Modern social science is the creation of theorists who explored the role of religion in many areas of social life, giving it an explanatory role; in doing so, they discussed a very wide range of acts, beliefs, feeling and experiences. Sociology and anthropology proposed from their very beginnings a scientific study of religion. I want to talk about some attempts to understand religion in this ‘scientific’ way. And, at the end, I am going to take the rise of a social science of religion as itself something worth trying to understand.

We'll start with a classical canon of anthropology and sociology of religion, in the works of E. B. Tylor (1832–1917), Max Weber (1864–1920) and Émile Durkheim (1858–1917), who all died towards the end of the second decade of the twentieth century. Tylor's account of religion, which is the earliest, can be found in his great work Primitive Culture, published in 1871. This is the book that entitles him to be regarded as the first cultural anthropologist; and it made his reputation, gaining him the academic stature that led to his eventual appointment, a quarter of a century later, as the first Professor of anthropology at Oxford University. Then, I shall discuss Durkheim and Weber, in each case focusing on a major work: Durkheim's Les formes élémentaires de la vie religieuse (1912) and Weber's discussion of religion in Economy and Society, the magnum opus published in 1922, two years after his death. Tylor was raised as a Quaker, Weber's family was Calvinist, and Durkheim was the son and grandson of rabbis. Between them these three – an English Quaker, a German Calvinist, and a French Jew – laid the foundations of the modern social sciences of religion. At the end I'll discuss a representative piece of recent work in evolutionary psychology that aims to explain how various features of religion developed historically. We'll see that it addresses the very elements that were central in the work of these three founding fathers.

Tylor's great contribution to the social science of religion is the theory of animism he proposed in Primitive Culture, when he offered his ‘minimum definition of religion’, as ‘the belief in Spiritual Beings’.Footnote ¹ This definition seems clearer, I think, than it actually is. For looking through the hundreds of pages of Tylor's compendium of examples, his ‘Spiritual Beings’ are a motley crew. Many Confucians believe in ancestral spirits. The Romans had their manes, spirits of the dead; and their household gods, the lares and penates. But Romans also believed in nature-spirits, in nymphs and satyrs, as well as in Jupiter and the other Olympians, and the hundreds of gods named in the indigitamenta, the priestly lists of gods. The Maori, on the other side of the planet, have Maui, who played a leading role in Tylor's book long before he was discovered by Disney to star alongside Moana; but they also have a sky god and an earth goddess and gods of the sun, the sea and the stars, of peace and war, beings with evocative names: Ranginui and Tangaroa. They have nature spirits as well, identified with trees, lakes or mountains.

Many Jews have believed not just in Yahweh but also in golems, creatures made from animated mud; Christians traditionally believe in demons; orthodox Moslems believe in Allah, of course, but also in jinn, of whom only one is well known outside the Moslem world … as the genie of Aladdin's lamp. But in all the Abrahamic traditions you will find people who worry about ghosts and have faith in angels. If, as Tylor clearly thought, all these are ways of believing in spirits, and spirits are so diverse, don't we need to say something about what, conceptually speaking, they all have in common?

Tylor's central idea encoded in the definition of religion as belief in spiritual beings, is that religion involves extending belief in the human soul, which accounts for the animation of a living body, in two conceptual directions. First, we develop the idea that what animates us can live outside our body and, finally, outlast it: this is the human soul. And second, we suppose there are other entities in the world that are soul-like. These two extensions together produce, in Tylor's scheme, the idea of a ‘spiritual being’. Once one has belief in spirits, then, with powers over aspects of the world, worship, one aspect of the practice of religion, is just a natural human response to the spirits we are interacting with. Just as we placate people who have power over the things we care about, so we propitiate the spirits. The belief is primary. Once we have belief, practice can be seen as the reasonable reflection of it. In extending souls into the wider world, we generate naturally a system of quasi-social relations beyond the world of living human beings.

Under the influence of Descartes, you may be tempted to suppose that spirits must be immaterial. But Tylor gives a cosmopolitan compendium of cases of traditions in which the soul is represented as a material substance, including, say the Tongans who ‘imagined the human soul to be the finer or more aeriform part of the body, … comparable to the perfume and essence of a flower as related to the more solid vegetable fibre’.Footnote ² The extension of souls beyond humanity goes far beyond their ascription to animals and plants, which like human beings can pass from life to death, and so may need a life force to explain that difference. For many things, including artefacts like drums or spears, can have spirits too; and some of the principle spirits, as with the Maori, inhabit lakes and mountains, the Sun and the stars. Tylor gets closer to an account that covers all the thousands of cases he examines in Primitive Culture, when he says, speaking more abstractly, ‘Spirits are simply personified causes’.Footnote ³

Now Tylor presents this belief in spirits as a perfectly natural response to early human experience of the natural world. He doesn't think our ancestors are to be condemned for it. We could say, in a modern philosophical idiom, that Tylor's primitive men and women adopted what Daniel Dennett, the American philosopher, has called the intentional strategy, ‘treating’, as Dennett put it, ‘the object whose behaviour you want to predict as a rational agent with beliefs and desires and other mental states exhibiting what Brentano and others call intentionality’. It's just that they did so over a much wider range than many of us now think is warranted and they based their appeal to these quasi-psychological states in the belief that they are the capacities of a sort of spiritual (which does not mean immaterial) substance. We should recall, in deciding what we think of our ancestors here, that we still apply the intentional strategy to things – such as our pets and our computers – that are not human.

So, using the vocabulary of the contemporary philosophy of mind, Tylor's claim is that the core of religion is the application of the intentional strategy to a substance widely instantiated in the non-human world. But human beings have explanatory schemas available other than the intentional stance. We understand certain events – the breaking of a tree limb by the wind – as a causal process, explicable without appeal to the will or the beliefs or the emotions of the tree or the wind. Everybody understands the push-and-pull of efficient causality. People also have the idea that some things have functions: that hearts are for pumping blood, knives for cutting.

In the contemporary world, with the rise of a scientific approach, we can give many more detailed causal explanations, allowing for much more successful prediction and control than was possible in earlier societies. Darwinism allows us to explain how functions arise without appeal either to agency or to design. And when things happen in the world around us, we moderns often turn naturally to an explanation neither in terms of agency – not that is, by way of the intentional stance – nor in terms of design, but in terms of natural causality. Why did our ancestors so often do otherwise? I will return later to this question why we humans have applied the intentional strategy to so many things other than ourselves.

Many of us, whether raised within one of the world's many religious traditions or outside them all, are likely to feel that Tylor makes too little of the role of intense emotion in religious life. Only at the end of the last of the chapters on animism, does Tylor observe that he has kept the ‘intellectual rather than the emotional side of religion … in view’. Yet even among ‘the rudest savages’ he concedes, ‘religious life is associated with intense emotion, with awful reverence, with agonizing terror, with rapt ecstasy…’ Recognizing this, Tylor admits, some will complain that he has ‘written soullessly of the soul’. Yes, he responds, there is emotion there, but the feelings are the natural responses to the spirits. So, it's the belief in spirits that is the heart of the matter. That Tylor doesn't talk much of religious emotion does not mean that his theory cannot encompass it, then. For belief in creatures with minds something like ours provides a quite reasonable basis for an emotional response.

There is another great absence in Tylor's treatment of primitive religion, one that you might think goes with the absence of feeling: Primitive Culture seems remarkably un-attuned to the role of religion in shaping the ethical life of its adherents, by which I mean its role in the enforcement of norms, in their creation and in their content. Tylor's defence here is that he is focusing on primitive religion, on the origins of religiosity; and he believes that ‘savage animism is almost devoid of that ethical element which to the educated modern mind is the very mainspring of practical religion’.Footnote ⁴ (Whether he was right about this is a question to which I will also return.)

In the works of the two great sociologists of religion I want to consider next, both ritual and community, play a much more central role: and the social meaning of religion extends far beyond the consequences of belief in spiritual beings into emotion and ethical life. But let's concede that Tylor demonstrated, with an astonishing range and depth of ethnographic materials from the human past and around the world, that a great deal of what we call religion can indeed be understood by seeing it as the projection of agency into the cosmos. In developing the theory of animism – the hypothesis that religion is, at heart, belief in spirits – Sir Edward Tylor delineated one paradigm for the scientific study of religion.

Max Weber's most famous work, The Protestant Ethic and the Spirit of Capitalism, is evidently about one sort of religion. But in that book, he didn't say what made the Calvinist ethos religious. All that mattered was that we could apply the label ‘Protestant’ (or ‘Calvinist’) to a group of people and associate the label with a commitment to certain values and the behaviour that flows from them. Weber went on to explore religious traditions from around the world, publishing The Religion of China and The Religion of India in 1916 and Ancient Judaism over the next two years. He was engaged in trying to complete the section on religion in his summa, Economy and Society, when he died in 1920, as one of tens of millions of victims of the Spanish flu pandemic. That work was published in 1922; and the section on religion is known today in English as The Sociology of Religion. And here Weber had to face directly the question what made all these earlier works studies of the same thing.

But Weber wanted the true nature of religion to emerge in the course of this study, so he didn't start with a definition. ‘The “essence” of religion is absolutely not our concern’, he wrote; our project, rather, is the investigation of ‘the conditions and effects of a particular type of social behaviour’.Footnote ⁵ For Tylor, individuals could seek the assistance of spiritual beings individually; what was social was only the source of their beliefs. Weber's interest is primarily in collective behaviour that is organized through communities and their shared institutions and concepts. Tylor's religion could be practiced alone. Weber's involves doing things together.

Weber aimed to sketch, like Tylor, a framework within which all the religions of the world could be accommodated: and it is remarkably reminiscent, to begin with, of Tylor's. He starts, for example, like Tylor, with the most basic historical forms of religion, and he asserts, like Tylor, that they are ‘oriented to this world’.Footnote ⁶ Religion begins, for both of them, with getting things done in the here and now, not in a search for some hoped-for life beyond.

Weber wants us to start with understanding the ideas that guide religious behaviour. So, is there, he wonders, a concept that identifies something in the mental lives of religious adherents around the globe that underlies their behaviour and that will provide a key to understanding this ‘particular type of social behaviour’? Weber's answer is that there is, and he calls it ‘charisma’.

Charisma began as a technical term in Jewish and Christian thought, referring to the gifts of God's χάρις, his grace, as in St. Paul's first letter to the Corinthians, where they are listed: wisdom, knowledge, faith, healing, miracles, prophecy, the discerning of spirits, and the ‘gift of tongues’.Footnote ⁷ Weber took this technical theological idea and applied it to special powers that could be found in certain particular objects and people.

When he first introduces the idea in The Sociology of Religion, Weber says simply that those who believe in magic acknowledge the ‘lesser everydayness’, of things like a stone that ‘is used as a fetish’, or of those people who can ‘achieve ecstatic states and through them produce meteorological, therapeutic, divinatory or telepathic effects…’Footnote ⁸ Powers like these, he says, have been given such special names as ‘mana’ in Polynesia, ‘orenda’ among the Iroquois, and the Iranian ‘maga’, (from which: magic), and Weber proposes to use the word ‘charisma’ for all such powers.Footnote ⁹

The special powers of charisma can be found in icons or statues of saints in contemporary folk religion in Europe, which operate in what Weber calls a ‘naturalistic, so-called pre-animistic way’. Having read Tylor, we know that he means by pre-animistic that these powers are supposed to work without imagining the presence within the charismatic object – the icon or the statue – of spiritual beings. Even though the process here is ‘pre-animistic’, Weber goes on immediately to say that in these cases, too, people have imagined ‘something “behind” the charismatically-endowed natural objects, artefacts, animals and human beings, concealed within them and somehow determining their behaviour: the belief in spirits’.Footnote ¹⁰ If we stopped here, we might think that Weber's understanding of religion returns to Tylor's. But Weber gives an account of spirits that is clear enough that we can see that it is different. In their basic form, his spirits are not like human souls. They are ‘something undefined: material yet invisible, impersonal and yet imagined as equipped with a kind of will, giving the concrete object its specific efficacy…’Footnote ¹¹ And Weber does not see the basic intellectual strategy of belief in spirits as the projection of the human soul and with it an anthropomorphic agency into the world. This is not the full intentional strategy. So, Weber starts in a different place – with charisma and with spirits less like us – and, as we shall see now, his emphasis on social behaviour takes him swiftly in an entirely different direction.

Weber thinks belief in charisma develops first in societies where magical powers are found in people with special qualifications, people he calls ‘magicians’. These professionals are permanently possessed of charisma: they are always capable of achieving ecstasy, which is the distinctive human state that both represents and mediates charisma. Ecstasy involves something different from what Weber calls the ‘rational magic’ with which he began – the pre-animist, naturalistic manipulation of objects with special powers. And it is available to laymen only within the social form of the orgy, which is the ‘primitive form or religious community’.Footnote ¹² The model Weber has in mind here is the Dionysian mystery cults, of Mycenae and later of Athens, in which initiates engaged in collective, secret practices – orgies – which led to ecstatic experience … to reverie, rapture, hallucination, possession, and every kind of enchantment.

Belief in the soul, then, as something distinct from the body, originates in this orgiastic experience, as the idea of something that leaves the body in ecstasy, but also in dreams, fainting and death; something that can also be found ‘behind, near or in’ natural objects. Gradually, over time, spirits come to be seen as normally invisible beings, governed by laws of their own, not living in or possessing objects but symbolized by them. So, the very idea of the human spirit, the soul, grows out of a religious practice – the orgy – rather than ante-dating and explaining it, as in Tylor. Weber's religion, unlike Tylor's, begins in a world of intense emotional experiences.

Though these spirits inhabit people and things, they are not yet fully personal. They need not have names; indeed, they may not endure beyond their operation in a single event. So, unlike Tylor's spirits, Weber's souls begin as something more like impersonal forces – albeit with wills – than people. Eventually some souls do become enduring personalities, bearing the names of heroes, say, and then we have gods. But this is not the really significant issue. What matters is that religion gives a role not just to ‘things and events that exist or actually happen’, but also to things that have a meaning beyond themselves. ‘Magic becomes’, Weber wrote, ‘… a symbolic activity’.Footnote ¹³

Now, Tylor too, had used the word ‘symbol’ a few dozen times in Primitive Culture: like any reasonable observer he noticed that religious acts are often, in some sense, symbolic. But it did not occur to him to see symbolism as central to religion, as Weber clearly does, or to insist on the most obvious reason why this must be so, even on an account that centres on the animist belief in spirits. For if these spirits ‘behind’ the world are, indeed, like persons, we will need to communicate with them; and that requires working ‘through means that speak to a spirit or a soul, something, therefore, that has meaning: through symbols’.Footnote ¹⁴

For Weber, a symbolic act has two dimensions: it has a form, which is often highly stylized or stereotyped, and it has a meaning. That meaning belongs to the world of the practitioner, so that what the acts mean they mean to him or to her. While the term ‘charisma’ belongs to Weber's theoretical vocabulary, this idea that some people and object have a special power, whether they call it mana or orenda, belongs to the religious adherents’ conceptual world. Weber thinks religion starts with the conviction that some people and things have extraordinary powers like these, derived from the presence of a spirit, which is something endowed with a kind of will.

Over time, however, as the spirits develop personality, they become gods, associated with various social groups for which they have a special concern and over which they exercise a special authority. But the important distinction is not Tylor's distinction between gods, which are aristocratic, and other lesser spirits, but rather in the ways we relate to them. It is, then, the social forms through which the gods are addressed that distinguish them from the spirits.

‘Prayer, sacrifice, and worship’, mediated by priests, are the marks of religion. Magic, managed by magicians, involves ‘compulsion’ through spells and the like. The spirits compelled are demons; the spirits worshipped are gods.Footnote ¹⁵

This differentiation between magic and religions, then, is connected with the distinction that Weber makes between priests, who develop later, and magicians, the earliest of the charismatic professionals, with whom he began. And the core difference is that magicians work alone, while priests are part of a profession operating a ‘continuously operating cult’.Footnote ¹⁶ Weber believed, famously, that human societies displayed a tendency towards what he called ‘rationalization’. And priests are important for rationalizing, through their shared activity in organized institutions – temples and churches – both theology and religious ethics.

Tylor had asserted that early religion was not much concerned with ethics. Weber agreed but argued that the gods became increasingly involved in making ethical demands in parallel with the processes that led to the increasing rationalization of norms, as social organization became more complex. With the rise of peaceful, law-governed states, the growth of a scientific understanding of the natural world, the increased regimentation of human interactions through conventional rules, and a growing dependence in economic activity of reliance on the word of others, people became more dependent on expectations based on personal attachment to a ‘cosmos of obligations’, which made their behaviour more predictable.Footnote ¹⁷ Now religions not only supported the moral norms necessary for the coordination of social and economic life, they generated their own forms of obligation: ritual practices whose point, for the actor, is that they are demanded by a god. Religion was not just the pursuit of ends given outside it, as in Tylor. Now it had become a source itself of ends.

Religion, then, is what happens when charisma becomes the institutional prerogative of a professional class. There is no need to worry about distinctions between monotheism and polytheism; indeed, in the most rarefied forms of Buddhism, as Weber understands them, there is no need for gods or for the soul, there is just the management of charisma. Without the concept of charisma, then, Weber's formal understanding of religion would be just the very belief in spirits that defined Tylor's animism. And Weber's treatment of Buddhism shows that he believed that you could have religion without non-human spirits, because even as the Buddhists abandon the Hindu gods, the priesthood and the management of charisma remain.

So, charisma has a lot of work to do. And yet, if one returns to his introduction of the concept, it is at least as puzzling as Tylor's account of spirits. Charisma, recall, was introduced simply as involving powers outside the everyday. But in many societies the sort of magic Weber discusses – the appeal by the Catholic peasant to the icon of the saint – in fact happens literally every day. What is distinctive about magic is not its frequency but the kind of causality it involves, the appeal to hidden powers associated with something like a will. And though Weber insists that this does not yet need to entail belief in spirits, the idea of a will does involve the projection into the world of an aspect of our human psychology. The difference between Tylor and Weber, then, is something like this: religion for Weber need not involve the full intentional strategy. Indeed, as we saw, Weber believes that the idea of the soul is the product not the source of religious practices; it derives, he says, from the experience of ecstasy in the orgy. And, as we also saw, the distinction between magic and religion for Weber is a distinction in the social functions of magicians and priests. It is the social role of the person with the extraordinary powers.

Tylor, Weber and Durkheim, to whom I turn more briefly now, offer an almost comical confirmation of national stereotypes: Tylor the commonsensical Englishman, Weber, the German, with his opaque profundities, Durkheim, seeking in the proper Gallic fashion to construct a logical system of gleaming Cartesian clarity. In the introduction to Les formes élémentaires de la vie religieuse, the French philosopher-sociologist lays out a series of propositions that aim to introduce a formal definition of religion.

Durkheim thinks religious beliefs, in societies both simple and complex, are in some sense about the division of the world of ideas and of things into two classes, which he calls sacred and profane.Footnote ¹⁸ ‘Beliefs, myths, dogmas, legends are representations or systems of representations that express the nature of sacred things, the virtues and the powers that are attributed to them, their history, their relations with one another and with profane things.’Footnote ¹⁹ Religion contains two main kinds of phenomena, though, of which belief is only one. The other is ritual, defined as behaviour governed by the rules that prescribe how one should relate to sacred things.Footnote ^20, Footnote ²¹

The sacred is to Durkheim what charisma is to Weber: the key to unlocking the door to religion. And, indeed, mana, orenda and magic all figure as examples of the sacred as they did for Weberian charisma. Durkheim is keen to point out that his theory explains why you can count the more intellectual forms of Buddhism as a religion, even though it involves no belief in a soul and no gods, because the Four Sacred Truths are sacred and so, therefore, are the practices that derive from them, which are, therefore, instances of ritual.Footnote ²² The first chapter of Les formes élémentaires offers a well-known definition:

And in the chapter's last sentence he insists that in showing that ‘the concept of religion is inseparable from the idea of a church’, he has intimated ‘that religion has to be something profoundly collective.’Footnote ²³ So religion involves both a division of the world into sacred and profane and an institutional community that is held together through rituals connected with that division.

Our three theorists assume, then, a human psychology and ask how, given that psychology, religious belief and action arise. But after Darwin, we see human psychology as itself something whose existence needs to be explained. We can ask why there are creatures like us – not taking us as we are, but asking how we came to be this way. Two central questions then suggest themselves. First, the question invited by Tylor's animist hypothesis: Why are there creatures like us that adopt the intentional stance over-broadly? Second, a question made natural by Weber and Durkheim: What about religious behaviour is adaptive, what explains its survival?

Let's consider, then, just one representative recent attempt to explain aspects of the evolution of religion, an explanation that displays Tylorean, Weberian and Durkheimian features. It's a story that aims to explain the origins of the very feature of our evolved psychology that Tylor took for granted: our propensity to overextend the intentional strategy.

The vocabularies of anthropology, philosophy and evolutionary psychology are often different. So rather than talking about the intentional strategy, as we philosophers do, or of belief in spirits, as Tylor did, some evolutionary psychologists ascribe to the evolved psychology of most modern human beings what they call an ‘Agent Detection Device’. Most brains are so formed, they think, that we have a module that leads us to respond to a wide range of experiences by attributing agency to something in the world. Very often, of course, that something is a person, an actual agent. And we understand agents by applying to them a folk psychological theory, that supposes that they have beliefs, desires and feelings. Evolutionary psychology calls that Theory of Mind.

It's easy to see why the complex of genes necessary to develop an Agent Detection Device might have evolved. In a social animal, making sense of others is crucial to individual survival. Since we are in fact agents – creatures that conform more or less to the folk Theory of Mind – then being able to apply that theory will help you get others right. But to do that you'll need to be able to identify agents in the first place; identify the things to seek to understand using Theory of Mind. Precisely because we are social creatures, dependent on coordinating our behaviour with others, individual survival will be enhanced by the possession of that device. But this does not explain why it increases the Darwinian fitness of individuals to have what has been called a hyperactive Agency Detection Device.Footnote ²⁴ Why, that is, should our tendency to ascribe agency have evolved to be triggered so often by things that are not, in fact, agents? Why should it behave precisely the way Tylor supposed?

The evolutionary biologist Dominic Johnson has proposed an answer. It begins explicitly, as all the theories we have so far considered did implicitly, with methodological atheism. There is no God. You might think it follows that there could be no individual selective advantage to be gained from believing there is one. You would be wrong, Johnson argues.Footnote ²⁵ In God's absence, the very heavy investment in the activity of the Durkheimian church is clearly going to disadvantage individuals. Prayer takes time away from farming or hunting, taboos limit diet and raise the cost of nutrition; the priests and the temples need support. So far, evolutionarily speaking, so bad. But suppose the advantages outweighed these admittedly significant costs. Johnson has a story about why they might.

Let's start with another example of a habit of the human mind that leads to error. We are far more likely to mistake sticks for snakes than the other way around. Why should that be? Because the costs of mistaking a snake for a stick are almost always way more serious than the costs of the reverse error. No detector, whether produced by evolution or human design will be perfect. It will make two kinds of errors: sometimes it will fail to detect the presence of something it should (a false negative), sometimes it will mistakenly announce the presence of something it shouldn't (a false positive). In general, then, according to what's called Error Management Theory, a well-designed detector should avoid the mistakes that are most costly, at the expense, if necessary, of increasing the rate of the less costly kind of error. In the case of snake-detection, the higher cost mistake is the false negative: mistaking a snake for a stick.

The hypothesis that a hyperactive Agency Detection Device can be adaptive requires that the more expensive error is the false negative: failing to identify an agent when one is around. But why? Johnson's answer starts with a solution to another problem about the evolution of sociability. The development of complex societies clearly requires a great deal of cooperation.Footnote ²⁶ A recent line of literature demonstrates with mathematical models that one way to support this is by establishing an inclination that in fact exists in most human beings, to punish those who breach social norms, even when we are not ourselves their victims. The bottom line is that it is reasonable to defect from a social norm in many contexts, only if you can reduce the costs of detection.

Now among the many significant consequences of the development of Theory of Mind and of language are these: people get better at identifying the motivations of others, including their incentives to defect, and they are able, through language, to communicate the fact of defection to others. The result is that in creatures like us, the risks associated with defection are significantly higher than in social creatures with neither language nor Theory of Mind. Even if you think you are not being watched, you may be; or others may infer your defection from its traces, including its later traces in your own behaviour. You may think you will get away with the lie, for example. But the cognitive task of keeping track of your lies can be high and may eventually show you up … as may the arrival of someone who knows the truth. And so not only is the likelihood of detection high, its costs are high, too, since they extend to the responses not just of those who witness or infer your defection potentially to all those with whom they can communicate. In these circumstances, the cost of wrongly thinking your defections will escape discovery, can be very high, and may be higher than the costs of conformity. You had better notice all the agents you can.

In these circumstances, Johnson argues, there may be individual selective advantage in having a tendency not to defect, in being inclined that is, to do things that are good for the community and genuinely biologically costly, because the expected imposed social costs of detected defection are so high. And belief in invisible spirits would be a mechanism for securing such prosocial behaviour, because, of course, if the spirits are always around and can be invisible, and if, like human beings, they can both punish you and communicate your defections to others, you will always worry that defection risks detection and punishment.

This is an argument for the proposition that Tylorean animism is the consequence of a hyperactive Agency Detection Device that is individually selectively advantageous, because we cannot be sure that we will evade detection, even when we think there is no one around. Not because the spirits are indeed watching, of course, but because human beings equipped with Theory of Mind and with language are way more likely to learn about and punish our defections than is true among our fellow primates.Footnote ²⁷

Like the founding fathers of the social sciences of religion, modern evolutionary psychologists have conceptions of religion. They include belief in spirits (Tylor), the endorsement of prosocial norms (Durkheim), and ritual behaviour (Weber), none of which would have surprised the founders. But in the actual explanations, of which I have offered just a representative flavour, each of these elements needs to be identified separately to do its explanatory work. In religions where one or the other was absent, parts of the explanation could still be appealed to. Just as in Weber's account of the Protestant ethic, the idea of religion doesn't play an explanatory role, so, in Johnson's account of why error management theory suggests we might have a hyperactive Agent Detection Device, what's needed is just belief in invisible agents and prosocial behaviour, not the whole panoply of actual religious life.

For Tylor, Weber and Durkheim, the real question as they looked out from Europe on the wider world was, in effect, ‘What do they have instead of Christianity?’ And if that is right, it won't be surprising that religion is a paradigm of what Wittgenstein taught us to call a ‘family-resemblance’ concept: each religion, like each member of a family, is, no doubt, like every other, in some respects, but there need be no distinctive characteristics they all share.Footnote ²⁸ Remember what Weber said: the ‘“essence” of religion is absolutely not our concern’.Footnote ²⁹ That's a good thing if it doesn't have one.Footnote ³⁰

The Christianity that surrounded our three founding fathers was a set of traditions that had been shaped over millennia. But their sources for thinking about religion outside the great literate traditions were largely materials assembled by Europeans in their global expansion since 1492: and, by and large, they had called anything a religion that was in competition with the Christian ideas, practices and traditions they had left behind. The result is that, many people in our intellectual tradition unwittingly take as the paradigm of religion one particular set of religious sects – those that developed out of medieval Christendom in the coincidence of the age of exploration and the reformation. That's why, for example, belief in God and the soul is taken to be one mark of religion, even though there are atheistic forms of Buddhism that have no place for either God or the soul.

A Christian model of religion is going to look for more than creeds, of course: it will expect churches, priests, prayer, collective worship, moral codes, and yes, charisma and the sacred. It will reject magic, even when it practices it. And it may pay less attention to things that people in other traditions might think obviously important: dietary rules or ancestral cults, which had largely disappeared in European Christianity by the end of the nineteenth century.

When we approach a society that is not our own, seeking to understand its thought and practices in a scientific way, we will always have to begin with an enterprise of translation. Social science begins, as Weber insisted, with understanding the conceptual world of those whose behaviour we seek to explain. The sociology of a Christian society would naturally use Christian concepts to try to make sense of non-Christian ones. But translation isn't just mapping concepts into concepts; it involves triangulating both with reality. To see this, let me offer an example from the history of chemistry.

From a current scientific point of view, late eighteenth-century chemistry looks as though it classified some things – acids and bases, say – by and large correctly, even if a lot of what they said about those things was pretty badly wrong. Also from the point of view of current theory, an acid is, roughly, a proton-donor.Footnote ³¹ Our recognition that the classification of acids and bases was in itself an intellectual achievement is recorded in the fact that we are inclined to say that when, in the eighteenth century, Antoine Lavoisier – who, not having any idea of the proton, could hardly be expected to have understood the notion of a proton-donor – used the word ‘acid’, he was nevertheless talking about what we still call acids.

In explaining why, it seems proper to think that Lavoisier was referring to the things we call proton-donors – even though much of what he believed about acids is not true of proton-donors – philosophers of science have drawn on the ‘causal theory of reference’ in the philosophy of language. The proposal is simple enough: if you want to know what object a word refers to, find the thing in the world that gives the best causal explanation of the central features of uses of that word. If you want to know what the name ‘New York’ refers to, find the object in the world that is at the root of most of the causal chains that lead to remarks containing the expression ‘New York’.

With acids, because we believe that the stuffs ‘out there’ in the world that really accounted for the central features of Lavoisier's ‘acid’-talk really were acids, we conclude that he wasn't simply talking about something else or about nothing at all, even though much of what he believed about acids was wrong.

So, a science formed on this model by Christian believers would proceed by mapping the ontologies of other societies as closely as they could into their own. And that, after all, is what the missionaries who came in the wake of Europe's empires to large parts of Asia, Africa and the Americas, did in the very societies about which the first anthropologists were writing. Where they found, as they often did, evidence of belief in what was called a ‘high God’, associated with the heavens, the sort of monarch of the spiritual world described by Tylor, they took this as evidence that the locals had recognized, however imperfectly, the existence of the God of Abraham and Isaac and Jacob. As for nature spirits or totemic gods, these were either idols – man-made objects mistakenly believed to contain or represent a lesser divinity – or the ‘powers’ for which the New Testament used the words δαιμόνια and πνεύματα, which the King James Bible translates as ‘devils’ or ‘spirits’.

Now I want in closing to suggest that there is a reason the sciences of religion can't be like this, and it has to do not with the nature of religions but with the ethos of science. Tylor, Weber, Durkheim: three nations, three religious traditions, and three languages in conversation. Science aims to be a discourse that transcends local identities; it is deeply cosmopolitan. To offer a local metaphysics as its basis would produce a discourse that was not recognizable as a contribution to the cosmopolitan conversation of the sciences. That is one reason why Soviet ideas of socialist science, which inspired Lysenko, or the even worse Nazi-inspired idea of a German science, belong somewhere in intellectual history but not in the history of the sciences. So, a science of religion cannot appeal to the entities invoked in any particular religion: which explains what I earlier called the methodological atheism of our three founding fathers.

This cosmopolitan ideal, the calling of the scientist, on the one hand, and the concern to understand the ideas of other cultures, on the other, can therefore pull in different directions. Understanding requires us to appeal to our own concepts but not to our own truths. In the explanations, though, truth – the universal shared reality – has to matter, because the scientific story of religion has to work for people of all faiths and none, precisely because it is cosmopolitan. But, as we saw with Lavoisier, we may need to appeal to the truth to figure out which of our concepts provides the best local analogue of the ideas of the stranger. So, the distinction isn't so sharp, precisely because concepts presuppose truths. If there really are no sprits, many conceptual questions about spirits don't make much sense. Are they material? Can they be in two places at once? What would that mean? Once you sense the paradoxes here you may lose your grip on the concept: you may decide it is like the word ‘witchcraft’, which many of us don't believe refers to anything real, rather than Lavoisier's word ‘acid’, which we think does.

As we have seen, not everything we call a religion will have historical Christianity's laser-like focus on ontological truth-claims. Still, so long as there are people making truth-claims about gods and spirits in the name of religion, there will be the possibility of a tension between the very idea of a science of religion and some of the multifarious collections of beliefs, practices and institutions that make up what we now call ‘religions’.