In the year 156 of the Christian era, Polycarp, Bishop of Smyrna, was arrested and brought before the magistrate, charged with being a Christian. He was in his eighties, and his age and frailty prompted the magistrate to offer him a quick discharge if he would acknowledge the divine spirit of the emperor and say 'Away with the atheists.' The latter, at least, you might think would not be difficult for a bishop; but of course at this period an atheist was someone who refused to take part in the civic cult of the empire, to perform public religious duties and take part in the festivals of the Roman city. Christians were atheists, by this definition; Polycarp had a problem after all. His response, though, was an elegant turning of the tables. He looked around slowly at the screaming mob in the amphitheatre who had gathered for the gladiatorial fights and public executions, and, says our eyewitness chronicler, he groaned and said, 'Away with the atheists.'

The magistrate did not fail to grasp the theological point, and Polycarp was duly condemned to be burned alive. But this poignant story is one well worth pondering for reasons beyond the study of early Christianity. It is a reminder that 'atheism' may be a less simple idea than either its defenders or its attackers assume. People often talk as though 'atheism' were a self-contained system, a view of the world which gained its coherence from a central conviction – that there is no transcendent creative power independent of the universe we experience. But the story of Polycarp reminds us that to understand what atheism means, we need to know which gods are being rejected and why.

Thus an early Christian was an atheist because he or she refused to be part of a complex system in which political and religious loyalties were inseparably bound up. 'Atheism' was a decision to place certain loyalties above those owed to the sacralised power of the state. But, moving across the world of faiths, Buddhists are sometimes described as atheists by puzzled observers, aware of the fact that Buddhist philosophy has no place for a divine agent and that Buddhist practice concentrates exclusively upon the mind purifying itself from self-absorption and craving; here, 'atheism' is a strategy to discipline the mind's temptation to distraction by speculative thought. Whether or not there is a transcendent creator is irrelevant to the mind's work; preoccupation with this is a self-indulgent diversion at best, and at worst a search for some agency that can do the work only we can do.

Neither of these has much in common with the atheism characteristic of Western modernity, which draws much of its energy from moral protest. The God of Jewish and Christian faith is seen as an agent who has the power to prevent the world's evil yet refuses to do so, so that there is the appearance of a moral incoherence at the heart of this tradition. Or he is seen as an arbitrary tyrant whose will is inimical to the liberty of human creatures; or else as an impotent and remote reality, a concept given a sort of ghostly existence by human imagination. In all these instances, it is clear that the refusal of belief in God is something essential to human liberation. We cannot live with a God who is responsible for evil; we cannot grow up as human beings if what is demanded of us is blind obedience; we cannot mortgage our lives and our loving commitment to an animated abstraction. Atheism here is necessary to maturity, individually and culturally.

Even those who argue at length about the simply conceptual inadequacies, as they see it, of Western religion – classically, writers in the Bertrand Russell style – will frequently deploy the language of moral revolt as well. 'Protest atheism', as it is often called, has become a familiar element in the armoury of modern intellectual life, perhaps more often repeated than expounded, but culturally very powerful. The more austere objection to belief found in the positivism of the early to mid twentieth century – it is equally without meaning to affirm or to deny the existence of an agency whose existence could never be empirically demonstrated – has an ironic resonance with Buddhism, but is another component in the mind of Western modernity, even when the philosophical system from which it arises no longer has much credibility. This is atheism as the mark of supreme intellectual detachment, with the intellect defined as a mechanism for processing checkable information only, with everything else reduced to emotive noise. But the other great modern version of atheism is that which exposes religious talk as ideological – as an instrument of social control whose surface conceptual structure is designed to obscure its real function and to divert thought, emotion and energy from real to unreal objects. This is the essence of Marxist atheism, but it also has some relation to Nietzsche's unforgettably eloquent polemic against Christian faith.

The point is that atheism is to be defined as a system only by some dramatic intellectual contortions. A number of intellectual and spiritual policies involve or at least accompany the denial of certain versions of the divine, especially the divine as an active and intelligent subject; but in each case the denial is not intelligible apart from a specific context of thought and image, representation and misrepresentation of specific religious doctrines, and the overall system of which the denial is a part is not necessarily shaped by it. This is why the recent proposal in the United Kingdom that religious education in schools should give attention to 'atheism and humanism' as 'non-faith belief systems' alongside the traditional religions was based on some serious conceptual confusions and category mistakes. In the background is the pervasive assumption of modernity that the intellectual default position is non-religious; but what this fails to see is that non-religiousness is historically and culturally a complex of refusals directed at specific religious doctrines, rather than a pure and primitive vision invaded by religious fictions. And if this is so, either religious education has to locate non-religious positions in relation to what it is that they deny, or it will end up treating atheism as the only position not subject to critical scrutiny and the construction of a proper intellectual genealogy: not a welcome position for a rationalist to be in.

In fact the incorporation of critical positions into religious education is to be applauded. To see where the points of strain are to be found in a religious discourse and to seek to understand how a thoughtful and self-critical tradition can respond to them is essential to a proper grasp of religious identities. One of the weaknesses of the kind of religious education now common in schools (in the UK at least) is that it tends to describe the positions of faith communities as finished systems for which questions have been answered rather than (to borrow Alastair Macintyre's phrase) 'continuities of conflict', in which the moral, spiritual and intellectual tensions constantly press believers towards a fuller, more comprehensive statement of their commitments.

'If you meet the Buddha, kill him' is a well-known Zen dictum, from a tradition deeply aware that personal agenda and history are easily capable of distorting any supposedly clear vision of where enlightenment is to be found. Any conceptual form that can be given in the abstract to the Buddha (i.e. to the enlightened awareness)will take its shape from the unenlightened awareness, and so has to be dissolved. But this is not that different from the conviction of much Hindu thought, that the divine is 'not this, not that', never identifiable with a determinate object, or from the principle, deeply rooted in the Abrahamic faiths, that God cannot be given an 'essential' definition, classified as a kind of object. This may be expressed in the form of the apophatic theology of an Ibn Sina or Maimonides or Nicholas of Cusa: Ibn Sina (like Aquinas and all that flows from him) insists that there can be no answer to the question, 'What makes God divine?' as if some 'quiddity' could be identified that grounded a divine definition. God is God by being God – by being the necessary, uncaused active reality he is; nothing else. But the same point is made in wholly different idioms by twentieth century writers such as Karl Barth and Simone Weil. For Barth, all systems for which God is an object are unsustainable: he always speaks before we have words to answer, acts before we can locate him on some intellectual map. He is never 'available', though always present. And Simone Weil, in an argument of some complexity, concludes that when the human ego says 'God', it cannot be referring to any reality to which the name might be truthfully applied. Because the 'I' that says 'God' is always self-directed and so wedded to untruth, God cannot properly be spoken of. Any God my selfish mind can conceive is bound to be a false, non-existent God. The true God is known only in ways that cannot be reduced to theory or third-person language. If you meet God (in the language of systematic theology or metaphysics), kill him.

It seems that, in differing degrees, most major religious discourses require and cultivate unbelief – that is, unbelief in a divine agent who can be thought about as an agent among others, an instance of a type, a kind of life that can be defined in terms of something other or prior. Thus when we try to consider and understand atheism of any kind, our first question has to be what it is about some particular piece of speech about God is causing trouble, and whether it is in fact essential to a religious tradition's understanding of what it means by God or the divine. It may be, of course, that what is objected to really is what a religious tradition believes; but even if it is, it is crucial to explore where the points of strain are felt, so that convictions may be tested and if possible reinforced. So the challenge of atheism in its various guises is one that has the potential to deepen what is said about our commitments; not for nothing did Olivier Clement, the French Orthodox theologian, write about 'purification by atheism'. To come to the point where you disbelieve passionately in a certain kind of God may be the most important step you can take in the direction of the true God.

But what I want to suggest specifically in connection with the dialogue between the world faiths is that we spend more time looking at what is disbelieved in other religious discourses. A few years ago, an American theologian wrote a book (Christopher Morse, Not Every Sprit. A Dogmatics of Christian Disbelief) about Christian doctrine as a series of 'unbeliefs': what does Christianity commit you to denying about God and Christ? I wonder if the same method might not be illuminating as we look across the faiths. Just as in the case of atheism generally we learn what we are and are not really committed to, so in dialogue or trialogue or whatever between faiths, we might be able to learn from each other's disbeliefs, to be 'purified' by encountering and examining the protests and denials, the 'atheisms', of each other's views.

Let me try to illustrate; I shall concentrate chiefly upon the Abrahamic faiths, but it should be clear at the very start of this reflection that Buddhism is of special significance in its denial of any personal agency outside the bounds of the world. If we ask why such a denial is made, we must conclude, as suggested earlier, that there is an anxiety that projection on to an external deity fatally weakens the incentive to dealing with our own distraction and selfishness; and there is an anxiety that the very act of affirming the existence of such a deity constitutes an escape from the severely practical analysis of the mind's liberation. In response to this, all three of the Abrahamic faiths have to examine themselves carefully. If we believe in a source of energy, forgiveness and love independent of ourselves, how exactly do we prevent it from becoming a belief that weakens our responsibility and imprisons us in fantasy? We shall need (to say no more in detail) to establish that we are not looking for a supernatural agent to fill the gaps in our imperfect self-awareness and willingness to change, a consoling personality who is there to serve the needs of our idle and needy selves. We shall need to examine carefully those aspects of our language which themselves warn against just such a misunderstanding – and those which might most easily suggest it or nourish it. Our faith becomes self-aware in a fresh way; even if we end by saying – as we probably shall – that the Buddhist refusal of a personal God assumes that ascribing personality and objectivity to God must always be a simple projection of our need, and if we argue – as we probably shall – that nothing is served by denying the fact of our dependence on what lies beyond the world, we shall have been warned, sharply and constructively, of just how we may use our faith to reinforce what is least converted in us.

But how might this apply to the conversation between the Abrahamic traditions? Here are some suggestions.

The Jew disbelieves propositions like the following: God is free to disregard or rewrite the solemn promise made to a specific people at a point in history; God makes no specific demands on those God chooses to hold in the closeness consequent upon such a promise; God cannot deal adequately with the world by revealing the divine will, but has to intervene in allegedly more 'intimate' or direct ways.

The Christian disbelieves propositions like these: God needs to be persuaded by our virtue to love us or to act on our behalf; God is a solitary individual with a personality comparable to that of a human individual; God is metaphysically incapable of acting in and as a created and dependent being; God's action can have no impact upon physical processes.

And the Muslim disbelieves propositions like these: God is the compound of several distinct divine agents (whether an indeterminate number or just three); God wills that the divine purpose be realised only in the lives of a limited segment of society or humanity; God's will can be divorced from the supposed cultural limitations of its earliest definitive and complete expression; God is known by a complex of human approximations to truth.

As will be evident, there is overlap in these configurations; and all three traditions agree in disbelieving in a God who is one of the items that exists within the universe, who is subject to time and change as finite beings are, who shares the same conceptual territory as do the limited agents we are familiar with. Although Christianity and Judaism have increasingly been willing to entertain images suggesting vulnerability and suffering in the divine life, this is largely a modern development whose conceptual relation to the definitive doctrines of the religions is a rather uneasy business. Powerful devotional metaphors require careful handling in this context, and they should not obscure one of the most significant convergences that exists between the Abrahamic faiths – and indeed between these faiths and others – on the conviction that God is not a member of any class of existent beings. You'll recall the earlier reference to Ibn Sina, echoed to the letter by Aquinas.

That being said, however, the respective systems of disbelief I have sketched so briefly pose equally significant mutual challenges. Faced with the disbeliefs of another discourse, each of the three participants in the Abrahamic conversation should be prompted to ask whether the God of the other's disbelief is or is not the God they themselves believe in. If the answer were a simple yes, dialogue might be a great deal more difficult than it is; the reality of dialogue suggests that we do not in fact have to do with a simple 'atheism' in respect of the other's models of God. And part of the fascination and the spiritual significance of dialogue is the discovery of how one's own commitments actually work, and specifically how they work under pressure. Is Christianity that which Judaism as such denies? Is the affirmation of Christianity identical to the denial of what Jews believe as Jews? And so with Islam also; one of the darkest and most tragic parts of our history in relation to other faiths ('our' history being, for these purposes the history of all the Abrahamic faiths) is the construction of the other as the opposite. To pick up an idea which I have tried to develop elsewhere, we have to put behind us a picture of the world of faiths in which each is seen as answering the same questions, so that the respective 'performance' of different traditions can be categorised in terms of right and wrong answers to these questions. Binary oppositions do not serve us at all here.

So to some of the particulars, though we can only take a few examples. I begin with two Muslim disbeliefs and their impact on Christian-Muslim encounter. The Muslim disbelieves in a plurality of divine agents: so the Christian has to ask whether his or her belief is properly so characterised – and if it is not, to examine why and how it could come to be read that way. Thus the Christian may say, in the tradition of Augustine and Aquinas, that belief in the Trinity is not belief in three self-subsisting individuals sharing a divine nature; it is to claim that the life that is divine life exists as three utterly interdependent 'streams' of agency, which cannot be reduced to each other – an originating agency, a responsive agency, an excess of creative and eternal agency always free to replicate the pattern of origin and response. The Greek tradition avoids calling them 'persons' and prefers 'subsistents' (hupostaseis). This is a very abstract rendering indeed of the doctrine of the trinity, but one which avoids the distorting impression that Christians believe in some kind of divine society of individuals; and there are aspects of both theology and popular devotion that can give such an impression of a belief which Muslims (and Jews) find incomprehensible and inconsistent with belief in the oneness of God. The important clarification for the Christian is that divine oneness is not the oneness of an individual (where there may logically be more than on of its kind) – and this is actually something that can be agreed by the Jew and the Muslim, who (at least in the shape of their mainstream philosophical systems) would equally deny that God is an individual in that sense. The Muslim challenge pushes the Christian, now as in the Middle Ages, to clarify a fundamental point of belief.

Imagine, next, the Christian picking up a Muslim unbelief and challenging it. The Muslim does not believe in a 'church' that is socially a separate body from the political community at large as organised by divine law. The Christian however has a long tradition of expecting the body of believers to be in significant respects different from political society (think of Polycarp again). Is the Muslim attitude not tantamount to saying that nothing but a theocracy can express Islam? To which the Muslim might respond by saying that if God is a God who has the capacity to make known the divine will, and if there is ultimately one good for human society which is to be found in following that will, there can be no stability or justice in a society that is not founded upon revealed law. But this does not at all mean that 'religious' authorities must dominate the state, or that the free exercise of different faiths is unthinkable. First, there are no simply religious authorities in the Western and Christian sense: there is a community (a political community, naturally, since that is how communities organise themselves) of those who have willingly submitted to revealed law. Second, only free submission to God's law is a proper foundation for the 'House of Islam': it may be necessary to combat the unbeliever as a political assailant, but this is not to deny the liberty in principle of any human being to be subject to God or not. Even the issue of voluntary abandonment of Islam is a subject that needs to be looked at with nuance; this is by definition a political offence, yet it is not wholly clear in Muslim jurisprudence that it merits an extreme political penalty. But the Muslim might emerge from the discussion conscious of a question about why and how the Christian might see this as a denial of human liberty; and the upshot could be a deeper recognition of the logic of free submission, and the unavoidably paradoxical nature of a political community governed by law which also assumes that loyalty and obedience to this community cannot be secured by external sanctions that seek to constrain the will by threat. And so the Muslim, challenged about a disbelief by the Christian, is taken back to the most fundamental defining question of Islam, the character of obedience.

Is what the Muslim denies what the Christian affirms? It seems less obvious than at the beginning of the argument. The Muslim is not a theocrat as the Christian West might understand the term: the denial of a 'secular' space is not a claim to impose religious authority over some other kind, but the acknowledgment that only one basis exists for coherent political life of any kind. The Christian may in fact agree; but will argue that in the realities of a historical existence where levels of submission to God are varied (to say the least), there is bound to be a tension between the community which lives professedly by God's law and the turbulent and unstable succession of social orders which arise in turn around it. Dealing with the Muslim's refusal of belief in any 'church-like' body, existing as a distinct entity within civil society may clarify both the Muslim's view of freedom and obedience and the Christian's eschatological reserve about any historical political order. And as in my first instance, a language for the conversation appears as this clarification advances, a language about God's will to be known and the necessity of such knowledge for a social life free from incoherent rivalry and struggle and injustice.

So to the disbelief of the Jew. Both Christian and Muslim apparently hold that God is at liberty to revise the divine promises; and in such a God the Jew cannot believe. A God who changed his mind would be precisely a God whose freedom would be subject to limit and negotiation, a God whose word once spoken could be rescinded. Hebrew scripture explicitly rules out such a thought, and Jewish philosophy understandably regards the mutability of divine election as diminishing God. Is this a fair perception of the God of Muslims and Christians? The Christian, of course, has the entire argument of Christian Scripture to appeal to, especially the complex arguments of Paul in Romans. The promise is made to Abraham's children, but God has extended the definition to include those who become Abraham's children by imitating his faith not simply by lineage; thus those who enter the Christian Church are honorary or adoptive Jews. God is faithful - a point insisted upon by Paul against those who would indeed argue that God's mind has changed and the Jewish people are rejected. The Muslim will go back to the story of Abraham in the Qur'an, accepting that there is a history of some sort of covenant (though with Ishmael as well as Isaac); but what constitutes covenantal obligation on the human side is that obedience which is now given final form in Muslim revelation, of which all earlier prophecy and theophany is a foreshadowing.

Neither Christian nor Muslim believes in a mutable God; but both will be properly challenged by the Jew to look at the coherence of their own stories, especially in the light of the persistence of Jewish religion and nationality. The question about God becomes intimately associated with a question about power. If the Christian and the Muslim are incorporating Jewish history into their narrative, does the Jew have the right to speak for himself or herself, and to be heard? Jewish disbelief in a changeable God is linked (in a way that does not apply to Christians and Muslims) with Jewish self-belief, the confidence of the Jewish people that they are immutably a people. If Christian and Muslim theologies, even when they confess an immutable God, presuppose the mutability of Jewish identity or legitimacy, they claim a very specific kind of power, a power to declare someone else's history over. Jewish disbelief challenges at the deepest level the way such claims may and do emerge in the histories of the younger faiths. Can Christianity and Islam sustain themselves against the accusation of promoting a theological imperialism which has, from a Jewish point of view, nakedly and often murderously political implications?

And once again, there are answers that may emerge. Christians (Christians other than the extreme dispensationalist Christian Zionists anyway) will often find difficulty in offering a theologically positive valuation of the continuing identity of the Jewish people, but may still believe that it is necessary to work at this, if only in terms of the people of Israel as the radical sign of the Church's incompleteness and the priority of the covenant people into whom non-Jews are now believed to be incorporated. The Muslim – paradoxically, more than the Christian in some respects – has a powerful sense of a shared prophetic history, but is unlikely to compromise over the supersession of Torah by Qur'an. Yet the Muslim will also understand the inseparability of law and people in ways that a Christian might find harder, since it bears a certain similarity to the Muslim denial of neutral secularity and of a separated religious society; hence the often exemplary record of Muslim toleration of Jews as a nation within the nation.

Many other comparable exercises could be carried out in respect of the impact of 'disbeliefs' upon dialogue between the Abrahamic faiths, but I hope that the point of the discussion may be emerging. I am proposing that there is some analogy between the significance of particular unbeliefs upon the self-understanding of religious discourse in general and the significance of the 'unbeliefs' of particular religious discourses for each other. There is no such thing as a global system of 'atheism': there are denials of specific doctrines on varying grounds, and the examination of where the points of stress are in the exposition of these doctrines very importantly allows us to test the resources of what we say as believers – and, ideally, to emerge with a more robust sense of those resources. But equally, conversation between faith traditions can sometimes give the impression that part of the essence of one religious idiom is its disbelief in the God supposedly revealed in another; so that binary oppositions dominate our attitudes. Treat these disbeliefs, I am suggesting, as we might imaginatively and sensitively treat atheism; that is, try to see why what is denied is denied, and whether that denial is directed against what another tradition in fact claims. And in the light of that, try to discover what your own tradition commits you to and how it answers legitimate criticism from outside – criticism which often (as in the case of the mutability of God) could be raised intelligibly within the native tradition. What emerges is frequently a conceptual and imaginative world in which at least some of the positive concerns of diverse traditions are seen to be held in common.

But this is not at all to condemn interfaith dialogue to the sterile and abstract task so often envisaged for it, of identifying a common core of beliefs. The exercise I have been describing is not about finding a common core at all; it is about finding the appropriate language in which difference can be talked about rather than used as an excuse for violent separation. Just as in an encounter with atheists, it is sometimes possible to grasp the positive sense or expectation that leads an atheist to reject what he or she imagines is God, so that the conversation does not simply end in the positing of affirmation and denial, so here. We should certainly not be looking for a common core of belief between believer and atheist, but for a language in which to acknowledge and understand the difference. And in interfaith conversation, we continue to make the claims we make out of conviction of the truth, but seek to break through the assumption that everything can be reduced to whether people say yes or no to a set of simple propositions. Only in the wake of such a move can true dialogue proceed; it does not in fact happen when the 'common core' model is at work, because the hidden assumption is that what is common is bound to be what matters – in which case, difference is not really interesting, intellectually or spiritually. Nor does it happen when the relation between the faiths is seen as one between a set of correct answers and several sets of incorrect ones.

This lecture began as a reflection on the slippery character of the word 'atheism', an the need to resist the elevation of atheism to the level of a system – a danger which has been very publicly around in educational debates in the UK in recent months. But the more we recognise the variegated sense of atheism, the more important it may appear to approach the denials made by atheists as a way into understanding more thoroughly what doctrines and commitments do and don't entail. And on the basis of this, we have moved on to look at the denials, the unbeliefs of certain religious traditions, denials often assumed (both within and beyond a tradition) to be necessarily connected with the refusal of the truth of another faith, seen as a system constructed on the basis of what one tradition or another denies. To allow atheistic schemes to be examined as more than just the elaboration of a single denial, and to allow religious faiths to be examined as more than a map of mutual exclusions and incompatibilities are closely connected. Hence the suggestion, not after all so paradoxical, that we can learn better how to understand other religious believers if we learn better how to understand unbelievers. If both enterprises lead us back to an enhanced appreciation of the resource and complexity which our own faith both offers us and demands of us, so that we are more and not less confident in dialogue, we shall not have wasted our time.

© Rowan Williams 2004