The nature of knowledge and its relation to certainty, belief, and doubt has been a philosophical topic ever since philosophy began. Philosophers in ancient Greece established a number of truths about knowledge. (1) Knowledge can only be of what is true. (2) A belief is only knowledge if it can appeal to some kind of warrant. (3) One who claims knowledge must have a resolute commitment to the proposition claimed to be known. If I claim to know something, then I exclude the possibility of being at some later time rightly converted to a different view.

Ancient philosophers, however, tended to demand too high a level of warrant for knowledge. Lurking behind much ancient epistemology was a fundamental logical fallacy. “Whatever is knowledge must be true” may be interpreted in two ways.

1. (1) Necessarily, if p is known, p is true

2. (2) If p is known, p is necessarily true.

(1) is uncontentiously true; but if (2) is taken as equivalent to it then only necessary truths can be known. Such a position is suggested from time to time by passages in Plato and Aristotle.

From Augustine onwards philosophers of various schools agreed that in addition to knowledge of necessary truths there was also knowledge of contingent, empirical matters. For a truth to be known it did not need to be self-evident: it could also be evident to the senses. However, a modern epistemological fallacy took the place of the ancient one. Descartes and Locke thought that the way something struck one could guarantee its truth: Descartes spoke of “clear and distinct perception”, and Locke spoke of an “evident lustre” attaching to certain propositions. But there is no purely internal method of discovering which of one's beliefs deserve the name of knowledge. The best one can hope for is to acquire proficiency in attaching the right degree of commitment to each belief.

It is important for human beings to strike the right balance in belief. One can err by believing too much or believing too little. The person who believes too much suffers from the vice of credulity; the person who believes too little is guilty of excessive incredulity or scepticism. If you believe too much your mind will be cluttered with many falsehoods; if you believe too little you will be deprived of much valuable information. Let us call the virtue which stands in the middle between scepticism and credulity the virtue of rationality.

It was Aristotle who first shewed us that virtues stand in a mean, that is to say that each virtue is flanked by two opposing vices. Aristotle did not himself identify any virtue which had belief as its field of operation. That is because he focussed his attention on those mental conditions, such as knowledge and understanding, that have only truth as their object. Because only what is true can be known, there was no need for Aristotle to identify a virtue which was possessing just the right amount of knowledge: one cannot—in any literal sense—know too much. But belief, as Aristotle well knew, is a state of mind that may be true or false. If p is false, then I do not know that p, however much I may think I do; but a belief of mine may be false and yet remain a perfectly genuine belief. There is room, then, for a virtue that determines the mean of belief.

Plato in the Theaetetus offered a definition of knowledge as a true thought with a logos. He found himself unable, however, to explain what was this logos that turned true belief into knowledge. However, his definition began a tradition of defining knowledge as justified true belief. This definition was classical over millennia, but in the last century a number of philosophers have cast doubt on this traditional definition.

Peter Geach once wrote to me “Belief is a disposition, expressed in acts of judgement, though not only that way… .[It] is verbally expressed in assertion, which answers to acts of judgement… . Knowledge differs from belief by being a capacity not a disposition. No added factor can turn true belief into knowledge: the Theaetetus problem is a pseudo-problem.”

There is indeed knowing how as well as knowing that, so that one cannot define knowledge tout court as a kind of belief. But knowing that p does involve believing that p: that is to say, it involves a similar disposition to judge and assert that p. “I know that p but I don't believe that p” is absurd.

However, a true belief that p is not sufficient to constitute knowledge that p. I may believe that p, but be quite ready to change my mind if evidence turns up that not-p. But if I claim to know that p, I have a much stronger commitment to the truth of p. I am claiming that nothing should make me change my mind about it. No doubt I realise that I may, at some future date, change my mind; but so long as I am claiming knowledge I am claiming that I would be wrong to do so.

Of course, we often claim to know that p, and later find out that p is false. That shows that we did not ever know that p, however strongly we thought we did. I cannot then say “I knew that p, but p was false”; rather, I say, “I was certain that p, but p was false” Certainty involves the same degree of commitment as knowledge—I cannot say “I am certain that p, but p is false”, However, the concept of certainty does not have the same conceptual link to truth as the concept of knowledge.

In this paper I want to address the general epistemological problem of the nature of knowledge, certainty, and belief. But I want to take, as a focus for discussion one particular topic of belief, namely belief in God. I want to consider how far such belief is reasonable. As a text on which to hang the discussion I will take Richard Dawkins book The God Delusion. (Bantam 2006; henceforth GD) I find myself in agreement with perhaps 90% of what Dawkins says, and I shall have little to say about the areas of our agreement. But because of the 10% difference between us I end up in quite a different position with regard to the rationality of religion.

Though, like Dawkins, I am not myself a believer, I take a much more tolerant view than he does of the possibility that religious belief may be reasonable. I differ from him, that is to say, about the relationship between faith and reason. It is wrong to make too stark and simple a contrast between faith and reason. Indeed in these days of post-post-modernism it is often the proponents of faith who are loudest in their defence of the rights of reason.

Faith and reason are sometimes presented as two contrasting sources of information about religious matters. Thus a Christian theologian might maintain that there are some truths about God (e.g. that he is omnipotent) that can be discovered by unaided reason, while there are others (e.g. that there are three persons in a single God) that cannot be attained without the grace of faith. But at least since St Thomas Aquinas there has been a traditional Christian teaching that while some truths are not attainable by pure reason, no revealed doctrines are contrary to reason and faith is itself a reasonable frame of mind. Later in this paper I shall address the distinction between natural theology (the work of unaided reason) and religious faith (which is claimed to be the work of grace). For the present, like Dawkins, I will lump the two together as “belief in God”.

Let me begin by saying that I am in accord with Dawkins in rejecting traditional philosophical arguments for the existence of God, whether ontological, cosmological, or experiential.

I believe that critics of the ontological argument, from Aquinas to Frege, have show that it fails to establish the existence of God. If it were valid, then “God exists” would be an analytic proposition: “exists” would be a predicate that was tacitly contained in the subject “God”. But as Kant insisted, all statements of real existence are synthetic, and “exists” is not a predicate at all. Abelard in the twelfth century and Frege in the nineteenth century urged us to rephrase statements of existence so that “exists” does not even look like a predicate. “Angels exist” should be formulated as “some things are angels”. This has the advantage that it does not make it appear that when we say “Angels do not exist” we are first positing angels and then rejecting them.

However, this is not a final solution of the ontological argument, because the question may be raised about what counts as something. Does our quantifier range over possible as well as actual objects? If so, then, following the lead of some recent theist philosophers, we may argue thus. A necessary being is one that exists in all possible worlds. So defined, a necessary being must exist in our world, the actual world. Our world would not exist unless it were possible; so if God exists in every possible world, he must exist in ours.

I believe that the framework of this argument, the apparatus of possible worlds, is philosophically incoherent, though I must admit that it is accepted by many atheist as well as theist philosophers. I believe that Kant was right to insist that whether there is something in reality answering to a concept of mine cannot itself be part of my concept. A concept has to be determined prior to being applied to reality, otherwise we have nothing definite to try out on the world. That there is a God cannot be part of what we mean by God. (Still less can it be all that we mean by God, as is contended by those who tell us that God's essence is existence).

I also agree with Dawkins that God's existence cannot be established by appeal to experience. If there is a God with the attributes ascribed to him by Western theism then he cannot be the subject of any cognitive activity analogous to the operation of our senses. We cannot have a sixth sense that detects that God is here and not there, as we can see that something is red at one end and not at another, or which detects that God was here a moment ago and is not now, as we can hear a noise that suddenly stops. The whole context within which talk of sense-experience makes sense is lacking in the case of an alleged sensus divinitatis.

Aquinas' famous five ways are given a swift despatch by Dawkins. I think that he misdescribes them in some ways; but even if Aquinas is given a lot more rope, I agree that he hangs himself in the end. Nearly forty years ago I wrote a book to that effect (The Five Ways, Routledge 1969). However, one of Aquinas' proofs, the fifth way, deserves further treatment. It resembles the argument from design, Kant's physico-theological proof. Its case is that the ordinary teleology of non conscious agents in the universe entails the existence of an intelligent universal orderer.

In considering this argument we must begin by recognizing that there is a difference between design and purpose. Design differs from purpose because design is purpose preceded by an idea: a thought, or blueprint, in somebody's mind. If the world is designed, then there was a precedent idea in the mind of the creator—what, in the fourth gospel, is called the logos or Word. In our kind of mind, the idea that precedes an artefact is not anything simple or timeless, but something that gets built up by research and experiment. The argument from design can only establish its conclusion if it is possible for there to be a quite different kind of mind: a divine, extra-cosmic, simple and eternal mind. That, to my mind, is the greatest difficulty with the argument.

If we are to attribute intelligence to any entity—limited or unlimited, cosmic or extra cosmic—we have to take as our starting point our concept of intelligence as exhibited by human beings: we have no other concept of it. Human intelligence is displayed in the behaviour of human bodies and the thoughts of human minds. If we reflect on the actual ways in which we attribute words such as “know” “believe” “think” “design” “control” to human beings, we realize the immense difficulty there is in applying them to a putative being which is immaterial, ubiquitous, and eternal. With a degree of anthropomorphism we can apply mentalistic predicates to animals, computers, institutions; to organisms that resemble us or artefacts that are our creations; but there are limits to anthropomorphism, and an extra-cosmic intelligence appears to me to be outside those limits. It is not just that we do not, and cannot, know what goes in God's mind; it is that we cannot really ascribe a mind to a God at all. The language that we use to describe the contents of human minds operates within a web of links with bodily behaviour and social institutions. When we try to apply this language to an entity outside the natural world, whose scope of operation is the entire universe, this web comes to pieces, and we no longer know what we are saying.

Most critics of the argument from design—particularly since Darwin—quarrel not with the conclusion but the premises of the argument. They object to the appeal to the apparent existence of teleology in the world around us.

In the course of history teleology has received drastically different treatments by different great philosophers. One of Aristotle's four causes was the final cause, the end, the purpose, the good to be achieved by some action. Aristotle, more than any other philosopher, emphasised the importance of teleology in the world; but he was no supporter of the argument from design, because he was well aware of the distinction between design and purpose. The budding of a rose and the building of a spider's web were no less teleological than human activities; but Aristotle knew better than to attribute consciousness to roses and spiders. Teleology, for him, was a basic fact about the cosmos, and no extra-cosmic designer was needed to explain it.

It was Aquinas who formulated the argument from purpose to design. Things without awareness, he argued in the fourth way, do not tend towards a goal unless directed by something with awareness and intelligence, in the way that an arrow is aimed by an archer. The ultimate designer, the arch-archer, we call God.

Descartes revolutionised philosophy by expunging teleology altogether. He eliminated, outside the human realm, purpose as well as design. Final causation, he thought, was a piece of scholastic nonsense. He rejected the explanation of gravity in terms of attraction between bodies, on the grounds that this postulated in inert entities knowledge of a goal or terminus.

With Darwinism we return to the position of Aristotle: the world contains purpose but not, outside the human realm, design. It is sometimes thought that Darwinism gave the final death-blow to teleology; but that is the opposite of the truth. Darwinian scientists have not given up the search for final causes. On the contrary, contemporary biologists are much more adept at discerning the functions of structures and behaviour than their ancient, medieval, or Cartesian predecessors. Darwin agreed with Aquinas against Aristotle that teleology was not a basic fact, but something needing explanation. His achievement was to make teleological explanation respectable by offering, in natural selection, a recipe for translating it into a naturalistic explanation that made no call on design.

Natural selection, however, cannot be offered as the sole and sufficient explanation of the history of the universe. When neo-Darwinians offer to explain the entire cosmos, I find difficulties at three main points: the origin of language, the origin of life, and the origin of the universe.

Language is conventional, that is to say it is governed by rules. Being rule-governed is being different from being governed by causal laws. Rules differ from such laws in several ways. Rules may be broken; not so causal laws, short of a miracle. To be governed by a rule you must be conscious of it at least to the extent that you are aware of any breach of it. But there is no need for you to be aware of being governed by a causal law: the planets know nothing of Kepler's laws.

The problem with an evolutionary account of language is this. Explanation by natural selection of the origin of a feature in a population presupposes the occurrence of that feature in particular individuals of the population. Natural selection might favour a certain length of leg, and the long-legged individuals in the population might outbreed the others. But for that kind of explanation of features to be possible, it must be possible to conceive the occurrence of the feature in single individuals. There is no problem in describing a single individual as having legs n metres long. But there is a problem with the idea that there might be a single human language user.

It is not easy to see how the human race may have begun to use language because language using individuals outbred the non-language users. This is not a difficulty in seeing how spontaneous mutation could produce a language-user; it is the difficulty of seeing how anyone could be described as a language-using individual at all before there was a community of language-users.

Of course, there are animal systems of communication that have some similarity with human language, and human pets can respond to names and obey commands. But t human language is separated by a gulf from the communicative abilities of other animals. The essential logical feature of human language is not that it contains nouns and verbs, but rather that it contains negation, conditionality, quantification, and modality. It is the words “not” “if” “some” and “therefore” that are the marks of rationality; and there are no animal equivalents of these.

If it is difficult to see how language could originate by natural selection, it is even more difficult to see how life could originate that way. However successful natural selection may be in explaining the origin of particular species of life, it clearly cannot explain how there came to be such things as species at all. That is to say, it cannot explain how there came to be true breeding populations, since the existence of such populations is one of the premises on which explanations in terms of natural selection rest as there starting point.

This is a point which is accepted, and indeed stressed, by Dawkins in his recent book. The origin of life, he says is a question for chemists, and he is no chemist. The chemists haven't been able to replicate the origin of life, but may be they will soon. In the meantime. Dawkins is happy to accept that it came about by a lucky chance: after all, if there are a billion billion planets, then the odds are a billion to one that life will have started on a billion of them, of which the earth is one. But when he turns to the ten million different species of life on our plant, each well adapted to a particular way of life, and asks “Could we get away with the “huge number of planets” argument to explain all these separate illusions of design? No we could not, repeat not. Don't even think about it. This is important, for it goes to the heart of the most serious misunderstanding of Darwinism. It doesn't matter how many planets we have to play with, lucky chance could never be enough to explain the lush diversity of living complexity on Earth in the same way as we used it to explain the existence of life here in the first place. The evolution of life is a completely different case from the origin of life.”(GD, 138.)

To explain the origin of life, Dawkins invokes a planetary version of the anthropic principle. He states it thus.

On the face of it, the planetary conditions for our support are immensely improbable and call for explanation. No, says this anthropic principle, far from being improbable they are necessary; and necessary truths call for no explanation.

We need to look more closely at the use of “necessary” here. There are at least two senses of the word “necessary”. Something is metaphysically necessarily if it follows from necessary truths. But there is also epistemic necessity, which is the counterpart of epistemic possibility. P is epistemically possible when we can say “for all we know to the contrary p may be the case.” If not-p is not epistemically possible, then p is epistemically necessary. Something may be epistemically necessary without being metaphysically necessary: as I write it is episemtically necessary, but not metaphysically necessary, that George W. Bush is President of the U.S. The anthropic principle argues from the epistemic necessity of conditions favourable to human life to their metaphysical necessity.

The anthropic principle is usually presented in a cosmic rather than a planetary version. Martin Rees has listed six fundamental constants that are believed to hold throughout the universe. If any one of these differed very slightly from its actual value life in the universe would be impossible. How are we to explain this fine tuning? It is here that the anthropic principle is invoked: the constants had to have that value or we would not be here to calculate them.

Dawkins is aware of the difficulty here, and borrows from John Leslie the analogy of a man sentenced to death by firing squad. It is just possible that all ten men of the firing squad will miss their victim. With hindsight the survivor who finds himself in a position to reflect upon his luck can cheerfully say “Well, obviously they all missed or I wouldn't be here thinking about it.” “But he could still” Dawkins says “forgivably, wonder why they all missed” I suggest that the word “forgivably” should be replaced by “most reasonably”.

Dawkin's answer to the puzzle is that (as Rees maintains, in company with Giordano Bruno) that there are many universes; the laws and constants of any one universe are by-laws. “The anthropic principle” as Dawkins puts it “kicks in to explain that we have to be in one of those universes (presumably a minority) whose by-laws happened to be propitious to our eventual evolution and hence contemplation of the problem”. (GD,145)

Dawkins is in fact offering two different answers to the problem. The anthropic principle says there is nothing to be explained—there was not improbability in the case, but necessity. The multiverse thesis says that there is indeed something to be explained, and it is to be explained by a priori probability. This can be done if we simply call into existence billions of universes—all, ex hypothesti, inaccessible to scientific inquiry.

The first explanation is a fallacy, resting on confusion between epistemic and metaphysical necessity. The second explanation is a piece of metaphysical speculation (in Rees no less than in Bruno). Its only merit seems to be that is to provides an alternative to intelligent design.

So far we have been discussing the kind of universe we have. But why is there a universe of any kind? The most fundamental reason of postulating an extra-cosmic agency of any kind is surely the need to explain the origin of the universe itself. Most philosophical arguments for the existence of God are only sophistications of the cry of the simple believer “God must exist, else where did the world come from?”

It is wrong to say that God provides the answer to the question “Why is there something rather than nothing?”. As Bede Rundle has shown in an engaging book with that title (OUP, 1999), that question is ill-conceived; the proposition “there is nothing” cannot be given a coherent sense, and therefore there is no need to ask why it is false. It is not the existence of the universe that calls for explanation, but its coming into existence.

At a time when philosophers and scientists were happy to accept that the universe had existed for ever, there was no question of looking for a cause of its origin, only of looking for an explanation of its nature. But when it is proposed that the universe began at a point of time measurably distant in the past, then it seems perverse simply to shrug one's shoulders and decline to seek any explanation. We would never, in the case of an ordinary existent, tolerate a blithe announcement that there was simply no reason for its coming into existence, and it seems irrational to abandon this principle when the existing thing in question is all pervasive, like the universe. If only an intelligent creator were conceivable, he would surely be a more persuasive solution to the problem.

My reaction to the difficulty of either proving or disproving God's existence is agnosticism: to say that we do not know either way. Often, both theist and atheist philosophers, instead of offering arguments, adopt a strategy that might be called grabbing the default position—that is, a tactic of throwing the burden of proof on the opponent. But it is agnosticism that is the true default position. A claim to knowledge needs to be substantiated, ignorance only has to be confessed. Moreover, a claim to know that God exists, or a claim to know that God does not exist, is an absolute commitment. The profession of doubt is no such thing.

Dawkins is contemptuous of agnosticism, and prefers atheism. He believes that he has an argument to disprove God's existence. A designer God, he maintains, cannot be used to explain organised complexity, because any God capable of designing anything would have to be complex enough to demand the same kind of explanation in his own right. (GD, 109) He calls this argument “The Ultimate Boeing 747 argument”, in tribute to Fred Hoyle who once said that the probability of life originating on earth is no greater than the chance that a hurricane, sweeping through a scrapyard, would have the luck to assemble a Boeing 747. God, according to Dawkins, is the ultimate 747.

A traditional theist would say that Dawkins' argument misrepresented the notion of God in two ways. First of all, God is as much outside the series complexity/simplicity as he is outside the series mover/moved. He is not complex as a protein is, nor for that matter is he simple as an elementary particle is. He has neither the simplicity nor the complexity of material objects. Secondly, he is not one of a series of temporal contingents, each requiring explanation in terms of a previous state of the Universe: unchanging and everlasting, He is outside the temporal series. What calls for explanation is the origin of organized complexity that is to be explained. But God had no origin and is neither complex nor organized.

The principle that what is complex must be designed by what is even more complex is a metaphysical principle, because it is meant to apply outside the world of science as well as within it. As a metaphysician myself I have no objection to the use of metaphysical principles. But is this one valid? Even on earth is it true that a designer must be more complex than his creation? Is it metaphysically impossible for a human to design a computer containing more bytes than there are cells in the human brain? This, it may be objected, is an inappropriate comparison; but doesn't that show that even at this level the notion of complexity is not a simple notion, and there are various kinds of complexity.

In particular, since design is purpose preceded by an idea, we need to consider the notion of complexity as applied to thinking. A thought, or idea, does not have the same complexity as its expression in a written design or blueprint would have. A thought does not have spatial or temporal parts in the way that the sentence that expresses it has. A thought is a unified whole, and is not made up of a succession of parts in the way that a sentence is made up of successive words. Even in our mundane world, therefore, there is reason to doubt Dawkins' principle.

However, Dawkins addresses the question not only at the metaphysical level. Elsewhere he claims that the existence of God is a scientific hypothesis like any other. A universe with a supernaturally intelligent creator, he says, is a very different kind of universe from one without. He agrees that the difference is not easy to test in practice but whether God exists is a scientific question to which one day we may know the answer and in the meantime we can say something about the probabilities. (GD, 48)

He sets up a spectrum of probabilities, between opposite certainties.

1. 1 I know there is a God

2. 2 I do not know but strongly believe

3. 3 I am uncertain but inclined to believe

4. 4 I think God's existence and non existence equiprobable

5. 5 I am uncertain but inclined to be sceptical

6. 6 I think God's existence is very improbable

7. 7 I know there is no God.

Dawkins places himself in category 6, as de facto atheist. He could equally well be described as an agnostic, if an agnostic is, as its etymology suggests, someone who does not know whether there is a God or not.

Interestingly, Aquinas in his treatment of faith (Summa Theologiae II-Iiae,2,1) sets up exactly the same spectrum. He calls (1) scientia (2) opinio (3) suspicio (4) dubitatio; which we may translate as (1) knowledge (2) conviction (3) guess (4) doubt.Footnote ¹

The scales given by Aquinas and Dawkins are an interesting way of classifying different degrees of commitment to a proposition, but they are inadequate to provide a total classification of epistemic states. States of mind are to be distinguished along not one but three axes: degree of commitment, conformity to the facts, type of warrant. Thus, e.g. knowledge differs from certainty not by degree of commitment, but by the fact that knowledge is only of the truth but certainty can be false. Again, in talking of certainty we must distinguish first, second, and third person. Contrast “I am certain that p but p is false” “You are certain that p but p is false” “It is certain that p but p is false”.

We must take into account not only commitment and veracity, but warrant. John Locke says we should not entertain any proposition with greater assurance than the evidence it is built on will warrant. “Whoever goes beyond this measure of assent, it is plain, receives not truth in the love of it, loves not truth for truth-sake, but for some other by-end”. (Essay 4, xvi) John Henry Newman contested this, and showed that knowledge need not be based on evidence. I may be justifiably certain of something and yet there be no evidence for it because it is itself more certain than anything that could be offered in evidence.

Wittgenstein, following Newman, identified a set of propositions of which we are certain, but where our certainty is not based on evidence and is not the result of any inquiry. He gave as examples “the earth has existed for many years past “cats do not grow on trees” “human beings have forebears.” He was not willing to say that we knew such propisitions, but he said that they stood fast in our picture of the world; they had a fundamental role which was not the result of, but prior to, our methods of inquiry and evidence gathering. (On Certainty, passim)

In a number of works Alvin Plantinga has argued that belief in God may be regarded as basic in this way: a person may be within her epistemic rights in believing in God without having any proof or evidence of God's existence. Bertrand Russell was asked what he would say if he died and was confronted by God asking why he had not believed in him. “Not enough evidence, God, not enough evidence”. If Newman, Wittgenstgein and Plantinga are right, Russell's answer does not deserve a pass.

Plantinga thinks that belief in God can be as basic an item in someone's noetic structure as the belief that other people have minds. I don't myself accept that belief in the existence of God can rightly occupy a basic role comparable to the existence of other people, but rather than explain why I think that, I want to make a distinction long overdue between simple belief in God's existence, and actual religious faith.

Faith, as I understand it, is the acceptance of the testimony of a sacred text or of a religious community. The two, in fact, go together, because if the sacred texts are taken as guides to practical life, their authority is inseparable from the authority of the religious officials whose role is to interpret them. In the Judeo-Christian tradition for instance the very notion of “the Bible” as a single entity depends on the various authorities throughout our history who have established the canon. However impressive individual books may be, to see them as elements of a single revelation containing some or all of the other books is already tacitly to accept a religious authority that defines the canon. One might gather together the works of Homer, Hesiod, Aeschylus, Sophocles, Euripides, Herodotus and Thucydides into an epitome of Greek thought. The anthology would share a common cultural tradition and cohere as well or ill as the bible does. But we would not treat it as a single book, to be treated differently from all other books, because there has never been a Hellenic rabbinate or episcopate to canonize such a collection.

The common characteristic of faith in almost all religious traditions is its irrevocability. A faith which is held tentatively is no true faith. It must be held with the same degree of certainty as knowledge. In some traditions the irrevocability of faith is reinforced by the imposition of the death penalty for apostasy, which is the abandonment of faith.

Aquinas, from whom I have drawn my account of religious faith, notes that it does not fit into the categorization of mental states which he, anticipating Dawkins, set out. “The state which is belief involves a firm adhesion to one side of the question. In this a believer is in the same position as someone who has knowledge or understanding; and yet his assent is not warranted by any clear vision, so that in that respect he resembles someone who doubts, guesses, or is convinced.”Footnote ²

Faith, then, resembles knowledge in being irrevocable, but differs from it in being a commitment in the absence of adequate evidence. This is well brought out by Newman. He says that the grounds on which faith is accepted—the evidence for there having been a revelation, for the divine authority of books or teachers—is a matter of conjecture, not of knowledge. “It starts from probability, yet it ends in peremptory statements… It believes an informant admit doubt, yet accepts his information without doubt.”

It is religious faith, rather than theistic belief, that is the main target of Dawkins' book. “What is really pernicious is the practice of teaching children that faith itself is a virtue. Faith is an evil precisely because it requires no justification and brooks no argument… Suicide bombers do what they do because they really believe what they were taught in their religious schools: that duty to God exceeds all other priorities, and that martyrdom in his service will be rewarded in the gardens of Paradise.” (GD, 308)

It is too much to say that faith requires no justification: many religious people offer arguments not just for belief in God but for their particular creed. What is true is that the kinds of arguments they offer cannot be claimed to have anything like the degree of warrant that would justify the irrevocable commitment of faith.

It is true that faith brooks no argument, not in the sense that the faithful are unwilling to offer responses to criticisms, but that no argument will make a true believer give up his faith, and this is something he is resolved on in advance of hearing any argument.

Some years ago I expressed an opinion similar to Dawkins complaint. “Faith” I wrote “is not, as theologians have claimed, a virtue, but a vice, unless a number of conditions can be fulfilled. One of them is that the existence of God can be rationally justified outside faith. Secondly, whatever are the historical events which are pointed to as constituting the divine revelation must be independently established as historically certain.” (What is Faith, OUP, 1992, 57)

It is a particular difficulty for the rationality of faith that there are so many alleged revelations that conflict with each other. One thing we know for certain is this: if any sacred text is literally true, then most are literally false. Of course, the incompatibility between conflicting revelations leaves it open as a logical possibility that just one of them is true while all the others are false. This is certainly not a possibility that can with decency be ruled out a priori by someone who believes that just by existing in this universe we are defeating odds of a billion to one. It seems to me, however, that if there is any truth in any religious revelation it is more likely that each of them is a metaphor for a single underlying truth that is incapable of being expressed in literal terms without contradiction. In this way religion would resemble poetry rather than science.

To say that religious language is not literal, and to say that different religious creeds therefore do not contradict each other, is not to say that all religions are of equal worth. The mode of utterance of Shakespeare and that of William McGonagall is poetic in each case; that does no mean that the writings of each of them display an equally valuable insight into human nature. But the fact that theological language cannot be literal provides a reason for toleration in religion. That is to say, theological propositions cannot contradict each other in the straightforward way in which empirical propositions do. Hence, there is not the head-on clash between different theologies, and different religions, which has been used to justify the persecution and killing of one religious group by others.

Dawkins' book suggests that all those who believe in God are unreasonable in so doing. I disagree. Those who claim to know that there is a God, I have agreed, are making a claim that is not justified; but so too are those who claim to know there is no God. But a belief in God, falling short of certainty, is not open to the same objection. A belief may be reasonable, though false. If two oncologists tell you that your tumour is benign, then your belief that it is benign is a reasonable belief even if, sadly, it is false. In the case of many people in many cultures, I maintain, religious belief, even if false, may well be reasonable. (So, too, may tentative, non-dogmatic atheism). But I think belief in God reasonable only if it is based on considerations available to all humans: not if it is claimed on the basis of a special message to oneself or to the group that one belongs.

Because I think belief in God can be reasonable I think Dawkins is wrong to object to bringing up children in belief in religion. The education of children is impossible without narrative and ceremony, and growing up is a matter of knowing what to discard and what to maintain. I doubt if Dawkins objects to telling children about Santa Claus; it does not lead to an adult society of bigoted Santaclausians. Many intellectual Christians, as they mature, abandon other bits of the Christmas story: I know priests in good standing who do not believe that Jesus was born in Bethlehem. Disbelieving in religious narratives, however, does not necessarily mean discarding them: it means, as I suggested earlier, removing them from the history section of one's mind into the poetry section. I agree with Dawkins that faith, as an irrevocable commitment, is not reasonable when given to a false proposition. But I see nothing unreasonable in believers having the degree of commitment to their church, synagogue or mosque that they might have to a political party or social community.

It is the degree of commitment involved in faith, rather than its religious object, that is what is really objectionable; and the history of nazism, fascism, and communism made this abundantly clear in the last century. Not all fanaticism is religious fanaticism, and I found unconvincing Dawkins attempt to show that Hitler was a closet Catholic.

In answer to the question whether religious people are better or worse than non-religious people, a friend of mine gave a sage answer. “Religion” he said “makes nice people nicer and makes nasty people nastier”.

When arguing, a moment ago, that religious language should be treated as metaphorical I mentioned that metaphorical propositions do not contradict each other in the way that bigots believe. But of course much religious language is in the imperative mood, and even if propositions don't contradict each other, commands certainly can. The question then arises, how should one deal with commands emanating from a religious position that one believes to be literally false. This is a practical, political, question, not a merely philosophical one.

Like Dawkins I believe that the state owes no special respect to religious beliefs as such. Religious conscientious objectors should be treated on equal terms with atheist objectors who oppose a war on moral grounds: general pacifism has no priority over qualified pacifism. Religious bodies, in order to obtain tax exemption as charities, should be under an obligation to prove public benefit—that is, benefit that can be assessed as such by others than co-religionists.

On the other hand, it may be said that one of the great benefits that religion has conferred on the human race is that it has instilled into people's hearts the idea that the governments they live under are not the supreme arbiters of right and wrong: that there is an authority superior even to the most omnicompetent totalitarian government. A world in which everyone believed that there was no moral authority superior to their rulers would not necessarily be preferable to a world in which everyone saw rulers as answerable to God.

Of course, there can be a world which is worse than either: a world in which rulers believe they have a special message from God authorising their policies in peace and war.