Moral philosophers should go to places like Kosovo if they want to know about right and wrong

A couple of weeks ago in the New Statesman, Richard Dawkins, rationalist extraordinaire, confessed that even he would have qualms about eating human flesh: "The taboo against cannibalism is probably the strongest we have," he wrote. "But even that needs to be looked at. We have to ask ourselves why we are against it."

I wonder if one reason is that we don't really want or need to and that it's not good for us. In his remarkable book, Guns, Germs and Steel, Jared Diamond writes about certain tribes in the interior of New Guinea who have the bad luck to live in an area that is both without large mammals or fish, and whose principal crop happens to be low in protein. As a result the people who live there have the distended stomachs that show they are permanently malnourished. A second result is that they desperately pursue any possible source of protein. Theirs is virtually the only culture in the world that takes the trouble to eat small rodents. It is also probably the reason why cannibalism has been relatively common there.

Remember the joke about how in New York City they have now started using lawyers instead of rats in lab experiments? There are three reasons: there are now more lawyers in New York than rats; the lab technicians sometimes become emotionally attached to the rats; and there are some things a rat won't do.

It's rather difficult to establish what a human being won't do. In the current New York Review of Books, the philosopher Stuart Hampshire reviews T M Scanlon's wonderfully titled study of morality, What We Owe to Each Other. To put it very crudely, Scanlon's argument is that being moral is really just being reasonable in our dealings with other people.

When people do moral philosophy, they generally try to establish some basic principles that all people, at all times, in all cultures, would agree on.

Scanlon writes: "There are many questions of right and wrong that seem clearly to have correct answers. There is no doubt, for example, that murder, rape, torture and slavery are wrong. No system of rules could be a system that people had reason to accept as an ultimate, normally overriding standard of conduct if it permitted these practices."

In his review, Hampshire plausibly objects to the inclusion of slavery, since this was so widely accepted before the early 19th century. What about the rest of Scanlon's list? I would exempt torture as well. It has been used by almost every country throughout history. And can anybody, other than a strict pacifist, deny that it is sometimes justifiable in the heat of battle? As for the rest, it seems a bad week to talk about questions of right and wrong that "clearly" have "correct answers". I was tempted to mention the case of General Pinochet, but we hardly need to look that far back.

In the argument about murder we all become linguistic philosophers. Almost nobody claims it is wrong ever to kill anybody. It's just that what some people call murder, others describe as judicial process, self-defence, expression of national self-determination, regrettable wartime casualties or, for that matter, abortion.

I've never heard anybody publicly defend rape as a legitimate form of warfare, but as part of the process of ethnic cleansing, it does what feminist commentators such as Susan Brownmiller always said it did: it represents not just violence but the debasement and humiliation against an entire group of women.

More than this, as systematically practised in Bosnia, it was a deliberately obscene statement about altering the racial balance of a region. Similarly, if your policy is, for whatever reason, to remove a group of people permanently from where they have lived for generations, then, whatever they may say in a seminar room, it makes sense to kill all the men after you've torched their villages.

Modern linguistic scholars now mostly accept that if you want to discover the grammatical structure, you collect lots and lots of examples of what people actually say and you derive the rules from these. It might be nice if the language made a neat distinction between disinterested and uninterested, but the fact is that the two words have always been muddled.

In just the same way, it might be useful if moral philosophers gathered more data about the way people actually behave at moments of what they perceive to be extreme need. I agree - of course I agree - that the Serbian troops are behaving wrongly in Kosovo. But it seems to be working.