But my preoccupation in this chapter is with Darwinian ultimate explanations. If neuroscientists find a 'god centre' in the brain, Darwinian scientists like me will still want to understand the natural selection pressure that favoured it. Why did those of our ancestors who had a genetic tendency to grow a god centre survive to have more grandchildren than rivals who didn't? The Darwinian ultimate question is not a better question, not a more profound question, not a more scientific question than the neurological proximate question. But it is the one I am talking about here.

Nor are Darwinians satisfied by political explanations, such as 'Religion is a tool used by the ruling class to subjugate the underclass.' It is surely true that black slaves in America were consoled by promises of another life, which blunted their dissatisfaction with this one and thereby benefited their owners. The question of whether religions are deliberately designed by cynical priests or rulers is an interesting one, to which historians should attend. But it is not, in itself, a Darwinian question. The Darwinian still wants to know why people are vulnerable to the charms of religion and therefore open to exploitation by priests, politicians and kings.

A cynical manipulator might use sexual lust as a tool of political power, but we still need the Darwinian explanation of why it works. In the case of sexual lust, the answer is easy: our brains are set up to enjoy sex because sex, in the natural state, makes babies. Or a political manipulator might use torture to achieve his ends. Once again, the Darwinian must supply the explanation for why torture is effective; why we will do almost anything to avoid intense pain. Again it seems obvious to the point of banality, but the Darwinian still needs to spell it out: natural selection has set up the perception of pain as a token of life-threatening bodily damage, and programmed us to avoid it. Those rare individuals who cannot feel pain, or don't care about it, usually die young of injuries which the rest of us would have taken steps to avoid. Whether it is cynically exploited, or whether it just manifests itself spontaneously, what ultimately explains the lust for gods?

Some alleged ultimate explanations turn out to be — or avowedly are — 'group-selection' theories. Group selection is the controversial idea that Darwinian selection chooses among species or other groups of individuals. The Cambridge archaeologist Colin Renfrew suggests that Christianity survived by a form of group selection because it fostered the idea of in-group loyalty and in-group brotherly love, and this helped religious groups to survive at the expense of less religious groups. The American group-selection apostle D. S. Wilson independently developed a similar suggestion at more length, in Darwin's Cathedral.

Here's an invented example, to show what a group-selection theory of religion might look like. A tribe with a stirringly belligerent 'god of battles' wins wars against rival tribes whose gods urge peace and harmony, or tribes with no gods at all. Warriors who unshakeably believe that a martyr's death will send them straight to paradise fight bravely, and willingly give up their lives. So tribes with this kind of religion are more likely to survive in inter-tribal warfare, steal the conquered tribe's livestock and seize their women as concubines. Such successful tribes prolifically spawn daughter tribes that go off and propagate more daughter tribes, all worshipping the same tribal god. The idea of a group spawning daughter groups, like a beehive throwing off swarms, is not implausible, by the way. The anthropologist Napoleon Chagnon mapped just such fissioning of villages in his celebrated study of the 'Fierce People', the Yanomamo of the South American jungle.⁷⁷

Chagnon is not a supporter of group selection, and nor am I. There are formidable objections to it. A partisan in the controversy, I must beware of riding off on my pet steed Tangent, far from the main track of this book. Some biologists betray a confusion between true group selection, as in my hypothetical example of the god of battles, and something else which they call group selection but which turns out on closer inspection to be either kin selection or reciprocal altruism (see Chapter 6).

Those of us who belittle group selection admit that in principle it can happen. The question is whether it amounts to a significant force in evolution. When it is pitted against selection at lower levels — as when group selection is advanced as an explanation for individual self-sacrifice — lower-level selection is likely to be stronger. In our hypothetical tribe, imagine a single self-interested warrior in an army dominated by aspiring martyrs eager to die for the tribe and earn a heavenly reward. He will be only slightly less likely to end up on the winning side as a result of hanging back in the battle to save his own skin. The martyrdom of his comrades will benefit him more than it benefits each one of them on average, because they will be dead. He is more likely to reproduce than they are, and his genes for refusing to be martyred are more likely to be reproduced into the next generation. Hence tendencies towards martyrdom will decline in future generations.

This is a simplified toy example, but it illustrates a perennial problem with group selection. Group-selection theories of individual self-sacrifice are always vulnerable to subversion from within. Individual deaths and reproductions occur on a faster timescale and with greater frequency than group extinctions and fissionings. Mathematical models can be crafted to come up with special conditions under which group selection might be evolutionarily powerful. These special conditions are usually unrealistic in nature, but it can be argued that religions in human tribal groupings foster just such otherwise unrealistic special conditions. This is an interesting line of theory, but I shall not pursue it here except to concede that Darwin himself, though he was normally a staunch advocate of selection at the level of the individual organism, came as close as he ever came to group selectionism in his discussion of human tribes:

To satisfy any biological specialists who might be reading this, I should add that Darwin's idea was not strictly group selection, in the true sense of successful groups spawning daughter groups whose frequency might be counted in a metapopulation of groups. Rather, Darwin visualized tribes with altruistically co-operative members spreading and becoming more numerous in terms of numbers of individuals. Darwin's model is more like the spread of the grey squirrel in Britain at the expense of the red: ecological replacement, not true group selection.

In any case, I want now to set aside group selection and turn to my own view of the Darwinian survival value of religion. I am one of an increasing number of biologists who see religion as a by-product of something else. More generally, I believe that we who speculate about Darwinian survival value need to 'think by-product'. When we ask about the survival value of anything, we may be asking the wrong question. We need to rewrite the question in a more helpful way. Perhaps the feature we are interested in (religion in this case) doesn't have a direct survival value of its own, but is a by-product of something else that does. I find it helpful to introduce the byproduct idea with an analogy from my own field of animal behaviour.