The 'mistake' or 'by-product' idea, which I am espousing, works like this. Natural selection, in ancestral times when we lived in small and stable bands like baboons, programmed into our brains altruistic urges, alongside sexual urges, hunger urges, xenophobic urges and so on. An intelligent couple can read their Darwin and know that the ultimate reason for their sexual urges is procreation. They know that the woman cannot conceive because she is on the pill. Yet they find that their sexual desire is in no way diminished by the knowledge. Sexual desire is sexual desire and its force, in an individual's psychology, is independent of the ultimate Darwinian pressure that drove it. It is a strong urge which exists independently of its ultimate rationale.

I am suggesting that the same is true of the urge to kindness — to altruism, to generosity, to empathy, to pity. In ancestral times, we had the opportunity to be altruistic only towards close kin and potential reciprocators. Nowadays that restriction is no longer there, but the rule of thumb persists. Why would it not? It is just like sexual desire. We can no more help ourselves feeling pity when we see a weeping unfortunate (who is unrelated and unable to reciprocate) than we can help ourselves feeling lust for a member of the opposite sex (who may be infertile or otherwise unable to reproduce). Both are misfirings, Darwinian mistakes: blessed, precious mistakes.

Do not, for one moment, think of such Darwinizing as demeaning or reductive of the noble emotions of compassion and generosity. Nor of sexual desire. Sexual desire, when channelled through the conduits of linguistic culture, emerges as great poetry and drama: John Donne's love poems, say, or Romeo and Juliet. And of course the same thing happens with the misfired redirection of kin- and reciprocation-based compassion. Mercy to a debtor is, when seen out of context, as un-Darwinian as adopting someone else's child:

Sexual lust is the driving force behind a large proportion of human ambition and struggle, and much of it constitutes a misfiring. There is no reason why the same should not be true of the lust to be generous and compassionate, if this is the misfired consequence of ancestral village life. The best way for natural selection to build in both kinds of lust in ancestral times was to install rules of thumb in the brain. Those rules still influence us today, even where circumstances make them inappropriate to their original functions.

Such rules of thumb influence us still, not in a Calvinistically deterministic way but filtered through the civilizing influences of literature and custom, law and tradition — and, of course, religion. Just as the primitive brain rule of sexual lust passes through the filter of civilization to emerge in the love scenes of Romeo and Juliet, so primitive brain rules of us-versus-them vendetta emerge in the form of the running battles between Capulets and Montagues; while primitive brain rules of altruism and empathy end up in the misfiring that cheers us in the chastened reconciliation of Shakespeare's final scene.

If our moral sense, like our sexual desire, is indeed rooted deep in our Darwinian past, predating religion, we should expect that research on the human mind would reveal some moral universals, crossing geographical and cultural barriers, and also, crucially, religious barriers. The Harvard biologist Marc Hauser, in his book Moral Minds: How Nature Designed our Universal Sense of Right and Wrong, has enlarged upon a fruitful line of thought experiments originally suggested by moral philosophers. Hauser's study will serve the additional purpose of introducing the way moral philosophers think. A hypothetical moral dilemma is posed, and the difficulty we experience in answering it tells us something about our sense of right and wrong. Where Hauser goes beyond the philosophers is that he actually does statistical surveys and psychological experiments, using questionnaires on the Internet, for example, to investigate the moral sense of real people. From the present point of view, the interesting thing is that most people come to the same decisions when faced with these dilemmas, and their agreement over the decisions themselves is stronger than their ability to articulate their reasons. This is what we should expect if we have a moral sense which is built into our brains, like our sexual instinct or our fear of heights or, as Hauser himself prefers to say, like our capacity for language (the details vary from culture to culture, but the underlying deep structure of grammar is universal). As we shall see, the way people respond to these moral tests, and their inability to articulate their reasons, seems largely independent of their religious beliefs or lack of them. The message of Hauser's book, to anticipate it in his own words, is this: 'Driving our moral judgments is a universal moral grammar, a faculty of the mind that evolved over millions of years to include a set of principles for building a range of possible moral systems. As with language, the principles that make up our moral grammar fly beneath the radar of our awareness.'

Typical of Hauser's moral dilemmas are variations on the theme of a runaway truck or 'trolley' on a railway line which threatens to kill a number of people. The simplest story imagines a person, Denise, standing by a set of points and in a position to divert the trolley onto a siding, thereby saving the lives of five people trapped on the main line ahead. Unfortunately there is a man trapped on the siding. But since he is only one, outnumbered by the five people trapped on the main track, most people agree that it is morally permissible, if not obligatory, for Denise to throw the switch and save the five by killing the one. We ignore hypothetical possibilities such as that the one man on the siding might be Beethoven, or a close friend.

Elaborations of the thought experiment present a series of increasingly teasing moral conundrums. What if the trolley can be stopped by dropping a large weight in its path from a bridge overhead? That's easy: obviously we must drop the weight. But what if the only large weight available is a very fat man sitting on the bridge, admiring the sunset? Almost everybody agrees that it is immoral to push the fat man off the bridge, even though, from one point of view, the dilemma might seem parallel to Denise's, where throwing the switch kills one to save five. Most of us have a strong intuition that there is a crucial difference between the two cases, though we may not be able to articulate what it is.

Pushing the fat man off the bridge is reminiscent of another dilemma considered by Hauser. Five patients in a hospital are dying, each with a different organ failing. Each would be saved if a donor could be found for their particular faulty organ, but none is available. Then the surgeon notices that there is a healthy man in the waiting-room, all five of whose organs are in good working order and suitable for transplanting. In this case, almost nobody can be found who is prepared to say that the moral act is to kill the one to save the five.

As with the fat man on the bridge, the intuition that most of us share is that an innocent bystander should not suddenly be dragged into a bad situation and used for the sake of others without his consent. Immanuel Kant famously articulated the principle that a rational being should never be used as merely an unconsenting means to an end, even the end of benefiting others. This seems to provide the crucial difference between the case of the fat man on the bridge (or the man in the hospital waiting-room) and the man on Denise's siding. The fat man on the bridge is being positively used as the means to stop the runaway trolley. This clearly violates the Kantian principle. The person on the siding is not being used to save the lives of the five people on the line. It is the siding that is being used, and he just has the bad luck to be standing on it. But, when you put the distinction like that, why does it satisfy us? For Kant, it was a moral absolute. For Hauser it is built into us by our evolution.