NPR Letters on the Biological Basis of Homosexuality

NPR Letters on the Biological Basis of Homosexuality

This is an early draft of an article I recently published, with Richard Pillard, in the "Harvard Mental Health Letter" entitled something like "The Innateness of Sexual Orientation." (Copyright be damned.) Unfortunately, I don't discuss Hamer's recent X-linkage finding in "Science." ______

Recently much media attention has focused on the question of what causes some people to be homosexual and others to be heterosexual. There are at least two reasons for this attention. First, gay rights issues have been prominent on the national agenda, and some commentators have argued that etiology is pertinent to their resolution. Second, scientists have recently reported several studies supporting the possibility that sexual orientation is primarily innate, rather than psychosocially determined. The worth and implications of these studies have been hotly debated. We believe that at least some of the studies are scientifically valuable and that biological theories of sexual orientation are the most promising currently being investigated.

Two lines of evidence suggest that sexual orientation is influenced by innate processes: neuroendocrine (including neuroanatomical) and genetic studies. Neuroendocrine theories are more fully developed and have spurred far more research. At the same time, however, cumulative empirical support for neuroendocrine influences is perhaps more tenuous than that for genetic influences.

Neuroendocrine Studies

Neuroendocrine views were originally motivated by the observation that gay men are like most women in being attracted to men; lesbians are like most men in their attraction patterns. The neuroendocrine view posits a process that is analogous to the differentiation of the external genitalia, about which much is known. At the risk of oversimplifying, the basic neuroendocrine hypothesis is that sexual orientation is determined by the early (probably prenatal) effects of androgens on relevant neural structures. If these structures are effectively exposed to high levels of androgens, then they are masculinized, and attraction to women will result in adulthood. If they are not exposed to high levels of androgens (either because there are low amounts of androgens or because the relevant tissues are insensitive to their effects), the structures do not masculinize, and attraction to men will result. Several lines of research support the likelihood of neurohormonal influences on human sexual orientation, and we consider them from the least to the most direct.

Studies of rodents have shown clearly that some sex-typical sexual behavior can be affected by altering early androgen levels. Adult female rats and mice that received male-typical doses of androgens sufficiently early will, under certain circumstances, display some male mating behaviors such as mounting and reduced lordosis. Conversely, males deprived of early androgens will show the opposite pattern of behavior. It has been widely acknowledged--by both sides of the biology debate--that neither mounting nor lordosis behavior is directly analogous to human sexual orientation, in which the sex of attraction is primary. Nevertheless, these studies have been heuristically valuable in suggesting how sex-typical behavior differentiates under the influence of early androgen exposure. Furthermore, some recent studies of various nonhuman species have supported the possibility of neuroendocrine influences on preference for male versus female partners.

Perhaps the single most reliable finding in all of homosexology is that gay men and lesbians recall, on average, substantially more gender atypical behavior in childhood than do heterosexuals of the same sex. Gay men often report that they were considered Òsissies,Ó and lesbians Òtomboys,Ó though there are plenty of exceptions in both groups. This supports the idea that homosexual people have been subject to some influences more typical of the opposite sex and is thus consistent with a neuroendocrine hypothesis. Furthermore, studies of primates including humans have shown that some patterns of sex-typical childhood behavior can be altered if fetuses are exposed to unusual levels of androgens. Female rhesus monkeys exposed prenatally to high levels of androgen show elevated rates of male-typical (rough-and-tumble) play behavior.

Similar findings have come from studies of girls and women with congenital adrenal hyperplasia (CAH). CAH is a genetic disorder that causes a fetus to secrete large amounts of androgens from the adrenal glands, enough so that newborn girls with the condition often have virilized genitalia. Several studies have found high rates of masculine behavior in girls with CAH, including preference for masculine toys, increased rough-and-tumble play activity, and other ÒtomboyishÓ behavior. Even more importantly, some studies have found elevated rates of homosexual feelings among adult women with CAH compared to control women. Women prenatally exposed to DES, a chemical that causes masculinization of sexual behavior in some animals, also have an elevated rate of homosexual feelings. At the same time, it should be noted that only a minority of women with a history of DES exposure or CAH admit to homosexual feelings. Furthermore, there is no model of male homosexuality that maps as neatly onto the hypothesized neuroendocrine route as does CAH for female homosexuality.

The study most relevant to neuroendocrine theories of male homosexuality, and certainly the most widely-discussed biological finding, is that of Simon LeVay. LeVay examined the brains of gay men who had died of AIDS, as well as those of presumably heterosexual men and women who had died of a variety of causes. He studied cell-groups (nuclei) in an area of the hypothalamus that had been implicated by animal work as important to sexually-dimorphic sexual behavior. Two of the nuclei had previously been shown to be larger in men than in women, and LeVay found the same sex difference for one of the nuclei (INAH-3) when he compared heterosexual men and women. More important for sexual orientation, he found that INAH-3 was as small in gay as in heterosexual women. The differences were statistical rather than absolute, i.e., there were some gay men and heterosexual women with INAH-3 as large as those of most heterosexual men. This shows that the size of INAH-3 cannot be the sole cause of sexual orientation; indeed, it may not be a cause at all. But it does suggest that some aspects of the brainÕs sexual differentiation are shared by gay men and heterosexual women, and it is noteworthy that these include a portion of the brain that animal work shows to be involved in sexual behavior.

LeVayÕs study has been intensively scrutinized, and several criticisms have been made. These include the concern that the difference may have been due to AIDS, that the ÒheterosexualÓ sample could have included some homosexual subjects (because medical records did not include sexual histories), and that the observed difference could have been an effect, rather than a cause, of homosexuality. These criticisms are unimpressive. Some of the heterosexual men died of AIDS as well, and their average INAH-3 volume was no smaller than that of heterosexual men who died of other causes. Recent surveys have suggested that the incidence of homosexuality is rather low, and so it is unlikely that the heterosexual control groups contained a significant proportion of homosexual people. (And if they had, LeVay would have been even less likely to find the results he did.) Finally, although the possibility that the anatomical differences resulted from behavioral differences cannot be dismissed altogether, research on some other mammalian species suggests that sex differences in the hypothalamus develop early in response to innate hormonal influences and are not modified by later experiences.

At least one other neuroanatomical study has found an association between male sexual orientation and brain structure that tracks the heterosexual sex difference. Laura Allen and Roger Gorski found the anterior commissure of the corpus callosum to be relatively larger in heterosexual women and homosexual men compared with heterosexual men. This portion of the brain is unlikely to be directly involved in sexual behavior. Rather, the study suggests that neuroendocrine influences may have more general effects.

Genetic Studies

Our own work has investigated the origins of sexual orientation using techniques from human behavioral genetics. We have attempted to elucidate the degree to which people who differ in their sexual orientations do so for genetic versus environmental reasons. Both male and female sexual orientation run in families, with gay men having more gay brothers and lesbians more lesbian sisters than do same-sex heterosexuals. (It is less clear at this point if gay men have more lesbian sisters, and vice versa.) Of course, this is insufficient to show that genes matter, because a trait can run in families for environmental reasons as well (e.g., Catholicism). In order to distinguish between genetic and familial environmental influences, more sophisticated designs are necessary. Perhaps the most widely used design in human behavioral genetics is the classical twin study, in which monozygotic (MZ) and dizygotic (DZ) twin pairs are examined for their similarity on the trait of interest. The rationale for this design is that because both MZ and DZ twins are typically reared together, they are equally similar environmentally. However, MZ twins are identical to each other genetically, while DZ twins are only as genetically similar as ordinary brothers. Thus, if MZ twins are more similar than DZ twins on average in their sexual orientation, the importance of genetic factors is supported. An explicit assumption of this design is that the environmental factors important for sexual orientation are no more similar for MZ than for DZ twins. This "equal environments assumption" has been criticized. However, in studies of other personality traits, those aspects of the environment that are particularly similar for MZ twins (such as being dressed alike) have not appeared to be very important in causing behavioral similarity.

There have been several twin studies of sexual orientation. The earliest, by Franz Kallmann in 1952, found an amazing 100% concordance rate among 37 male MZ twin pairs, compared to a much lower rate of 12% for 26 male DZ twin pairs. Kallmann's study had enough methodological problems that its specific results, particularly the 100% MZ concordance rate, cannot be taken very seriously. On the other hand, no one has provided a satisfying explanation of how the methodological problems could wholly discredit evidence for genetic influences. Thus, it is somewhat surprising to us that nearly 40 years passed before others attempted to follow up his promising results systematically.

The largest genetic studies to date have been our own, one with men and one with women. These studies included not only MZ and DZ twins, but a third group, homosexual subjects with adoptively-related same-sex siblings, who are genetically least similar and thus should be least similar in their sexual orientations. Our two studies obtained results that were quite similar to each other. In the male study, MZ twins of our gay index subjects had a 52% chance also of being gay, compared to a 22% rate for their DZ twins and an 11% chance for their adoptive brothers. In the female study, MZ twins had a 48% chance of also being lesbian, compared to a 16% rate for DZ twins and a 6% rate for adoptive sisters. Note that similarity in sexual orientation corresponded closely with genetic similarity in both studies.

Quantitative analyses provide heritability estimates of the relative proportion of genetic causation, and our heritabilities were consistently above 50%. Thus, both studies were consistent with moderate to strong genetic influences on sexual orientation. Of course, both studies yielded figures lower than the 100% suggested by Kallmann. Two newer twin studies more or less supported ours. One found concordance rates that were lower than ours, the other found higher rates, but both found MZ higher than DZ rates.

All recent studies find that MZ twins often differ in their sexual orientations, which shows that sexual orientation cannot be completely explained by genes. Environment must play its part. On the other hand, we emphasize that environmental pathways can be biological as well as psychosocial. One of the most interesting questions to stem from the recent twin studies is what environmental factor could affect MZ twins differently enough to give them opposite sexual orientations.

One other twin study is worth mentioning because it contained what many consider to be the ideal (if rare) subjects: twins reared apart. In this study both of two male pairs were concordant for adult homosexual feelings and behavior, which would be highly unlikely if genetic factors were unimortant. Although none of four female pairs was concordant, this sample was too small to be conclusive; of course it provided no support for genetic factors affecting female sexual orientation.

Although we believe the available evidence is strongly suggestive of genetic influence on sexual orientation for both men and women, one serious methodological limitation of available studies prevent them from being definitive. The studies have recruited subjects via advertising in gay- and lesbian-oriented publications. They may have had an overrepresentation of concordant pairs because such studies might be more appealing to gay men and women with gay twins. Such bias is not very serious unless it was stronger for MZ than for DZ twins, and we have no reason to suspect that such differential bias occurred. But only a study using systematic ascertainment could exclude this possibility, and unfortunately none exists.

Research that helps settle some questions raises new ones. By what pathway do genes lead to the expression of atypical childhood behavior and homosexual orientation? Our studies have usually been discussed as supportive of neuroendocrine theories, and genetic influences might indeed be neuroendocrine. One could imagine a gene that led a fetus to secrete unusually high or low levels of prenatal androgens during brain differentiation, or alternatively, to be especially sensitive or insensitive to androgens. But genetic influences on sexual orientation would not have to be neuroendocrine in nature. For example, genes could influence personality (say, independent thinking or unconventionality) that could increase the chance of adopting a homosexual identity. For reasons we discuss later, we doubt that the indirect route through personality is typical, and instead believe that genetic influences operate via a neuroendocrine pathway. But our studies have had nothing so far to say about this.

The Biology Debate

A thorough and thoughtful critique of the biological evidence has been provided by Byne and Parsons, in a recent version of Archives of General Psychiatry. They argue that the biological case is quite weak. Byne has asserted that reviewing the biological evidence is akin to "adding zeroes" because no one study is methodologically strong enough to establish the biological case definitively. While we agree that no one study has established the case for biological factors, we disagree that their sum is "zero" evidence in favor of biological hypotheses. It is difficult if not impossible to do perfect studies using human subjects. But if different kinds of studies with different research strategies (with different methodological inadequacies) converge to similar conclusions, then the critic who rejects the cumulative evidence begins to sound strained. For example, disregarding other research one might reasonably worry (as Byne and Parsons do) about the role of postnatal socialization of girls with CAH, since some are born with virilized genitalia and they and their parents know about their condition, and for this reason may be treated more like boys. But what if one knows that the only studies that have investigated this possibility have failed to confirm it? Furthermore, it has been well-established that androgen administered at critical periods can masculinize female rhesus monkey's play and sexual behaviors. How likely is it that the apparently analogous behavioral masculinization of some CAH females has a completely separate explanation?

A second complaint we have of Byne and Parsons' otherwise useful critique is its one-sided perfectionism. Scientific theories are not evaluated in a vacuum. Rather, their evidentiary basis and plausibility are compared to those of the competition. There have been two main psychosocial competitors to innate theories of sexual orientation: psychoanalysis and socialization theory. Psychoanalytic theories of homosexuality, which have stressed the role of family psychodynamics, suffer from the same well-known problems as does the corpus of psychoanalysis, particularly the inadequacy of impressionistic data collected during psychotherapy under uncontrolled conditions. Although there has been some support for the psychoanalytic prediction that fathers are somewhat distant from their gay sons, this is a rather small effect that is easy to explain on other grounds, such as fathers' intolerance of their sons' gender-atypical behavior.