in fact enjoy this kind of intellectual satisfaction about nature as a whole or about a single natural thing or person or event. All that is asserted is that the universe is such that a mind, which worked on the same general principles as ours but had indefinitely greater knowledge of detail and power of seeing logical connexions and keeping them before it without confusion, would find every fact about nature perfectly intelligible, in the sense in which everything in pure mathematics is perfectly intelligible to the mathematician. Now I do not see the least reason to believe this. Plainly it is not the kind of premiss for which there is or could be any empirical evidence. Nor is it self-evident or deducible from any premisses which are self-evident. Wherever we have this kind of completely satisfactory insight we are dealing with the formal relations of abstract entities, such as numbers or propositions, and not with the existence or the non-formal properties of particulars. There is no reason whatever to think that this kind of rational insight is possible in the latter case.
      (iii) I think that we can go much farther than this in the negative direction. We have seen that an indispensable condition, without which it is logically impossible for nature to be capable of satisfying the intellect in the sense defined, is that there should be some intrinsically necessary existential propositions. Now, in criticizing the Ontological Argument, we saw that 'So-and-so exists' is equivalent to 'There is something which has such and such a set of characteristics', where this set of characteristics constitutes the definition or description of a certain possible object. Therefore an intrinsically necessary existential proposition would be of the form 'There must be something which has the characteristics x, y, z, etc.', where this set of characteristics constitutes the definition or description of a certain possible object. Or, to put it the other way round, 'The set of characteristics, x, y, z, etc., must together belong to something' .
      Now it seems to me evident that there can be no intrinsically necessary propositions of this kind. Necessary propositions are always about the connexion (or disconnexion) of one attribute with another attribute or one proposition with another proposition, and they are always conditional. They are always of the form 'If anything had the attribute x, it would necessarily have the attribute y', or 'If p were true, then q would be true'. If I am right on this point, it follows that the conclusion of the Cosmological Argument is not only unproven but is false. And it follows that the suppressed premiss of the argument is false. That is, we can be quite certain that the universe cannot be of such a structure that the kind of intellectual satisfaction which is possible in pure mathematics might conceivably be attained about the things and persons and events of nature.
      Even if this objection be waived, an equally formidable one remains. Let us suppose, for the sake of argument, that the suppressed premiss is true. Then I think it is easy to show that, even if there were an existent or existents whose existence is intrinsically necessary, this would not in the least help to make nature theoretically intelligible in the sense required. The difficulty is as follows. Anything whose existence was a necessary consequence of its nature would be a timeless existent. If a certain set of attributes is such that it must belong to something, it is nonsensical to talk of its beginning to belong to something at any date, however far back in the past. It would be like talking of a date at which equilateral triangles began to be equiangular. Now nature is composed of things and persons and processes which begin at certain dates, last for so long, and then cease. But how could a temporal fact, such as the fact that there began to be a person having the characteristics of Julius Caesar at a certain date, follow logically from facts all of which are non-temporal? Surely it is perfectly obvious that the necessary consequences of facts which are necessary are themselves necessary, and that the necessary consequences of facts which have no reference to any particular time can themselves have no reference to any particular time.
      I may therefore sum up my criticisms on the Cosmological Argument as follows. The argument presupposes that nature must be, in principle, capable of satisfying the intellect in the way in which it can be satisfied in pure mathematics. It rightly denies that explanations in terms of ordinary causation, however far back they may be carried, have any tendency to produce this kind of intellectual satisfaction. It argues that such intellectual satisfaction about nature would be in principle obtainable if and only if the two following conditions were fulfilled:

that there is at least one particular such that the existence of a particular of that nature is an intrinsically necessary existential fact;
that all facts about the existence of such natural substances as do exist and about the occurrence of such natural events as do occur are necessary consequences of these intrinsically necessary existential facts.
The conclusion of the argument is that these two conditions must be fulfilled. Now the objections are these.

It is not in the least evident that nature must bc in principle capable of satisfying the intellect in this peculiar way.
The first of the two conditions which are necessary for the fulfilment of this demand appears, on reflexion, to be almost devoid of meaning and almost certainly incapable of realization.
Even if the first condition were fulfilled, it is self-evidently impossible that the second should be. For this would require that facts about the existence of things and the occurrence of events at certain dates should be necessary consequences of facts which are all without any temporal reference whatever.

      Suppose now that all these objections could be overcome. What kind of conclusion would the Cosmological Argument establish, and how is this argument related to the Ontological Argument? In answer to the first question there are two remarks to be made. (a) The Cosmological Argument, by itself, would not justify the conclusion that there is only one substance whose existence is a necessary consequence of its nature and from which alone the existence of everything else follows. It would justify only the less determinate conclusion that there is at least one such substance, and that from the existence of it or of them the existence of everything else follows. (b) If the conclusion of the Cosmological Argument be accepted, it follows that there are no really contingent facts. The fact that a person having the nature which I have was born at a certain time and place, and the fact that he sneezed at 11.15 yesterday, may seem contingent relatively to our ignorance. But, if we accept the Cosmological Argument, we know that these facts must be necessary consequences of facts which are all intrinsically necessary. Therefore they cannot really be contingent. All that is possible will be actual, and all that is actual will be necessary. This, as we all know, is the consequence which Spinoza drew in Book I of his Ethics, and it seems to me that Spinoza is one of the few people who have both accepted the Cosmological Argument and seen clearly the logical consequences of it.
      The relation of the Cosmological to the Ontological Argument may be stated as follows. The Ontological Argument specifies a certain property, viz. that of having all positive powers and qualities to the highest possible degree; and it professes to show that there must be something which has this property. The Cosmological Argument does not claim to do anything so definite as this. It claims only to prove that there must be at least one set of characteristics such that there must be something which has them all. It does not profess to mention any specific set of characteristics and to show that there must be something that has them. If the Ontological Argument were valid, the conclusion of the Cosmological Argument would certainly be true. But the Cosmological Argument might be valid, and its conclusion might be true, even if the Ontological Argument were invalid and its conclusion false.
      It only remains to consider what causes have made the Cosmological Argument seem valid to so many men of the highest intellectual power. It was accepted, e.g., by Aristotle, St. Thomas, Descartes, Spinoza, Leibniz, and Locke; and this certainly seems a sufficient guarantee of philosophical respectability. I think that there are two causes for this widespread delusion. One is the failure, which we have already noted in connexion with the Ontological Argument, to recognize the peculiarity of existential propositions and the fact that they are utterly unlike characterizing propositions in logical structure. So long as this difference remains unnoticed it does not seem absurd to talk of necessary existential propositions or facts. But, when once it is seen that all admittedly necessary propositions are of the form 'if this were the case, then that would be the case', and that no existential proposition is of that form, the temptation to think that there might be necessary existential propositions or facts is removed. A second cause is the very peculiar position which Euclidean geometry enjoyed for so many centuries. Here we have a science which seems to consist of propositions which necessarily follow from intrinsically necessary premisses, and yet to give us synthetic and categorical information about a certain important aspect of nature. This suggested the ideal of a completely rational knowledge of every aspect and every fact of nature; and it made this ideal appear to be intelligible even if the de facto limitations of the human intellect should forbid its being ever realized in detail. We know now that the necessity of Euclidean geometry, like all other necessity, is only conditional. The theorems follow necessarily from the axioms; but the axioms themselves are not intrinsically necessary, and therefore their necessary consequences are not themselves necessary propositions. So we are exempt from this temptation to which so many of our betters succumbed.
      So far I have been dealing with what may fairly be called the more 'metaphysical' arguments for the existence of God. I pass now to those which may be called more 'empirical'. In accordance with the classification already given, the arguments which remain to be considered are the Argument from Design, arguments which use specifically ethical but not specifically religious premisses, and those which use specifically religious premisses.
      The Argument from Design has been criticized very fairly and thoroughly by two of the greatest European philosophers, Hume and Kant. I have nothing to add to their criticisms, and I have seen nothing in the writings of those who have tried to rehabilitate the argument which effectively rebuts their adverse verdict. I shall therefore set this argument aside. As regards arguments from ethical premisses, I have said what I have to say on the logical and epistemological issues in Chapter XI of my book The Mind and its Place in Nature. That chapter is, indeed, concerned primarily with ethical arguments for human survival, and not for the existence of God. But the principles are the same in either case, and so I do not propose to treat the subject again here. I shall therefore confine myself in this article to specifically religious experience and the argument for the existence of God which has been based on it.
      This argument differs in the following important respect from the other two empirical types of argument. The Argument from Design and the arguments from ethical premisses start from facts which are common to every one. But some people seem to be almost wholly devoid of any specifically religious experience; and among those who have it the differences of kind and degree are enormous. Founders of religions and saints, e.g., often claim to have been in direct contact with God, to have seen and spoken with Him, and so on. An ordinary religious man would certainly not make any such claim, though he might say that he had had experiences which assured him of the existence and presence of God. So the first thing that we have to notice is that capacity for religious experience is in certain respects like an ear for music. There are a few people who are unable to recognize and distinguish the simplest tune. But they are in a minority, like the people who have absolutely no kind of religious experience. Most people have some slight appreciation of music. But the differences of degree in this respect are enormous, and those who have not much gift for music have to take the statements of accomplished musicians very largely on trust. Let us, then, compare tone-deaf persons to those who have no recognizable religious experience at all; the ordinary followers of a religion to men who have some taste for music but can neither appreciate the more difficult kinds nor compose; highly religious men and saints to persons with an exceptionally fine ear for music who may yet be unable to compose it; and the founders of religions to great musical composers, such as Bach and Beethoven.
      This analogy is, of course, incomplete in certain important respects. Religious experience raises three problems, which are different though closely interconnected.

What is the psychological analysis of religious experience? Does it contain factors which are present also in certain experiences which are not religious? Does it contain any factor which never occurs in any other kind of experience? If it contains no such factor, but is a blend of elements each of which can occur separately or in non-religious experiences, its psychological peculiarity must consist in the characteristic way in which these elements are blended in it. Can this peculiar structural feature of religious experience be indicated and described?
What are the genetic and causal conditions of the existence of religious experience? Can we trace the origin and development of the disposition to have religious experiences

in the human race, and
in each individual?
Granted that the disposition is present in nearly all individuals at the present time, can we discover and state the variable conditions which call it into activity on certain occasions and leave it in abeyance on others?
Part of the content of religious experience is alleged knowledge or well-founded belief about the nature of reality, e.g., that we are dependent on a being who loves us and whom we ought to worship, that values are somehow conserved in spite of the chances and changes of the material world at the mercy of which they seem prima facie to be, and so on. Therefore there is a third problem. Granted that religious experience exists, that it has such-and-such a history and conditions, that it seems vitally important to those who have it, and that it produces all kinds of effects which would not otherwise happen, is it veridical? Are the claims to knowledge or well-founded belief about the nature of reality, which are an integral part of the experience, true or probable? Now, in the case of musical experience, there are analogies to the psychological problem and to the genetic or causal problem, but there is no analogy to the epistemological problem of validity. For, so far as I am aware, no part of the content of musical experience is alleged knowledge about the nature of reality; and therefore no question of its being veridical or delusive can arise.

      Since both musical experience and religious experience certainly exist, any theory of the universe which was incompatible with their existence would be false, and any theory which failed to show the connexion between their existence and the other facts about reality would be inadequate. So far the two kinds of experience are in exactly the same position. But a theory which answers to the condition that it allows of the existence of religious experience and indicates the connexion between its existence and other facts about reality may leave the question as to its validity quite unanswered. Or, alternatively, it may throw grave doubt on its cognitive claims, or else it may tend to support them. Suppose, e.g., that it could be shown that religious experience contains no elements which are not factors in other kinds of experience. Suppose further it could be shown that this particular combination of factors tends to originate and to be activated only under certain conditions which are known to be very commonly productive of false beliefs held with strong conviction. Then a satisfactory answer to the questions of psychological analysis and causal antecedents would have tended to answer the epistemological question of validity in the negative On the other hand, it might be that the only theory which would satisfactorily account for the origin of the religious disposition and for the occurrence of actual religious experiences under certain conditions was a theory which allowed some of the cognitive claims made by religious experience to be true or probable. Thus the three problems, though entirely distinct from each other, may be very closely connected; and it is the existence of the third problem in connexion with religious experience which puts it, for the present purpose, in a different category from musical experience.
      In spite of this essential difference the analogy is not to be despised, for it brings out at least one important point. If a man who had no ear for music were to give himself airs on that account, and were to talk de haut en bas about those who can appreciate music and think it highly important, we should regard him, not as an advanced thinker, but as a self-satisfied Philistine. And, even if he did not do this but only propounded theories about the nature and causation of musical experience, we might think it reasonable to feel very doubtful whether his theories would be adequate or correct. In the same way, when persons without religious experience regard themselves as being on that ground superior to those who have it, their attitude must be treated as merely silly and offensive. Similarly, any theories about religious experience constructed by persons who have little or none of their own should be regarded with grave suspicion. (For that reason it would be unwise to attach very much weight to anything that the present writer may say on this subject.)
      On the other hand, we must remember that the possession of a great capacity for religious experience, like the possession of a great capacity for musical appreciation and composition, is no guarantee of high general intelligence. A man may be a saint or a magnificent musician and yet have very little common sense, very little power of accurate introspection or of seeing causal connexions, and scarcely any capacity for logical criticism. He may also be almost as ignorant about other aspects of reality as the non-musical or non-religious man is about musical or religious experience. If such a man starts to theorize about music or religion, his theories may be quite as absurd, though in a different way, as those made by persons who are devoid of musical or religious experience. Fortunately it happens that some religious mystics of a high order have been extremely good at introspecting and describing their own experiences. And some highly religious persons have had very great critical and philosophical abilities. St. Teresa is an example of the first, and St. Thomas Aquinas of the second.
      Now I think it must be admitted that, if we compare and contrast the statements made by religious mystics of various times, races, and religions, we find a common nucleus combined with very great differences of detail. Of course the interpretations which they have put on their experiences are much more varied than the experiences themselves. It is obvious that the interpretations will depend in a large measure on the traditional religious beliefs in which various mystics have been brought up. I think that such traditions probably act in two different ways.
      (i) The tradition no doubt affects the theoretical interpretation of experiences which would have taken place even if the mystic had been brought up in a different tradition. A feeling of unity with the rest of the universe will be interpreted very differently by a Christian who has been brought up to believe in a personal God and by a Hindu mystic who has been trained in a quite different metaphysical tradition.
      (ii) The traditional beliefs, on the other hand, probably determine many of the details of the experience itself. A Roman Catholic mystic may have visions of the Virgin and the saints, whilst a Protestant mystic pretty certainly will not.
      Thus the relations between the experiences and the traditional beliefs are highly complex. Presumably the outlines of the belief are determined by the experience. Then the details of the belief are fixed for a certain place and period by the special peculiarities of the experiences had by the founder of a certain religion. These beliefs then become traditional in that religion. Thenceforth they in part determine the details of the experiences had by subsequent mystics of that religion, and still more do they determine the interpretations which these mystics will put upon their experiences. Therefore, when a set of religious beliefs has once been established, it no doubt tends to produce experiences which can plausibly be taken as evidence for it. If it is a tradition in a certain religion that one can communicate with saints, mystics of that religion will seem to see and to talk with saints in their mystical visions; and this fact will be taken as further evidence for the belief that one can communicate with saints.
      Much the same double process of causation takes place in sense-perception. On the one hand, the beliefs and expectations which we have at any moment largely determine what interpretation we shall put on a certain sensation which we should in any case have had then. On the other hand, our beliefs and expectations do to some extent determine and modify some of the sensible characteristics of the sensa themselves. When I am thinking only of diagrams a certain visual stimulus may produce a sensation of a sensibly flat sensum; but a precisely similar stimulus may produce a sensation of a sensibly solid sensum when I am thinking of solid objects.
      Such explanations, however, plainly do not account for the first origin of religious beliefs, or for the features which are common to the religious experiences of persons of widely different times, races, and traditions.
      Now, when we find that there are certain experiences which, though never very frequent in a high degree of intensity, have happened in a high degree among a few men at all times and places; and when we find that, in spite of differences in detail which we can explain, they involve certain fundamental conditions which are common and peculiar to them; two alternatives are open to us. (i) We may suppose that these men are in contact with an aspect of reality which is not revealed to ordinary persons in their everyday experience. And we may suppose that the characteristics which they agree in ascribing to reality on the basis of these experiences probably do belong to it. Or (ii) we may suppose that they are all subject to a delusion from which other men are free. In order to illustrate these alternatives it will be useful to consider three partly analogous cases, two of which are real and the third imaginary.


Most of the detailed facts which biologists tell us about the minute structure and changes in cells can be perceived only by persons who have had a long training in the use of the microscope. In this case we believe that the agreement among trained microscopists really does correspond to facts which untrained persons cannot perceive.
Persons of all races who habitually drink alcohol to excess eventually have perceptual experiences in which they seem to themselves to see snakes or rats crawling about their rooms or beds. In this case we believe that this agreement among drunkards is merely a uniform hallucination.
Let us now imagine a race of beings who can walk about and touch things but cannot see. Suppose that eventually a few of them developed the power of sight. All that they might tell their still blind friends about colour would be wholly unintelligible to and unverifiable by the latter. But they would also be able to tell their blind friends a great deal about what the latter would feel if they were to walk in certain directions. These statements would be verified. This would not, of course, prove to the blind ones that the unintelligible statements about colour correspond to certain aspects of the world which they cannot perceive. But it would show that the seeing persons had a source of additional information about matters which the blind ones could understand and test for themselves. It would not be unreasonable then for the blind ones to believe that probably the seeing ones are also able to perceive other aspects of reality which they are describing correctly when they make their unintelligible statements containing colour-names. The question then is whether it is reasonable to regard the agreement between the experiences of religious mystics as more like the agreement among trained microscopists about the minute structure of cells, or as more like the agreement among habitual drunkards about the infestation of their rooms by pink rats or snakes, or as more like the agreement about colours which the seeing men would express in their statements to the blind men.

      Why do we commonly believe that habitual excess of alcohol is a cause of a uniform delusion and not a source of additional information? The main reason is as follows. The things which drunkards claim to perceive are not fundamentally different in kind from the things that other people perceive. We have all seen rats and snakes, though the rats have generally been grey or brown and not pink. Moreover the drunkard claims that the rats and snakes which he sees are literally present in his room and on his bed, in the same sense in which his bed is in his room and his quilt is on his bed. Now we may fairly argue as follows. Since these are the sort of things which we could see if they were there, the fact that we cannot see them makes it highly probable that they are not there. Again, we know what kinds of perceptible effect would generally follow from the presence in a room of such things as rats or snakes. We should expect fox-terriers or mongooses to show traces of excitement, cheese to be nibbled, corn to disappear from bins, and so on. We find that no such effects are observed in the bedrooms of persons suffering from delirium tremens. It therefore seems reasonable to conclude that the agreement among drunkards is a sign, not of a revelation, but of a delusion.
      Now the assertions in which religious mystics agree are not such that they conflict with what we can perceive with our senses. They are about the structure and organization of the world as a whole and about the relations of men to the rest of it. And they have so little in common with the facts of daily life that there is not much chance of direct collision. I think that there is only one important point on which there is conflict. Nearly all mystics seem to be agreed that time and change and unchanging duration are unreal or extremely superficial, whilst these seem to plain men to be the most fundamental features of the world. But we must admit, on the one hand, that these temporal characteristics present very great philosophical difficulties and puzzles when we reflect upon them. On the other hand, we may well suppose that the mystic finds it impossible to state clearly in ordinary language what it is that he experiences about the facts which underlie the appearance of time and change and duration. Therefore it is not difficult to allow that what we experience as the temporal aspect of reality corresponds in some sense to certain facts, and yet that these facts appear to us in so distorted a form in our ordinary experience that a person who sees them more accurately and directly might refuse to apply temporal names to them.
      Let us next consider why we feel fairly certain that the agreement among trained microscopists about the minute structure of cells expresses an objective fact, although we cannot get similar experiences. One reason is that we have learned enough, from simpler cases of visual perception, about the laws of optics to know that the arrangement of lenses in a microscope is such that it will reveal minute structure, which is otherwise invisible, and will not simply create optical delusions. Another reason is that we know of other cases in which trained persons can detect things which untrained people will overlook, and that in many cases the existence of these things can be verified by indirect methods. Probably most of us have experienced such results of training in our own lives.
      Now religious experience is not in nearly such a strong position as this. We do not know much about the laws which govern its occurrence and determine its variations. No doubt there are certain standard methods of training and meditation which tend to produce mystical experiences. These have been elaborated to some extent by certain Western mystics and to a very much greater extent by Eastern Yogis. But I do not think that we can see here as we can in the case of microscopes and the training which is required to make the best use of them, any conclusive reason why these methods should produce veridical rather than delusive experiences. Uniform methods of training and meditation would be likely to produce more or less similar experiences, whether these experiences were largely veridical or wholly delusive.
      Is there any analogy between the facts about religious experience and the fable about the blind men some of whom gained the power of sight? It might be said that many ideals of conduct and ways of life, which we can all recognize now to be good and useful, have been introduced into human history by the founders of religions. These persons have made actual ethical discoveries which others can afterwards recognize to be true. It might be said that this is at least roughly analogous to the case of the seeing men telling the still blind men of facts which the latter could and did verify for themselves. And it might be said that this makes it reasonable for us to attach some weight to what founders of religions tell us about things which we cannot understand or verify for ourselves; just as it would have been reasonable for the blind men to attach some weight to the unintelligible statements which the seeing men made to them about colours.
      I think that this argument deserves a certain amount of respect, though I should find it hard to estimate how much weight to attach to it. I should be inclined to sum up as follows. When there is a nucleus of agreement between the experiences of men in different places, times, and traditions, and when they all tend to put much the same kind of interpretation on the cognitive content of these experiences, it is reasonable to ascribe this agreement to their all being in contact with a certain objective aspect of reality unless there be some positive reason to think otherwise. The practical postulate which we go upon everywhere else is to treat cognitive claims as veridical unless there be some positive reason to think them delusive. This, after all, is our only guarantee for believing that ordinary sense-perception is veridical. We cannot prove that what people agree in perceiving really exists independently of them; but we do always assume that ordinary waking sense-perception is veridical unless we can produce some positive ground for thinking that it is delusive in any given case. I think it would be inconsistent to treat the experiences of religious mystics on different principles. So far as they agree they should be provisionally accepted as veridical unless there be some positive ground for thinking that they are not. So the next question is whether there is any positive ground for holding that they are delusive.
      There are two circumstances which have been commonly held to cast doubt on the cognitive claims of religious and mystical experience

It is alleged that founders of religions and saints have nearly always had certain neuropathic symptoms or certain bodily weaknesses, and that these would be likely to produce delusions. Even if we accept the premisses, I do not think that this is a very strong argument.

It is equally true that many founders of religions and saints have exhibited great endurance and great power of organization and business capacity which would have made them extremely successful and competent in secular affairs. There are very few offices in the cabinet or in the highest branches of the civil service which St. Thomas Aquinas could not have held with conspicuous success. I do not, of course, regard this as a positive reason for accepting the metaphysical doctrines which saints and founders of religions have based on their experiences; but it is relevant as a rebuttal of the argument which we are considering.
Probably very few people of extreme genius in science or art are perfectly normal mentally or physically, and some of them are very crazy and eccentric indeed. Therefore it would be rather surprising if persons of religious genius were completely normal, whether their experiences be veridical or delusive.
Suppose, for the sake of argument, that there is an aspect of the world which remains altogether outside the ken of ordinary persons in their daily life. Then it seems very likely that some degree of mental and physical abnormality would be a necessary condition for getting sufficiently loosened from the objects of ordinary sense-perception to come into cognitive contact with this aspect of reality. Therefore the fact that those persons who claim to have this peculiar kind of cognition generally exhibit certain mental and physical abnormalities is rather what might be anticipated if their claims were true. One might need to be slightly 'cracked' in order to have some peep-holes into the super-sensible world.
If mystical experience were veridical, it seems quite likely that it would produce abnormalities of behaviour in those who had it strongly. Let us suppose, for the sake of argument, that those who have religious experience are in frequent contact with an aspect of reality of which most men get only rare and faint glimpses. Then such persons are, as it were, living in two worlds, while the ordinary man is living in only one of them. Or, again, they might be compared to a man who has to conduct his life with one ordinary eye and another of a telescopic kind. Their behaviour may be appropriate to the aspect of reality which they alone perceive and think all-important; but, for that very reason, it may be inappropriate to those other aspects of reality which are all that most men perceive or judge to be important and on which all our social institutions and conventions are built.

      (ii) A second reason which is commonly alleged for doubt about the claims of religious experience is the following. It is said that such experience always originates from and remains mixed with certain other factors, e.g., sexual emotion, which are such that experiences and beliefs that arise from them are very likely to be delusive. I think that there are a good many confusions on this point, and it will be worth while to begin by indicating some of them.
      When people say that B 'originated from' A, they are liable to confuse at least three different kinds of connexion between A and B.

It might be that A is a necessary but insufficient condition of the existence of B.
It might be that A is a necessary and sufficient condition of the existence of B. Or
it might be that B simply is A in a more complex and disguised form.
Now, when there is in fact evidence only for the first kind of connexion, people are very liable to jump to the conclusion that there is the third kind of connexion. It may well be the case, e.g., that no one who was incapable of strong sexual desires and emotions could have anything worth calling religious experience. But it is plain that the possession of a strong capacity for sexual experience is not a sufficient condition of having religious experience; for we know that the former quite often exists in persons who show hardly any trace of the latter. But, even if it could be shown that a strong capacity for sexual desire and emotion is both necessary and sufficient to produce religious experience, it would not follow that the latter is just the former in disguise. In the first place, it is not at all easy to discover the exact meaning of this metaphorical phrase when it is applied to psychological topics. And, if we make use of physical analogies, we are not much helped. A mixture of oxygen and hydrogen in presence of a spark is necessary and sufficient to produce water accompanied by an explosion. But water accompanied by an explosion is not a mixture of oxygen and hydtogen and a spark 'in a disguised form', whatever that may mcan.
      Now I think that the present rather vaguely formulated objection to the validity of the claims of religious experience might be stated somewhat as follows. 'In the individual religious experience originates from, and always remains mixed with, sexual desires and emotions. The other generative factor of it is the religious tradition of the society in which he lives, the teachings of his parents, nurses, schoolmasters, etc. In the race religious experience originated from a mixture of false beliefs about nature and man, irrational fears, sexual and other impulses, and so on. Thus the religious tradition arose from beliefs which we now recognize to have been false and from emotions which we now recognize to have been irrelevant and misleading. It is now drilled into children by those who are in authority over them at a time of life when they are intellectually and emotionally at much the same stage as the primitive savages among whom it originated. It is, therefore, readily accepted, and it determines beliefs and emotional dispositions which persist long after the child has grown up and acquired more adequate knowledge of nature and of himself:'
      Persons who use this argument might admit that it does not definitely prove that religious beliefs are false and groundless. False beliefs and irrational fears in our remote ancestors might conceivably be the origin of true beliefs and of an appropriate feeling of awe and reverence in ourselves. And, if sexual desires and emotions be an essential condition and constituent of religious experience, the experience may nevertheless be veridical in important respects. We might merely have to rewrite one of the beatitudes and say 'Blessed are the impure in heart, for they shall see God'. But, although it is logically possible that such causes should produce such effects, it would be said that they are most unlikely to do so. They seem much more likely to produce false beliefs and misplaced emotions.
      It is plain that this argument has considerable plausibility. But it is worth while to remember that modern science has almost as humble an ancestry as contemporary religion. If the primitive witch-smeller is the spiritual progenitor of the Archbishop of Canterbury, the primitive rain-maker is equally the spiritual progenitor of the Cavendish Professor of Physics. There has obviously been a gradual refinement and purification of religious beliefs and concepts in the course of history, just as there has been in the beliefs and concepts of science. Certain persons of religious genius, such as some of the Hebrew prophets and the founders of Christianity and of Buddhism, do seem to have introduced new ethico-religious concepts and beliefs which have won wide acceptance, just as certain men of scientific genius, such as Galileo, Newton, and Einstein, have done in the sphere of science. It seems somewhat arbitrary to count this process as a continual approximation to true knowledge of the material aspect of the world in the case of science, and to refuse to regard is as at all similar in thc case of religion. Lastly, we must remember that all of us have accepted the current common-sense and scientific view of the material world on the authority of our parents, nurses, masters, and companions at a time when we had neither the power nor the inclination to criticize it. And most of us accept, without even understanding, the more recondite doctrines of contemporary physics simply on the authority of those whom we have been taught to regard as experts.
      On the whole, then, I do not think that what we know of the conditions under which religious beliefs and emotions have arisen in the life of the individual and the race makes it reasonable to think that they are specially likely to be delusive or misdirected. At any rate any argument which starts from that basis and claims to reach such a conclusion will need to be very carefully handled if its destructive effects are to be confined within the range contemplated by its users. It is reasonable to think that the concepts and beliefs of even the most perfect religions known to us are extremely inadequate to the facts which they express; that they are highly confused and are mixed up with a great deal of positive error and sheer nonsense; and that, if the human race goes on and continues to have religious experiences and to reflect on them, they will be altered and improved almost out of recognition. But all this could be said, mutatis mutandis, of scientific concepts and theories. The claim of any particular religion or sect to have complete or final truth on these subjects seems to me to be too ridiculous to be worth a moment's consideration. But the opposite extreme of holding that the whole religious experience of mankind is a gigantic system of pure delusion seems to me to be almost (though not quite) as far-fetched.