God Genes and Other Overblown Claims

One of the advantages of being at ScienceBlogs is that when confronted with idiocy like this NY Times article about the genetic basis of behavior, you can count on your fellow bloggers to tear it to shreds. Thankfully, Jonah and Dave do a wonderful job. It seems every so often this sort of sloppy genetic determinism graces the pages of the Grey Lady. The last time this happened Nicholas Kristof was expounding on the 'virtues' of The God Gene by Dean Hamer. Here's a post about that from the archives of the Mad Biologist; I think some of the points might be germane to the NY Times story.

(originally published Mar. 15, 2005)

So I've finally found some time to review Dean Hamer's The God Gene (some preliminary thoughts can be found here and here). I decided to do something unusual in discussing this book: I actually read it, and checked out some of the source literature.

Here's what the book isn't: a eugenicist screed. Here's what it is: some very sloppy science. Before I get to some particular points, I'm not allergic to the idea that some personality traits could have a genetic component, at least in certain environments, although I don't think the genetic effect would be very large (but I can accept being proven wrong). On the other hand, I need to see rock solid evidence, largely because the consequences of accepting lousy science are horrendous (case in point: don't even get me started on The Bell Curve).

The first problem is that, despite labelling his book The God Gene, Hamer admits that:

The term "God Gene" is, in fact, a gross oversimplification of the theory. The are probably many different genes involved, rather than just one. And environmental influences are just as important as genetics. Finally, spirituality, in its broader meaning, is about much more than belief in a particular God... Nonetheless, I felt it was a useful concept.

I don't think it's very useful at all, for it leads to statements like this by Kristoff:

Instead, modern science is turning up a possible reason why the religious right is flourishing and secular liberals aren't: instinct. It turns out that our DNA may predispose humans toward religious faith.

I think calling the book The God Gene, in spite of the evidence to the contrary, is self-promotion, not effective shorthand. Ok, onto the science. The crux of the data is a very large twin study in which (pp. 46-48):

scientists compared the similarity of responses [to a test that measures "self-transcendence"] of responses in identical twins to fraternal twins...

The self-transcendence score of identical twins were more alike by far than those of fraternal twins. For males, the correlations came in at 40 in monozygotic twins, compared to 18 in dizygotic twins; for females, the corresponding numbers were 49 and 26. In other words, the ratio was close to 2 to 1 in both sexes, which is just what one would expect for a genetically mediated trait, since identical twins are twice as similar at the DNA level as fraternal twins.

Essentially, what the researchers did is measure self-transcendence for each subject. Then they looked at how correlated scores were: if you scored high, how likely is it that your twin also scored high? Typically, in these analyses, heritability is measured as the difference between the correlations multiplied by two. For example, in males, heritability of self-transcendence would be (0.4 - 0.18) x 2 or 44% of the variation in spirituality is genetically based.

There are two problems with this approach. First, the data they collected were part of a much larger questionaire. We would expect that twins would show the expected pattern for some questions simply due to random chance (because Hamer doesn't present the data, there's no way to evaluate whether this is the case). This might simply be a fluke. Second, there aren't adequate controls. The appropriate control is not looking for correlations between random people. I think the appropriate control is to compare very different groups of people and see what the correlations between the two groups are. In other words, what are the 'heritabilities' of a congregation of Southern Evangelicals or a group of Northern atheists? (no offense meant to the Southern atheists or Northern Evangelicals). If you find that the correlations within these groups are in the range of ~0.1 -0.4 (and assuming the subjects aren't related), one has to wonder whether Hamer's observation is, in fact, real.

Another thing that bothered me about the twin studies was the following:

First, they evaluated the data, using a modeling technique that took into account three main sources of variation in self transcendence: genetic influence, shared environmental influences, and unique environmental influences...The analysis indicated that genes are responsible for 48 percent of the variation in self transcendence... The remaining 52 percent of variance was due to environmental factors.

Later on Hamer points out that shared environmental factors had no effect on self-transcendence. It is absolutely unclear how environmental differences between twins had an effect, whereas differences among pairs of twins did not. I understand the statistical methods, but I'm skeptical regarding the result.

One huge problem is that measurements of heritability are always taken in the context of a particular environment. In other words, a trait isn't 'heritable', it's heritable under certain conditions. If you change the conditions, you can change or eliminate the heritability of that trait. It's absolutely unclear what the environment context in these experiments is. To put it another way, would the estimates of heritability change if one were to study evangelical twins? (I'll completely ignore the issue of whether genotype by environment interactions are real or epiphenomena).

The most damning part of the book is the following (p. 77):

What I meant to say, of course, was "a" God gene, not "the" God gene. It wouldn't make sense that a single gene was responsible for such a complex trait. Beside, the numbers didn't add up. The twin studies showed that 40 to 50 percent of self-transcendence is heritable. But our analysis of the VMAT2 polymorphism ["the God gene"] showed that it raises self-transcendence sorces by only a single point, or 7 percent of the mean-less than 1 percent of total variance. That means that most of the inherited effects on self-transcendence can't be explained by VMAT2. There might be another 50 genes or more of similar strength.

An alternative explanation is that there's nothing there. This reminded me of The Bell Curve. When you actually looked at the data in the appendices (which are nowhere to be found in The God Gene), the correlations of IQ with race were typically less than 1%. Since the error in taking the test is higher than that, it didn't seem worth getting all het up about it. If Hamer wants to convince me that his less than 1% is significant, he's going to have to replicate this several times. Right now, VMAT2, the protagonist of his book, looks like a phantom effect.

There are some other problems with the book. Most of the critical studies are found in the journal Twin Research. Why aren't these studies in higher profile journals? Surely, the editors and reviewers would be interested in these topics? This alone doesn't mean anything, but I'm a little nervous.

A second problem is his chapter "The DNA of the Jews." Hamer describes how 'Cohanim', who are Jews thought to descend from Aaron the first high priest, for the most part appear to share a common ancestor (this is not new). He then suggests that religiosity, which was first acquired by cultural learning, could become genetically heritable. In principle, this could happen, by a process known as canalization. The best example of this is ether resistance in fruit flies: first, some flies resisted the effects of ether due to non-heritable physiological changes, but after several generations in the ether environment, ether resistance became genetically heritable (this was discovered by Waddington). However, Hamer presents absolutely no evidence that this is the case with religiosity.

Overall, this book involves a lot of hand-waving around very little data. Hamer suggests that spirituality could have a genetic basis (although not with VMAT2), but he hasn't presented the data demonstrating that it does.