Field of Science

Why we are all different (and not all religious)

So, on to the paper by Robert Rowthorn, which I see now has even been picked up by the Denver Post!

Just to explain a bit of the background. Rowthorn is an economist, and his paper is basically a model of what would would happen if you have a gene (strictly speaking [and for Bjørn's benefit], an allele) that predisposes for membership of a group, and if that group has high reproduction.

What he shows is that the gene spreads incredibly quickly - just 10 generations after it appears, 80% of the population have it. After 20 generations, 95% have it, and it keeps increasing until that figure reaches 100%.

Because the gene spreads, membership of the group increases. It starts off a little slower, and never quite reaches 100% (because even gene carriers can opt out in this model).

Since there is a gene for religion, and the religious (especially the ultra-conservatives) have more kids, you can easily see how this model directly relates to the real world.

And that, Dear Reader, is why 95% of the world's population are Orthodox Jews!

Oh. Alright there is clearly something wrong with the model, but that's OK. Models are usually wrong, and the fun part comes in trying to figure out why. And this model is particularly interesting because it touches on a number of popular misconceptions.

Some are specific to this model but there's also a much bigger issue underlying all this: just why is there so much variation in how religious people are, if (as some theorists will have you believe) religion is so beneficial?

But first, let's look at some specific problems with the assumptions in this model. First off, it isn't really a model of religion, despite the title of the paper. It's a model of conservatism.

Rowthorn starts from some basic assumptions. That global birthrates have fallen, but they have fallen more slowly among the most religious (and not at all among certain sects like Orthodox Jews and the Amish). That conservatism and religion are inextricably linked (and that they have a simple genetic basis). And that religion is inextricably linked to high birth rates.

A quick survey just of European history will quickly show that the last two assumptions don't hold. There have been countless examples of religious anti-conservative movements - the Protestant reformation is just the most obvious example, but there are numerous others, like the anti-slavery movement and the 12th century reformation.

Religion is invented by people, and religion can be radical and innovative - according to their needs.

And of course religion does not necessarily promote fertility. Throughout most of its history, the Catholic Church has been at pains to promote the moral value of 'sexual continence'. The most religious people eschewed sex, and were packed off in their tens of thousands to monasteries and nunneries - places not renowned for their fecundity! Not just Europe either - Burma even now has around 400,000 monks.

So the link between religion and fertility is a cultural one, and it's a product of our recent history. Cultural innovation in recent decades has been tied to non-religion. Conservative people by their nature are likely to shun innovative lifestyles, and are more likely to adopt the lifestyles prevalent in the past. From the perspective of the early 21st century, that means things like religion and traditional families.

One of the problems with Rowthorn's model is that assumes that there are only two groups. A high fertility, religious (conservative) group, and a low fertility, non-religious (liberal) group. People who carry the 'gene' can switch out of the religious group, but their descendants will still be more likely to switch back in to that same group.

But what happens if the genes for conservatism and religion are not inextricably linked? What happens if new cultures arise that are religious, but do not promote fertility, or which are conservative, but do not promote religion?

And, of course, what happens if there is not really a gene for religion at all? Are the Amish really genetically disposed to be Amish? Or is it simply that they have been brought up to be Amish? Even if they are, will their offspring be more likely to rejoin the Amish, or will they simply be more likely to join some other traditional cultural group (steam engine enthusiasts, perhaps).

This is not to knock Rowthorn's paper, which gives a great insight into gene-culture interaction. It shows very nicely that a gene can spread faster than the culture that promotes it.

But it is a caution against running away with conclusions about a complex world based upon a simple, unvalidated model.

It raises a much bigger question, too. Just suppose there really is a culture that is 'best' (from an evolutionary perspective). And just suppose that there is a personality that is predisposed to that culture, and a gene that predisposes to that personality. Well then, if such a thing exists, why isn't everybody like that?

After all, evolution is ruthlessly efficient. A gene that reduces fitness by only 1% will be eliminated in just 10 generations, and will be carried by just 100 individuals. There's an excellent paper published a few years ago that tackles question of why on earth people have different personalities (see the footnote for reference).

After ruling out genetic drift and random mutation, the authors conclude that the most likely reason different people have different personalities is that there is no one 'optimal' strategy. In different times, or in different places, different approaches might be best.

This effect is called 'balancing selection', and it results in a number of different 'evolutionary stable strategies' - i.e. personalities.

What's more, these stable strategies often show negative frequency dependence. What that means is that the benefit of having a particular personality decreases the more that other people have the same personality.

In other words, human society creates a kind of ecosystem, and different personalities occupy different niches in that ecosystem.

Even then there are some nuances. Reproductive fitness is not just a question of numbers of offspring. Each organism has to make a choice of whether to invest resources in growth, in reproduction, or in survival. The optimum balance depends on - you guessed it - the particular environment (both physical and cultural).

Of course, they also go on to point out that there is no direct link between genes and personality. Rather, they suggest that multiple genes act to influence 'personality mechanisms' - like the 'startle reflex'. Depending on the environmental setting, these contribute to personality traits (the startle reflex has been linked to political conservatism). But even these personality traits are then further influenced by the environment to produce the end product - behaviour.

In other words, genes act at a very basic level to create a brain tool kit. These tool kits can generate a number of different personalities (and, ultimately, behaviour) depending on the environment.

And the reason you have the toolkit that you have - which is unique to you and may, depending on your environment, have resulted in you being religious or nonreligious - is because the components of your particular brain toolkit helped your ancestors to survive.

None of us, whether religious or non-religious, are maladaptive - from an evolutionary perspective, at least!

(Well, except those of you out there reading this who have congenitally low IQ - but that's a different story and you'll have to read the paper yourself to find out why!)
Rowthorn R (2011). Religion, fertility and genes: a dual inheritance model. Proceedings. Biological sciences / The Royal Society PMID: 21227968

Penke, L, Denisson, J, & Miller, GF (2007). The Evolutionary Genetics of Personality European Journal of Personality, 21, 549-587

Creative Commons License This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.


  1. Wow, Tom, that post had all kinds of interesting ideas in it. Thank you.

  2. Interesting stuff. I'll have to give the full paper a read.

    Incidentally, the ideas you mention towards the end - about personality and evolutionarily stable strategies and frequency-dependent selection etc - are very close to my own research on the evolution of morality.

    Because there's no one 'perfect' strategy for managing social interaction and keeping groups strongly bound together yet flexible enough to deal with a changing environment, we've evolved a bunch of mechanisms that predispose us towards a range of 'strategies' - some 'liberal' and some 'conservative'.

    And frequency-dependent selection (among other things) works to maintain a pluralism of strategies within a group.

    The end result: different people are born predisposed to different moral (and political) attitudes. Some are nice, some are suspicious, some are nasty. And often they'll just end up talking past each other.

  3. Tim, in case you are interested there was a supplement in Royal Soc B in December last year on the "Evolutionary Genomics of Animal Personality". Couple of papers there look very relevant - and a bit more up to date than the 2007 one I cited!

  4. Knowledge of advanced statistics is lacking with Rowthorn's work. He is applying a model which does not work with human evolution, simply because of the assumptions it makes.

    He takes the average offspring number of the made up religiosity gene, and then applies that to typical evolutionary models to "demonstrate" that the gene will fix in the population. His model suggests that it will fix in less than a dozen generations! So, either the gene is new (a few hundred years old), or is has already fixed in the population. Maybe you can already see this model is not accurate.

    The inaccuracy arises when using the average offspring number as a meaningful measure. I will point you toward NN Taleb's work to find out why this does not work. As an economist, he falls into a common trap, for them and biologists, that averages matter.

    If some gene has a high average number of offspring, but also a high variance in offspring number, then this gene may actually grow slowly in the population, or not at all. An economist should know that averages do not describe geometric growth (g). A simple (economic) estimate of "g" is g=sqrt[(mean)^2-(variance)^2]

    This is a HUGE hole in his model. Another hole is that environments change, and this "gene" may not always and forever have a reproductive advantage (if it does have one). He also ignores the fact that humans have class structured populations. For these reasons, and more, this work should be disregarded as it is very inaccurate.

    Forgive me for remaining anon here, but as a scientist it allows me to openly discuss my opinions. Thank you for your post.


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS