why I no longer believe religion is a virus of the mind

Guardian | Are religions viruses of the mind? I would have replied with an unequivocal "yes" until a few days ago when some shocking data suggested I am wrong.

This happened at a conference in Bristol on "Explaining religion". About a dozen speakers presented research and philosophical arguments, mostly falling into two camps: one arguing that religions are biologically adaptive, the other that they are by-products of cognitive mechanisms that evolved for other reasons. I spoke first, presenting the view from memetics that religions begin as by-products but then evolve and spread, like viruses, using humans to propagate themselves for their own benefit and to the detriment of the people they infect.

This idea began with Richard Dawkins's The Selfish Gene, was developed in his later article "Viruses of the mind" and taken up by others, including myself in The Meme Machine and other works. It is one version of "dual-inheritance" theory in which genes and culture are both seen as evolving systems.

The idea is that religions, like viruses, are costly to those infected with them. They demand large amounts of money and time, impose health risks and make people believe things that are demonstrably false or contradictory. Like viruses, they contain instructions to "copy me", and they succeed by using threats, promises and nasty meme tricks that not only make people accept them but also want to pass them on.

This was all in my mind when Michael Blume got up to speak on "The reproductive advantage of religion". With graph after convincing graph he showed that all over the world and in many different ages, religious people have had far more children than nonreligious people.

The exponential increase in the Amish population might be a one off, as might Catholics having lots of children, but a comparison of religious and nonaffiliated groups in the USA, China, Sweden, France and other European countries showed that the number of children per woman in religious groups ranged from close to zero (for the Shakers) to between six and seven for the Hutterites, Amish and Haredim, while the nonaffiliated averaged less than two per woman – below replacement rate.

Data from 82 countries showed almost a straight line plot of the number of children against the frequency of religious worship, with those who worship more than once a week averaging 2.5 children and those who never worship only 1.7 – again below replacement rate. In a Swiss census of 2000 the nonaffiliated had the lowest number of births at 1.1 per woman compared with over two among Hindus, Muslims and Jews.