Comment

IT'S A matter of size. Or perhaps movement. The perception of risk, that is. The Paddington train crash has brought out once more the peculiar way in which human beings assess risk, and how differently we behave from the way the best theories tell us to. The best theories are elegant, but their foundations are simple enough. The assessment of risk is part of the wider subject of gambling on outcomes good and bad; and the basic logic is this: how much it's worth putting on a horse is a matter of calculating the probability that the beast will win (or place) and multiplying that by the pay-off. Of course, once you try, you discover that it is no accident that most bookies are richer than their clients. Conversely, the amount of money it's worth investing in reducing risk is a matter of multiplying the (negative) pay-off by the probability of the event occurring.

After the Paddington rail crash, the usual arguments flew about. "You can't set a price on life", met its old friend "of course we set a price on life". So far as that goes, it is obviously true that we do in fact set a price on life; but we do it in a funny sort of way. Ahead of time, we are quite happy to engage in cost-benefit analysis; when we are in the middle of trying to save even one life, we behave quite differently. The NHS would be criminally irresponsible if it allocated its facilities without trying to ensure that it got the biggest possible bang for the taxpayers' buck.

How do you measure the reduction of deaths from breast cancer against the alleviation of long drawn-out incapacitating aches and pains from displaced discs and fallen arches? None of the obvious answers stay obvious for long. Price an illness in terms of lost production, and the deaths of the retired and the not yet employed drop out of the equation; leave out the economic spin-off from restoring the incapacitated to good working order, and you ignore the fact that a healthier population can pay for a better NHS in future. Try to measure the pain and distress of the bereaved in monetary terms, and you open the gates to American juries handing out millions to people whose lawyers tell a good story about heartbreak. Measure only the economic loss, and the families of the unproductive get nothing.

But even when we have a view about the value of an outcome, we are in trouble. We all know the familiar contrasts; we fear flying more than we fear the drive to the airport, but shouldn't. We discount risks that we feel are under our control, even though we know that we are unskilled and under-trained drivers and the pilot of the plane is highly trained and highly skilled. We also bury what you might call "background" dangers deep below the level of conscious anxiety. We know that the Paddington crash killed fewer people than three ordinary days of road accidents. If we minded most about reducing the danger of death from travelling to work, we would spend the extra billion pounds that John Prescott wants Railtrack to spend on road rather than rail safety.

So why don't we? Why do we over-estimate the chances that we shall die in large, violent, catastrophic events and under-estimate the risk of breaking our necks when we stand on a chair to reach the top shelves in the kitchen? Are we just hopelessly irrational? It is possible that we are. We are as a species not good at assessing evidence; when we try to bring evidence to bear on an issue, we reach for salient, colourful examples of what we are concerned with rather than carefully sifted and boring statistics. We are not very good at distinguishing between what readily comes to mind and what is likeliest to happen in the world outside. Much of the history of science has consisted of the process of forcing ourselves to reject colourful coincidences in favour of dreary but real connections.

Of course, the economists can (always) find a way to avoid impugning our rationality. They simply note that we set a higher value on dying in ones and twos and in cars, than in twenties and thirties or two or three hundreds in trains and aeroplanes; plug in those utilities and disutilities, and the sums again make sense. My own guess is that we are, yet again, the victims of our evolutionary history. A mute animal, with little memory, and almost no resources for sifting evidence, does about as well as it can by watching out for the large, sudden, novel movement against a background that is mostly ignored. No doubt our pre-human ancestors were more likely to perish of infection or from the violence of their family members than from the attentions of large, carnivorous animals. But given their sensory and intellectual apparatus, they were doubtless right to focus on the sabre-toothed tiger before they thought about E. coli. Having inherited most of that apparatus, we have to work against it while we work with it.

The writer is Warden of New College, Oxford