Richard Green. Journal of Gay & Lesbian Mental Health. Volume 12, Issue 1-2. 2008.
Introduction
The association between gender-typed behavior in childhood and sexual orientation in adulthood is staggering. A metanalysis of 28 studies, with a median sample size of 189, determined the statistical effect size associating sexual orientation with childhood gender-typed items (Bailey and Zucker, 1995). The studies covered rough-and-tumble play, toy and activity preferences, role playing, cross-gender dressing, sex of peer group, appellations of “sissy” or “tomboy,” and gender identity. Statistical effect sizes of childhood behaviors distinguishing adult heterosexuals and homosexuals were “among the largest effect sizes ever reported in the realm of sex-dimorphic behaviors” (p. 49).
Gender nonconformity is the term often used to describe childhood cross-gender behavior. However, it is not defined in all studies of boys as “feminine-type” behaviors. Some studies refer to gender nonconformity as the absence of “masculine-type” behaviors (Hockenberry and Billingham, 1987). Childhood gender-type behaviors have been contrasted between groups of heterosexual versus homosexual subjects as well as within samples of primarily heterosexual persons with minor degrees of homosexual interest or behavior. For males, their degree of reported disinterest in boyhood rough-and-tumble play and sports correlated with degree of current homosexual interest (McConaghy, Buhrich and Silove, 1994). Among males and females self-identified as heterosexual, differences in gender nonconformity were recalled by those “completely” heterosexual from those reporting rare same-sex sexual behaviors, but no homosexual attraction (Dunne et al., 2000).
Why the powerful association between childhood cross-gender behavior or childhood gender nonconformity (CGN) and adulthood same- sex eroticism? A first response answers, why not? Nearly all children who behave like girls (most of whom are girls) are later attracted to men, and nearly all who behave like boys (most of whom are boys) are later attracted to women. The association may therefore not be surprising. However, its basis remains enigmatic.
Biologically Based Theories
There are no satisfactory non-evolutionary explanations for oppo- site-sex erotic attraction. It seems to “come about” in nearly every non-human animal and in all human societies. Certainly, unless a species were hermaphroditic, or could reproduce by binary fission, it would not have had much of a future without heterosexual mating. So, species that survived the Darwinian obstacle course mutated and evolved with this advantage. That heterosexuality must therefore be essentially “automatic” (innate) may mean that homosexuality is also “automatic.” A prevalence of less than 5% does not dent Homo Sapiens’ population escalation, more at risk from over-zealous mixed-sex mating.
What could be the basis of this “automatic” homosexual outcome? Extensive clinical research has documented a relation between prenatal levels of sex steroids and postnatal gender-typed behaviors. Obviously, nearly all boys and girls with normal genitalia at birth had different levels of androgen exposure prenatally. But, girls with congenital virilising adrenal hyperplasia (CAH) are exposed to higher levels of androgen in utero than typical females. Their childhood behaviors are masculinized (essentially “tomboyish”) (Berenbaum and Hines, 1992) and more of these women are later bisexual or homosexual than expected by chance (Money, Schwartz and Lewis, 1984; Meyer-Bahlburg et al., 1996). However, most of those children with a prenatal excess of androgen and substantially virilized genitalia evolve a female identity and are sexually attracted to male partners only. Further, in most persons with extensive cross-sex identity (transsexuals) or those with exclusive same-sex erotic attractions (homosexuals), if prenatal sex steroid levels are implicated, they did not act at a time to alter genital development. That genitally normal appearing infants are usually those who develop a homosexual orientation does not rule out atypical prenatal sex steroid levels as influential. Hormones active at different critical developmental periods prenatally can affect different systems, behaviorally and anatomically. With rhesus monkeys, exposing the female fetus to androgens at an early and late time period masculinized different sex-typed behaviors and late exposure virilized genitalia (Goy, Bercovitch and McBriar, 1988).
However, this favored intellectual pursuit of prenatal sex steroids could turn out to be a chase “a la wild goose.” There is evidence, in the mouse, for the sexually dimorphic pattern of expression of genes in embryonic brain development to be independent of, and occur before, the availability of sex hormones. There may be a direct genetic influence on brain sexual differentiation (Dewing et al., 2003, Arnold, 2002). However, in one clinical example where genes and hormones contradict each other, the complete androgen insensitivity syndrome (AIS), hormonal influences dominate. There the Y chromosome is present but cells are unable to utilize androgen. The result is a female-appearing body, female sexual identity, and usually sexual attraction to males (Wisniewski et al., 2000).
An alternative non-hormonal explanation being pursued is genomic imprinting (Bocklandt and Hamer, 2003). Here, whether a trait is expressed is dependent on whether it is transmitted by the mother or father to the child. Our work with male-to-female transsexuals finding a deficiency of maternal uncles versus aunts (Green and Keverne, 2000) and the work of Turner (1995), finding the same deficiency with male homosexuals, is consistent with genomic imprinting. It appears to effect a lethality factor in one generation and pass on a gene with sex-atypical features to the next.
Neuroanatomical or gene locus bases of sexual identity and sexual orientation are tantalising but also premature as a satisfactory explanation. The post-mortem study of a nucleus of the anterior hypothalamus that shows a heterosexual male-female size difference and a homosexual male versus heterosexual male size difference has been replicated only in part. The INAH 3 nucleus, larger in males, was smaller on average in 19 homosexual males versus 16 heterosexual males (Le Vay, 1991). The long-awaited replication of this study confirmed the size difference but found that the number of neurons in the nucleus of the two groups did not differ (Byne et al., 2001).
A portion of the bed nucleus of the stria terminalis also shows a male-female size difference. A post-mortem study of transsexuals found it to be female-sized in 6 male-to-female transsexuals and male-sized in 1 female-to-male transsexual (Zhou et al., 1995). The sex difference is not evident until young adulthood, after many transsexuals manifest cross-sex identity. Further, as this sex-size difference develops postnatally, it is possibly correlated with transsexualism, without being causally related. No independent replication is reported.
A gene locus on the X chromosome at Xq 28 was found to be shared more commonly by homosexual brothers than by homosexual and heterosexual brothers. These were families where there were other homosexual males on the mother’s side, the mother presumably passing the X chromosome and gene locus to her sons (Hamer et al., 1993). In a sample without the maternal family history of male homosexuality, the gene locus sharing between homosexual brothers was not found (Rice et al., 1999).
Twin studies addressing both sexual orientation and childhood gender-typed behaviors can be very informative. A colossal study of 4900 twins found a genetic basis for both cross-gender behavior in childhood and homosexual behavior in adulthood. Childhood gender nonconformity, retrospectively recalled, was measured by 24 items. For male monozygotic (MZ) twins, sexual orientation correlated 0.51 between twins and CGN correlated 0.54. Among male dizygotic (DZ) twins, the correlations were only 0.11 and 0.14, respectively. With female MZ twins, sexual orientation and CGN correlated 0.49 and 0.42 between twins, respectively, but with female DZ pairs the respective correlations were 0.45 and 0.06. For males, the within-twin, cross sex and gender correlation (homosexual orientation and childhood gender non-conformity) was 0.57; for females, 0.33 (Bailey, Dunne and Martin, 2000).
As childhood gender nonconformity could be considered a genetically determined contributor to homosexuality, itself also genetically influenced, a test of this hypothesis could be examining the sexual orientation of relatives of homosexual persons with CGN. Although they should have been found to have a higher rate of homosexuality, this was not the case (Bailey and Pillard, 1991). If homosexual men who were gender nonconforming as children are more genetically homosexual, they should be those most likely to have homosexual co-twins. This should be the case whether they are MZ or DZ twin pairs. However, this too, was not confirmed. Homosexual twins with high CGN scores did not differ from those with low CGN scores as to whether their co-twin was homosexual. Further, for MZ concordant homosexual twin pairs, CGN scores were similar. By contrast, the CGN of homosexual DZ twins correlated poorly with that of heterosexual co-twins (Bailey et al., 1993).
A related strategy compared CGN between homosexual non-twin brothers. They report similar degrees; both brothers tended to be either feminine or masculine. This suggests that CGN may distinguish familial-genetic subtypes of male homosexuality (Daewood et al., 2000).
Socially Based Theories
Socialization forces are also offered in explanation for the association between CGN and adult homosexuality. If most of a child’s peer group adopts a specific socio-sexual pattern of romantic crushes and erotic behavior, socialization may proceed on the same track even if a child is a different sex to that peer group. So the cross-gender boy whose peer group is girls finds other boys to be “cute,” etc. A related explanation for the early socialisation effect on later sex life has a nice ring to it: “exotic becomes erotic” (see Bem, 1996; 2008). Genes or prenatal hormones code for childhood temperament such as aggression and activity level. These influence peer group composition as male or female. It is posited that children experience heightened non-specific physiological arousal in the presence of peers from whom they feel different. This is transformed into erotic attraction. Other factors than gender behaviors are also posited to effect same-sex attraction, such as physical disability, illness, or lack of contact with same-sex peers (Bem, 1996).
As with biological theories, these early cross-gender socialization theories do not explain later same-sex attraction in children with gender-conventional behaviors. Although the association between the two phenomena is possibly the highest of any two life point behaviors, it is not a perfect fit. Some conventionally gender-behaving children evolve as homosexual and some gender nonconformists evolve as heterosexual. A possible explanation for the former is that their “natural” behaviors are modified to societal conformity by pressures from parents and peers. Similarly, in adulthood, some persons primarily attracted to same-sex partners function heterosexually in response to family, peer and other societal forces. However, transsexualism remains an enigma even if the children were concealing their early “natural” tendency for cross-gender behavior. Adult transsexuals recall extensive cross-gender behavior in childhood, accompanied by discomfort in being a child of their birth sex. They attempt to remedy this discord in adulthood through surgical sex reassignment. However, about a third of male-to-female transsexuals and a tenth of female-to-male transsexuals in our Clinic are sexually attracted to partners of the opposite birth sex.
Psychodynamic theories have stressed the cross-sex parental identification of pre-homosexual males and females. Thus feminine, protohomosexual boys identify with mothers and masculine, proto-homosexual girls identity with fathers (Bieber et al., 1962; Green, 1974). If this is a powerful factor, then the later sexual desires of the child may reflect what was observed in the adult identity model. The processes of identification by children are complex. Some evidence points to a given degree of similarity between child and adult as facilitating identification (Kagan, 1958). Thus temperamental differences in infancy or early childhood may promote same-sex versus other-sex identifications that are the bases for gender-typed behaviors in later childhood and then sexual behaviors in adulthood.
Diversity of Outcome for Childhood Gender Nonconformity
Why don’t most cross-gendered boys become transsexual and not homosexual? The boyhood histories provided by adult male-to-female transsexuals inspired my long-term study of cross-gendered boys. These boys were behaving in ways recalled by my male adult transsexual patients. Consquently, Volume 1 of this study was titled “Sexual Identity Conflict in Children and Adults” (Green 1974). It juxtaposed different people at two life points. However, the title of Volume 2 reveals the evolution of the same boys: “The ‘Sissy Boy Syndrome’ and the Development of Homosexuality” (Green, 1987). Why the common outcome of transsexualism and homosexuality from childhood crossgender behavior? Perhaps some previously feminine boy homosexual males with continuing femininity who desire masculine heterosexual partners and are conflicted over homosexuality reinvent themselves as women to attract masculine men. This controversial developmental process is advanced by Bailey (2003).
Why weren’t more of the cross-gendered groups subsequently transsexual? One possibility is probability. The incidence of male homosexuality is about 3-4 in 100; the incidence of male-to-female transsexualism about 1 in 10,000 (Bakker et al., 1993). So, given a small sample of a few dozen people and the inability to distinguish pre-homosexual from pre-transsexual children, the chances of homosexual outcome are substantially greater. Another possibility is the influence of having been brought by parents for an assessment, and in some cases, intervention, for the cross-gender behavior. Recognition by parents of behaviors of concern to them, and attempts by parents and perhaps clinicians to modify the behaviors, may have resulted in their substantial reduction. This may have enabled these children to socialize as persons of their birth sex, without the need for sex change. Further, while parents and clinicians may think they know how to modify gender-typed social behaviors, they are probably clueless and helpless for how to modify later erotic behaviors.
A recent longitudinal study of boys with Gender Identity Disorder (GID) found more continuing with gender dysphoria into adolescence than the “Sissy Boy” study (Zucker, 2003). They appear on course for adult GID or transsexualism. Why should more recent series include more patients who remain gender dysphoric beyond childhood? The answer may reflect a cohort effect. Thirty-five years ago there was less tolerance of both adult homosexuality and childhood cross-gender behavior. The children assessed then may have been less extensively cross-gender behaving than those currently assessed because the threshold for clinical referral has risen. If so, children with greater gender dysphoria referred more recently might be expected to continue to be unhappy in their birth sex beyond childhood.
Two Twin Pairs
Studying monozygotic twin pairs discordant for gender identity or sexual orientation for their childhood gender-typed activities could help dissect the genetic contribution of the earlier and later behaviors. Thirty years ago, I initially reported a pair of male monozygotic twins discordant for boyhood gender-typed behaviors. As their mother described them, one “walked like a clod hopper,” and the other was “rather feminine and on the prissy side.” One associated with boys and was always outdoors; the other would play at home with his sister and her dolls. In “mother-father” games, one role-played as father; the other as mother or sister.
Earlier experiential factors could have influenced their disparate development of gender-typed behavior. The pre-feminine twin developed glandular tuberculosis involving axillary nodes. For two years he was driven by his mother to hospital for treatment while the co-twin was engaged in sports with his father. After the pre-feminine twin underwent axillary node dissection, the father “gave up” trying to have him throw a ball. At follow-up in their 20s, the previously feminine twin was bisexual, but predominantly homosexual. The previously masculine twin was bisexual, but predominantly heterosexual. Arguably, the genetic extent of their bisexuality was modified by the extent of boyhood gender-typed behaviors, not genetically determined factors (Green and Stoller, 1971; Green, 1987).
Also 30 years ago, Stoller reported a pair of female monozygotic twins where one was seeking sex reassignment. As children that twin had a masculine nickname, refused to wear dresses, excelled in sports and reportedly wished that she had been born male. The other was conventionally feminine. Although the early childhood health related experiences of the female twins were not as dramatically distinct as in the case of the male twins, the feminine twin had a minor orthopaedic foot defect requiring treatment for several years. As children, the larger, healthier twin was the masculine twin. As young adults, one was heterosexual and wanted to have children. One was sexually attracted only to females and wanted hormonal and surgical treatment to live as a man (Green and Stoller, 1971). This case argues for a non-genetic basis of both childhood gender-typed behavior and sexual identity and sexual orientation but maintains the link between childhood gender nonconformity and adulthood sexual orientation.
Political Considerations
What are the implications of the association between CGN and adult homosexuality? Controversy envelops the inclusion of Gender Identity Disorder in Children (GID) in the Diagnostic and Statistical Manual of Mental Disorders (DSM) (Bartlett, Vasey and Bukowski, 2000). Because GID is a disorder associated in most cases with a behavior in adulthood that is not a disorder (homosexuality), the argument goes that the childhood diagnosis is incoherent. More than incoherent, it is also argued that the diagnosis stigmatizes children whose gender behaviors do not conform to societal stereotypes.
For a set of behaviors to be included in DSM, they must be associated with subjective distress and social disadvantage. If the child has “gender dysphoria,” the subjective distress of being male or female, and is stigmatized by the peer group in consequence of the gender atypical behaviors, the criteria for DSM are met. If the gender dysphoria persists beyond childhood, transsexualism (or GID in adolescents/adults) is the condition. This is less controversial as a diagnosis, though not without advocates for its inclusion as a less stigmatizing neurological disorder, if disorder it must be. Where the gender dysphoria is no longer present years later, for whatever reason, intervention or otherwise, this may not mean that the earlier diagnosis was in error if the two criteria for inclusion remain. Gender dysphoria does not commence in adolescence or adulthood if the psychosexual histories given by teenagers and adults with GID are valid.
An argument for continuing GID in childhood in a diagnostic manual is that it may encourage intervention to interrupt the development of transsexualism. Living as a homosexual adult is generally easier than living as a transsexual. Further, the concern that the diagnosis may stigmatize a child may be responded to by the contention that careful adherence to the diagnostic elements of GID should not permit a false positive diagnosis. Androgynous children who are not gender dysphoric will not be labelled with GID.
Conclusion
The explanation for childhood cross-gender social behavior and adulthood homosexual behavior may merely be that each is the age appropriate expression of the same mysterious underlying phenomenon. Few 5 year olds are engaged in interpersonal genital sex. Even fewer 30-year-olds play with dolls or toy trucks. Whatever the theory, a fundamental obstacle to explanation is that much remains to be proven about the etiology of gender-typed social behaviors and much remains to be proven about the etiology of sexual behaviors. The consequence is conundrum.