23 March 2011

In which several evolutionary psychologists still don't understand evolution

ResearchBlogging.orgJesse Bering has responded to criticism—by me, Jon Wilkins, and P.Z. Myers, among others—of his post about Gordon Gallup's hypothesis that fear of homosexuals is favored by natural selection, in the form of an interview with Gallup. The result is informative, but probably not in the way intended.

To recap: Gallup proposed that homophobia could be adaptive if it prevented gay and lesbian adults from contacting a homophobic parent's children and—either through actual sexual abuse or some nebulous "influence," making those children homosexual. In support of this, he published some survey results [$a] showing that straight people were uncomfortable with adult homosexuals having contact with children.

I pointed out that all Gallup did was document the existence of a common stereotype about homosexuals—he presents no evidence that believing this stereotype can actually increase fitness via the mechanism he proposes, or that it is heritable.

Homophobia. And, um, everyone-else-phobia, too. Photo by yksin.
So now Gallup and Bering have responded, although they have not, I think, improved their case. There's a lot for me to address here, so I'll try to break it up into sections, and follow the order of the interview.

In which Gordon Gallup is not a homophobe

In the response post, Gallup (and Bering, who contributes quite a lot to the argument in his role as interviewer) takes issue with the collective objections of working biologists, but manages not to actually address those objections. Bering starts the conversation on the moral high ground:

BERING: Let’s address the elephant in the room. It’s embarrassing for me to even ask this of you, since the answer is so obviously "no" to me. Is your theory a justification of your own homophobia?

GALLUP: A lot of people think that if a person has a theory it’s a window unto their soul. I have lots of theories. (See CV (pdf).) I have a theory of homophobia, I have a theory of homosexuality, and I have a theory of permanent breast enlargement in women, just to mention a few. So that would make me a homophobic, homosexual who is preoccupied with women’s breasts.

Neither I, nor any of the other critics I've seen have called Gallup a homophobe. He may be uniquely bad at understanding how societal homophobia nullifies his interpretation of his survey results, but that doesn't make him a homophobe. Thanks for clearing that up, though, guys.

Gallup then demonstrates that he either hasn't actually read any of the latest criticism, or has missed the point entirely:

... It is interesting how my critics tip-toe around the fact that my approach is based on a testable hypothesis, and how they go out of their way to side-step the fact that the data we’ve collected are consistent with the predictions. Whether it is politically incorrect or contrary to prevailing social dogma, is irrelevant. In science, knowing is preferable to not knowing. Minds are like parachutes, they only function when they’re open. If I were a homosexual, I’d want to know about these data.

I certainly didn't tiptoe around the testability of Gallup's hypothesis—I wrote that (1) the data he presented do not test his hypothesis, and (2) the data we do have regarding the probable fitness benefits of homophobia and its heritability contradict his hypothesis. I'm entirely prepared to revise my conclusions given new data, but Gallup doesn't have any.

In which at least one of us doesn't understand heritability

In his next question to Gallup, Bering accuses me of "bungling" the definition of heritability, linking to evolutionary psychologist Rob Kurzban, who says that my brief definition of heritable as "passed down from parent to child more-or-less intact" is wrong because heritability is actually "the extent to which differences among individuals are due to differences in genes."

Wow, dude. You are aware that what you just said means exactly the same thing as what I originally said, right?

Let's go to the textbooks that Kurzban says I'm contradicting. Here's the passage on heritability from Douglas Futuyma's gold-standard undergraduate textbook Evolution (page 209):

One way of detecting a genetic component of variation, and of estimating VG [trait variation attributable to genetic differences] and h2 [the proportion of total trait variation explained by genetic variation], is to measure correlations* between parents and offspring, or between other relatives. For example, suppose that in a population, the mean value of a character in the members of each brood of offspring was exactly equal to the value of that character averaged between their two parents (the MIDPARENT MEAN) (Figure 9.20A). So perfect a correlation clearly would imply a strong genetic basis for the trait. [Bold text and bracketed notes mine; otherwise sic.]

The asterisk in that quote leads to a footnote pointing out that regression, rather than correlation, is more typically used. This is the definition of heritability that I learned in my undergraduate and graduate courses. It's also the definition I've just helped teach to a class of third- and fourth-year undergraduate biology students in my capacity as a teaching assistant on a course in population biology.

In non-statistical terms (the kind I try to use on this blog), a positive regression between a parent's traits and those of their offspring means, in fact, that the parent's traits are passed on to their offspring, um, more-or-less intact.

Parent-offspring regression is widely used to estimate heritability [PDF], but you can also do similar analyses using trait measurements for siblings, or multiple generations on a pedigree. In all of these cases, known parental or sibling or familial relationships are proxies for genetic similarity—you can estimate heritability without knowing anything about specific genes. (In fact, sometimes biologists use genetic data to reconstruct pedigree relationships, then estimate heritability from the pedigree.) As implied in the quote from Futuyma's textbook, this approach is statistically equivalent to showing that there is a significant portion of trait (phenotype) variation explained by genetic variation—which is where Kurzban seems to have become confused.

Wild parsnip, mostly here to break up the wall of text. Photo by Bas Kers.
Here's a specific example near and dear to my field of study, species interactions: To determine whether parsnip webworms could be under natural selection to resist nasty chemicals produced by their food plant, the wild parsnip, May Berenbaum and Arthur Zangerl estimated the genetic component of variation [$a] in the worms' capacity to choose food with lower levels of the toxins, and to tolerate the toxins they did eat. To do this, they raised webworm larvae of known parentage in the lab, and tested them on controlled diets. Their actual statistical analysis tested for an effect of the worms' sibling relationships (parentage) on their ability to avoid toxins and survive them.

In all of Gallup's lengthy response to Bering's question about heritability, he doesn't say a word about this kind of data with regard to homophobia. That's because it doesn't exist, and, as far as I can tell from the interview, he has no intention to try and collect it. To be completely fair, it's harder to collect heritability data on humans than on webworms—but it's hardly impossible. As Gallup notes, there are studies documenting heritability for, of all things, human grip strength [PDF].

Kurzban's critique is correct in one very specific regard, which Bering doesn't touch on. It is relatively difficult, both for logistic and resource-related reasons, to estimate a trait's heritability and determine whether natural selection is acting on it within the same study. (Although there are plenty of exceptions—here's one example [$a] pulled at random from my reference library.) That's why I said, in my original post, that biologists expect evidence for heritability or fitness benefits in support of an initial claim that a trait or behavior is adaptive. The study I cited as an example of support for adaptation—which shows that horned lizards' horns prevent predator attacks [PDF]—demonstrates fitness benefits, but not heritability. This point should be familiar to anyone who regularly reads the evolutionary biology literature.

Grip strength: known to be heritable. Homophobia: not so much. Photo by West Point Public Affairs.
So, again, Gallup has no data on the heritability of homophobia. The rest of his interview shows that he still doesn't have any data to demonstrate fitness benefits for it, either.

In which evidence of fitness benefits also remains absent

Gallup then comes to the question of whether a child who would otherwise be straight can be "converted" to homosexuality by early same-sex sexual contact.

As detailed in my 1996 reply to Archer, we’ve collected data from male homosexuals that show that most gay males don’t report getting a clear sense of their homosexual orientation until they have their first same-sex postpubertal sexual experience.

Most gay men don't know for sure that they're gay until they've actually, you know, tried gay sex? Quelle surprise. This is absolutely classic mistaking of correlation for causation, and it suggests that Gallup doesn't know much about the actual experience of sexual minorities. When you grow up surrounded by straight people, it often takes very direct evidence to convince you that you're attracted to people of the same sex. If same-sex activity shortly after puberty can cause homosexuality, wouldn't parents be most concerned about homosexuals having contact with teenagers? At the risk of sounding like a broken record, this is yet another thing we can't tell from Gallup's survey data—he asked about pre-pubescent children, and in one case 21-year-old children, but not children who have just passed puberty.

Finally, Gallup deals with the relative risk that homosexuals will molest children. He does this by moving the goalposts for pedophilia:

There is also evidence that shows that the propensity to have sex with minors is positively correlated with promiscuity among homosexual males. Unlike heterosexual pedophiles, homosexuals who have sex with minors target young postpubertal victims.

That's not pedophilia Gallup is talking about—that's violation of age-of-consent laws. The comparison between heterosexual-identified pedophiles, who target children, and homosexuals who have sex with post-pubertal teens under the age of consent is, frankly, intellectually dishonest. By definition those are two different groups. The comparison to make is that between all homosexuals who have had sex with minors and all heterosexuals who have had sex with minors. I would imagine that, as Gallup basically admits in his next sentence, those two groups look much more similar.

So that's where we stand: still no evidence that homophobia is heritable, and still no evidence that it provides a fitness benefit by preventing the homophobe's children from becoming homosexuals. Gallup's only data are still, over fifteen years after his initial publication, a set of survey responses that are consistent with any number of hypotheses for the origins of homophobia. Claiming that those data demonstrate an adaptive function for hatred of homosexuals doesn't just fail the standards of evidence for evolutionary biology, it's bad scientific reasoning.

In which we come to a conclusion of sorts

In a coda to the interview, Bering accuses me and his other critics of failing to engage with Gallup's results. I think my previous discussion, and Bering's response to it, speak for themselves. Bering has demonstrated to me that he doesn't understand undergraduate-level biology, and that, as Will Wilkinson suggested, he's more interested in ginning up controversy than scientific rigor. (On which point he wins, I suppose. D&T's visit count went through the roof when P.Z. Myers linked here.)

Bering also makes some conspicuously uninformed speculations about my own experience and motivations. I won't dignify that with a response except to say yes, Jesse, I'm gay, and you don't know the first thing about what I have or haven't encountered in the way of "palpable disapproval." First and foremost, though, I'm a scientist. Contrary to what you seem to think, I love a good counterintuitive, paradigm-shifting hypothesis, but I also expect it to be supported with data.

Bering, however, is convinced that he's established himself as a hard-nosed scientific iconoclast in opposition to all us stodgy, dogmatic, evidence-demanding biologists. He concludes,

So, I’ll continue to dredge up any old theory, no matter how meager the supporting data ...

Clearly, Jesse, I can expect nothing more of you.


References

Arden, N., & Spector, T. (1997). Genetic influences on muscle strength, lean body mass, and bone mineral density: A twin study. Journal of Bone and Mineral Research, 12 (12), 2076-2081 DOI: 10.1359/jbmr.1997.12.12.2076

Berenbaum, M., & Zangerl, A. (1992). Genetics of physiological and behavioral resistance to host furanocoumarins in the parsnip webworm. Evolution, 46 (5), 1373-84 DOI: 10.2307/2409943

Young, K. (2004). How the horned lizard got its horns. Science, 304 (5667) DOI: 10.1126/science.1094790

Campbell, D. (1996). Evolution of floral traits in a hermaphroditic plant: Field measurements of heritabilities and genetic correlations. Evolution, 50 (4), 1442-53 DOI: 10.2307/2410882

Futuyma, DJ. (2005). Evolution. First ed. Sunderland, MA: Sinauer Associates. Google Books.

Gallup, G. (1995). Have attitudes toward homosexuals been shaped by natural selection? Ethology and Sociobiology, 16 (1), 53-70 DOI: 10.1016/0162-3095(94)00028-6

Mousseau, T., & Roff, D. (1987). Natural selection and the heritability of fitness components. Heredity, 59 (2), 181-97 DOI: 10.1038/hdy.1987.113

No comments:

Post a Comment