Editors: Norio Fujiki, M.D. & Darryl R.J. Macer, Ph.D.
Associate Clinical Professor of Medicine, Stanford University, Palo Alto, California 94304, USA
This paper is a revision of another which was delivered at a symposium organised by the American Association for the Advancement of Sciences in Boston (1993). The current changes are minor. In response to that paper, a prominent scientist in the field of behavioural genetics, Dr. Dean Hamer, wrote to several of my colleagues suggesting that some participants in that event considered my comments "homophobic". He further suggested that this fact would limit my ability to engage in this field's activities (Beckwith, personal communication 1993).
This unfortunate incident may be illustrative of important issues which the scientific study of behaviour pose. Science reflects the environment in which it is conducted and the interests of those who do it. Scientific progress arises as fact and interpretation develop out of insight, data and analytic conflict. Biomedical research and behavioural studies, because they can be so personally relevant to those conducting this research, may experience frequent distortions of normal processes especially in the absence of credible data. Though all scientific information can be used in what are essentially political processes, character assassination and "political correctness" need not be part of the analysis and review of any aspect of biomedical research.
Genes are involved in the development and functioning of those organs and tissues which mediate and participate in human behaviour. Though this is self evident, it is also supported by experimental data showing the heritability of human traits, model neurodevelopmental systems which demonstrate the importance of single genetic mutations, and clinical conditions like PKU and Huntington disease where single gene alterations are frequently associated with mental retardation and psychiatric sequelae.
Unfortunately, the insight that genetics provides in behavioural analysis stops abruptly at this point. Duchenne muscular dystrophy, an X-linked single gene disorder, is associated in 20% of cases with behavioural and mental findings in affected boys. Genetic and biological studies of dystrophin, the protein altered in this disorder, have not explained the cases with behavioural abnormalities. This failure in explanatory power of single gene analysis may be amplified when genetic methods are applied to the study of normal behaviour and common behavioural disorders. In these situations, phenocopies and phenotypic heterogeneity are frequent. The genetic component of observed variance is also heterogeneous. In fact, such traits are influenced by multiple genes with redundant functions which are environmentally modified, and interact in complex cascades over developmental time.
To demonstrate the difficulties of applying genetic methods and thinking to complex human behavioural traits, this paper will focus on the study of variation in human sexual orientation. This will be a critical analysis of published research on the biological basis of homosexuality. Though it will argue that the work of laboratories headed by Drs. Simon Levay, Roger Gorski, Michael Bailey, Richard Pillard, Dean Hamer and others have not established the genetic causation of sexual preference (or orientation), the work that these researchers have contributed is a substantial improvement in the field of behavioural genetics. These scientists have carefully pointed out the limitations and problems with their own research. This has not often been true in the genetic analysis of behaviour in the past. A general analysis of the field of behavioural genetics has been published previously by my collaborators and myself (1). We have also recently commented on the controversy surrounding genes and sexual orientation specifically (2).
This paper will make both general and specific criticisms of the types of data which are now being used to justify the "genomic search" for a gene or genes involved in sexual preference. It will place this information in the historical and scientific traditions from which it arises. It is clear that the fervour of the researchers and the "sexiness" of the subject, not the rigour of the data nor the probability of finding a major gene, is motivating the current genetic research. In addition, this essay will pose the question, "why now?"; why is this research being conducted at this moment in the United States and receiving such widespread attention? Finally, several suggestions will be offered to improve the scientific outcome of the study of behavioural genetics.
The field of behavioural genetics, and the specific study of the genetics of sexual orientation, has been persistently confronted by its inability to precisely define the traits being studied. Homosexuality is clearly not a single entity. It is, for instance, true that until the 1970s homosexuality was defined as a "disease" for medical and psychiatric purposes in the United States. It is still treated as if it were an illness in regions of the United States and other countries.
Should homosexuality, for research purposes, be characterized as one or more same sex sexual encounters? Or the absolute absence of heterosexuality? The presence of homo-erotic fantasies? Or a specific type of sexual activity at a specific age? An additional complication affecting the definition is its susceptibility to environmental and social responses. Is the expression, and thus characterization, of sexual preference the same in San Francisco and Salt Lake City, where social responses to homosexual individuals may be quite different? Finally, though traditionally genetics has been most successful in explaining dichotomous traits, sexual orientation is a continuous characteristic of human populations. Males and females can be defined as heterosexual, homosexual, bisexual, asexual or otherwise. The range of behaviours within any two groups created for research purposes will either reflect selection (and thus not be representative), or will overlap substantially.
Evidence for major gene effects will likely only be described in groups created by scientists or specifically to find them. Because of this bias towards studying cohorts of limited diversity, the explanatory power of genetic insights will be lessened. Thus, it may be impossible to conduct research on homosexuality using genetic methods, or to genetically analyze any human characteristic, when the studied traits cannot be reliably ascertained in a large number of individuals, across a broad range of environments.
The published experiments from the laboratories of Simon Levay and Roger Gorski attempt to correlate brain structural changes or polymorphisms to sexual preference designations (2). These approaches are not new and involve a small number of subjects. The samples analyzed were confounded by pathology associated with HIV infection, and their retrospective study design raises serious questions about whether the designation of sexual orientation was correct. Various areas of the brain have been implicated. They imply that structural variation, and its genetic causation, determine subsequent sexual orientation. The genetic methods used cannot establish this as fact.
Whether reviewing data on brain structural polymorphisms such as these laboratories have described or DNA polymorphisms (see ref. 3) in relation to human behaviour, the tradition of phrenology must be remembered. Phrenologists attempted to correlate skull polymorphisms with a variety of behaviours. The biases of these researchers as well as the ambient prejudices of the 19th century obscured the absence of any evidence suggesting that skull differences caused the behaviours being studied. The current research under review may represent a drift towards a modern, possibly molecular phrenology.
The study of twins published by Michael Bailey and Richard Pillard, like most of the work using this classic genetic method, has problems related to how the sample was recruited. The twins were not randomly ascertained, and thus again represent a selected view of homosexuality. In addition, the findings of approximately 10% concordance in adopted or non-twin siblings, 20-25% concordance in dizygotic twins, and 50% concordance in monozygotic twins can be interpreted as showing that twins experience similar environmental influences more than non-twin siblings. This effect may be stronger on monozygotic twins as a result of their physical similarity, responses to their "identity", and/or other factors. To interpret this data as an indication of significant genetic effects, and use it as a justification for a search using DNA methods for related genes, is wishful.
Studies using twins have repeatedly been shown to harbour biases (1). Experiments using this genetic method have been hard to reproduce. In addition, approaches using twins can not locate genes. And, twin studies can never resolve the critical issue surrounding whether 1, a 100, or a 1,000 genes may be involved (directly or indirectly) in the trait under study. Their use in pronouncing that major genes affect behavioural phenomena has persisted despite the many methodological limitations and the noticeable failure to find major genes producing these traits.
Therefore, neither structural studies nor twin methods have provided firm evidence for the genetics of sexual orientation. Given this situation and in the absence of a testable hypothesis concerning how genes might determine this trait, a genomic search for such a gene, using a mapping method will likely produce falsely positive findings. This reflects simple Bayesian reasoning; the prior probability of a major gene causing homosexuality is exceptionally low. Falsely positive identifications of genes have been the result when localization approaches have been applied to the study of manic depressive illness, schizophrenia and alcoholism (1).
If significance using linkage or positional techniques is defined at 95% confidence limits and LOD scores of 3, then 5% of positive findings will represent falsely positive linkages or identifications. The linkage method, utilized in this way, improves the probability of linkage of a single gene from approximately 1 in 50 to 1 in 1000. This has prompted my colleagues and I to advocate a higher standard of LOD score significance in behavioural studies, particularly when there is no tenable hypothesis (or testable functional model) explaining the genetic regulation of a particular behavioural trait (1). If no real data direct the search for genes underlying a definable trait, and a random search is made, the current unequal distribution of testable sites across human chromosomal material make X-linked probes most likely to be falsely linked to the phenomenon under study.
The recent paper by Dean Hamer's group reflects progress in this field, in that it employs verifiable and reproducible molecular techniques and is appropriately circumspect in its conclusion and discussion (3). Unfortunately, though it is the best this area of study has to offer, these experiments remain seriously flawed.
First, these experiments have not solved the phenotype problem. In both parts of their study (a population and sib pair analysis), these investigators are assaying a selected population of gay males. In fact, there may be differences between the experimental groups. In addition, they employ a sib pair analysis method which can only associate regions of DNA or candidate genes with traits. Employing such a method in the analysis of characteristics where the prior probability of finding a major causal gene is low, is exceptionally unlikely to yield a true positive result.
Such a method might, for instance, be applied to familial forms of arthritis. The result could be an association with the HLA region of chromosome 6. Such a finding would not reveal what caused the arthritis nor have significant predictive power.
The power of Hamer's population study of homosexuality is weak. Therefore, their assumption that a significant gene resides on the X chromosome is ad hoc and biased. Their data only suggest that in a subgroup of those who identify themselves as gay, there may be genes on autosomes and X involved. The probability of any single positive association representing a falsely positive finding is high.
They then create a sib pair study design which eliminates any possibility of detecting falsely positive associations between X-linked DNA segments and the sexual orientation trait. Hamer's study neither tested the non-gay siblings in the maternally transmitted kinships nor included paternally transmitted subjects who should show no X-linked association.
Thus, a highly biased study group was created, multiple sensitive DNA tests applied and a positive result reported. This finding is unlikely to reveal a major gene with significant predictive value, may be a falsely positive association, and thus only represents progress in this field by focusing attention on an X-linked region. This site will likely be eliminated as the location for the "gay gene" by further experimentation, conducted on different subjects, by other interested researchers.
Since many of these methodological criticisms of research on sexual behaviour are well known, the question may be posed as to why this work is occurring now, and is so newsworthy? Is this field worthy of valuable research time and federal grant support? Several points seem relevant to such a line of inquiry.
First, scientific activity embodies a universal hope of understanding and sometimes controlling phenomena. In addition, the indomitable curiosity of scientists drives the investigative enterprise. Many dedicated individuals have invested long careers, and their hopes for professional success or academic promotion, in the frustrating analysis of behaviour and its origins.
Second, as new paradigms are entertained or accepted, the scientific analysis of any relevant field may be renewed. The currently dominant model of analysis in biomedical science is molecular genetics -- reviewing all fields through its prism will occur. This development is important to human geneticists and to those selling genetic information and products. Whether all the subjects being studied in this manner will benefit from the reductionist approach which genetic methods necessitate, and avoid the problems of misunderstanding which genetic methodology can bring, is uncertain. Even complex analytic approaches like those suggested by Eric Lander and David Botstein do not alleviate many of these long standing concerns (1).
Third, social, political and economic changes in the United States have encouraged the notion that the origin of difference is biological or genetic. These alterations may also explain why homosexuality is the trait under discussion, not heterosexuality. By promoting the idea that human traits are genetic and amenable to scientific analysis--ideas and approaches which emphasize difference--the determinative nature of such traits is highlighted, and social or political parties are relieved of responsibility. The outcome, which is certainly favoured by some (consider Pat Buchanan), is unequal treatment of groups justified by their biological variation, and the sociopolitical absolution of injustice, ignorance, fear or social/political failures as causal. Though biological difference may be easier to protect under current civil rights law, these arguments may have unforeseen pitfalls (4). The enforcement of civil rights in the absence of political will, understanding or tolerance within society is unlikely.
Finally, the suffering and costs associated with the AIDS epidemic fuel not only a desperate search for cures for this wretched illness but also new attempts at understanding the communities that are being devastated. Not only is there a marshaling of the hope associated with the scientific enterprise, but simple coping can be a motivation fostering investigations into homosexuality, a condition at risk for AIDS. This apparent psychological adaptation to the premature death of legions of individuals may be influencing the interest of scientists and other individuals both within and outside the gay community in "studying" and "understanding" homosexuality.
With these comments in mind, the following suggestions are offered in order to improve the outcome of the genetic analysis of behaviour. Generally, research in behavioural genetics and specifically sexual orientation must meet high standards of scientific responsibility. The limitations of the methods used (particularly their inability to indicate causality), the preliminary nature of early reports which have a high likelihood of being false, and the absence of testable models to study genetic findings must always be emphasized. Over-selling of new findings by scientists, university publicists and reporters must be discouraged by the scientists first. In addition, scientific responsibility dictates the researchers become aware of certain facts which affect their enterprise:
1) Genetic discrimination. Though it is argued that research in behavioural genetics will remove stigma, it is more likely to simply replace one form of prejudice and adverse consequences with another. Individuals labelled with genetic information in this society suffer a range of unfortunate discrimination (5).
2) This is not the first "genetic age". Earlier in this century, genetics was exceptionally popular in many countries. During this period behavioural genetics was studied using methods involving twins and heritability calculations. Results of this research were the German Eugenic Hygiene Laws and the legal sterilization of approximately 60,000 people for such traits as "imbecility" and "deviance" through the 1960s in the United States. The causes and forces which led to these outcomes are largely unknown. It can not be excluded that such results might reoccur or become manifest in other, possibly more modern ways. Eugenics remains overtly practiced in many countries today by groups who have political power and governmental license.
3) Conflict of interest. Given the recent rhetoric about life style preference and family values at the last Republican National Convention, the vote limiting gay and lesbian rights in Colorado, and the reiteration of overtly discriminatory hiring practices by the Armed Forces (a Federal Agency), there exists a distinct possibility of misuse of genetic research concerning sexual behaviour particularly if it concentrates, as it currently stands, on non-heterosexual orientation. Researchers conducting this work with federal support may therefore be directly or inadvertently aiding governmental programs including discriminatory practices. Until such conditions and practices have ended, researchers in behavioural genetics should consider: not accepting federal grants; returning currently allocated funds; and declining support from foundations with a history of using biological data to promote views and practices leading to unequal treatment of any group.
4) The scientific enterprise is international. The thin veneer of equal protection under law is barely operation in the United States and certainly should not be assumed to extend further than its borders. It is only in recent times that Afro-Americans and women have gained some measure of legal protection in our nation. Research data produced in the United States is influential in Germany where sexual behaviour variation has been "treated" by psychosurgery and hormone injections. Behavioural genetic research conducted in the United States is noted in legislative proceedings in Malaysia and Singapore in order to justify eugenic social policies. Electro-convulsive therapy is used on homosexual men in China, a practice which behavioural genetic research may indirectly support. The embargoing of research results and ending scientific collaborations should be considered until adverse practices arising from defining a normal, human trait as a "biological entity", a "disease" or "illness" ceases.
The political philosopher, J.S. Mills commented on behavioural genetics when he considered the possibility of intelligence differences between men and women. At that time, he noted that such comparative work could not be reliably done until several generations of equal education of women had been accomplished (See ref. 6). The study of behaviour and in particular, sexual orientation, may yield insightful and important findings. But, well meaning scientists curious about the biology of sexual behaviour should first dedicate themselves to equal protection, and the right to access goods and entitlements, for all individuals, irrespective of sexual tastes. They must do this in order to guard against the misuse of their research work under current national and international conditions. It is not acceptable to ignore or minimize the tawdry history of this field. Fairness and equal rights are matters of scientific responsibility, justice and morality, not genetics.
1. Billings, P. et al. (1992) Social Science Medicine, The Genetic Analysis of Human Behavior: A New Era? Social Science & Medicine 35: 227-238.
2. Billings, P. & Beckwith, J. (1993) Born Gay?Technology Review 96: 60-62.
3. Hamer, D. et al. (1993) A Linkage between DNA Markers on the X Chromosome and Male Sexual Orientation. Science 261: 321-3.
4. Halley, J. (1994) Genetics, Homosexuality and Civil Rights. Stanford Law Review (In Press).
5. Billings, P. et al. (1992) Discrimination as a Consequence of Genetic Testing. American Journal of Human Genetics 50: 476-482.
6. Sterling, A.F. Myths of Gender: Biological Theories About Women and Men (Basic Books, 1992).