Posted by: Chris | July 14, 2011

On the Origins of Homosexuality

As regulars of this blog might know, the biological origins of homosexuality* (and the evolutionary accounting for the same) are a popular topic in these parts.  My preferred hypothesis is a mix of neurological genes and prenatal hormones for the former and benevolent pleitropic and carrier effects for the latter, which also seems to be the dominant view of the relevant scientific community.  However, a post on the issue by Razib Khan and the comments therein got me following a mess of alternate biological explanations for the phenomenon, complete with their own discreet Darwinian explanations.  I did not emerge all too convinced by any, but I thought they were all interesting and deserving of wider circulation.

The first, which I had only encountered cursorily before, comes from Gregory Cochran, a biological anthropologist and former co-blogger with Razib when Gene Expression was over at Science Blogs.  One of Cochran’s preferred hypotheses is that many conditions we initially assumed to be genetic are in fact precipitated by viral infections, which are in turn made more likely by genetic factors.  He cites narcolepsy, which researchers suspect might be caused by an virally-triggered auto-immune disorder that attacks specific neuron involved in the production of hypocretins (a sleep-related neurotransmitter) and only seems to affect people with a specific HLA (a super locus of genes involved in self-recognition in cells).  The hypothesis is that homosexuality works similarly, with a viral infection and auto-immune irregularity wiping out, in a targeted fashion, neurons involved in proper orientation.  Cochran reasons this likely because its high prevalence does not resemble other deleterious genetic disorders:

 Most common and serious diseases that have been around a long time and hit in early life are caused by germs – bacteria, viruses, parasitic worms, etc. Evolution doesn’t necessarily make them rare, because evolution is playing on both sides in this struggle: they’re evolving too.    In much the same way, evolution doesn’t just make zebras faster, it makes the lions faster too.   Lions can continue to be a major problem for zebras over millions of years – and in the same way, malaria can continue to be a problem for humans indefinitely. 

So if a disease is common (> a tenth of one percent), hits in early life, has been around a long time (so we know it’s not caused by some new industrial chemical or whatever), and it’s not restricted to people from the malaria zone – it’s probably caused by some bug.  

 But what about homosexuality?   Well, from this biological perspective, it’s surely a disease. Disinterest in the opposite sex reduces reproduction quite a bit – around 80% in American conditions. Does it hit in early life?    Sure. Has it been around a long time?    Certainly.

It is an interesting hypothesis for sure, but I do not find the adaptionist rationale as compelling as Cochran does, largely because it has been amply established that defects with beneficial secondary effects on heterozygotes could stabilize in a population at percentages as high as five or ten percent, especially if thrust that high initially through drift.  What more, the additional evidence that Cochran musters to defend his claim are dubious at best.  Cochran implies that the pathogen jumped from sheep to humans during their domestication in Anatolia and spread throughout humanity from there, in a similar manner as pathogens like small pox.  Though this might account for the lamb-loving Greeks, it runs afoul of the prevalence of homosexuality found in many pre-agricultural societies.  Likewise, Cochran cites the differential between openly gay people in urban compared to rural environments, which he sees as reflective of a more conducive environment for a viral transfer, but to me seems to simply reflect differences in social mores.  Further, the auto-immune nature of the proposed disorder does not seem strong enough to account for the observed heritability of homosexuality.

The second explanation comes from an interested layman who became gradually disenchanted with Cochran’s hypothesis and groped about for a better explanation.  He seems to have settled on an explanation that hinges on chimerism as an explanation.  Chimeras are organisms who, during development, absorbed stem cell lines from a genetically distinct source, usually an undeveloped twin zygote or, occasionally, the mother.  This results in certain organs or parts of the body being genetically distinct from each other and could account, for instance, for people who have different colored eyes or ambiguous sexual organs.  This all seems quite esoteric and was assumed to be quite rare in humans (thought it is common in some species of mammals, like marmots).  However, recent research suggests that chimerism is far more common in humans than previously suspected.  The author suggests that this process could account for homosexuality as well:

 Some physiological differences have been found between heterosexual and homosexual men suggesting a slight, overall feminization of homosexual men, but the most compelling and obvious difference is behavior, which is neurologically determined. Humans are extraordinarily complex beings, and much of what we are is expressed by our minds. Our likes and dislikes, our tastes — both artistic and physical — are determined by our neurological makeup. Some of us are timid, some bold. Some are meticulous and some spontaneous. And, as it happens, some of us prefer women and others men. In most cases this can be determined by sex, i.e., most men prefer women and most women prefer men, but this obviously isn’t a hard-and-fast rule: some men clearly prefer men to women and some women prefer women to men…

It is likely that many stem cells, even from siblings, would eventually be rejected as foreign in a developing fetus, or would be weeded out when the child is born and begins to develop a mature immune system. However, certain tissues are exempt from this, including those in the brain. This means that stem cells that made their way into the incipient brain of an embryo could stay there safely, even if they are eliminated from other tissues. In fact, they could, and probably do, persist in the brain throughout the life of the organism. Female stem cells in a developing male infant’s brain could have profound implications for its ultimate form. Just as much of the rest of the body is shaped by our gender, so is the brain, and tests with rats have shown that hormones alone are not sufficient to masculinize the brain. The anatomical differences discovered in regards to homosexual brains, most of which show feminization, could be explained by pockets of genetically female cells, or even a few interspersed throughout areas critical to masculinization. Chimerism in the brain is probably more common than in other regions, and does not usually result in homosexuality, but can when cells colonize certain regions.

Another interesting theory, but it too stumbles while trying to explain some observed phenomena with regards to male homosexuality.  To me, the biggest stumbling block is in the numbers.  The author seems to have misinterpreted the growing consensus of chimerism as common to mean essentially a widespread condition, afflicting a majority of humanity.  He assumes (without support) that chimerism occurs in between 50-70% of all pregnancies.  However, as the Cochrane article previously cited makes clear, when scientists and medical researchers use the word common, they are referring to something slightly different from the average person.  Cochrane defines it as anything occurring in a population at a rate greater than 1 in 1000.  The best I could tell, the current estimates for chimerism in humans as a whole run between 1% to 8%, with the latter number representing a clear upper bound of situations where chimeras could arise (multiple pregnancies resulting in a single birth).  However, remember that the author is arguing that one specific chimeric variant, which replaces set neurons in a certain hypothalmic nucleus, accounts for observed homosexual behavior.  Even given the most generous estimates, this variant cannot account for more than perhaps 1% of extant human chimeras (think of the numerous possibilities for just neurological replacements).  There is no way that this .01%-.08% of the population  can account for the roughly 3-8% of males that are gay. 

There are further problems with this hypothesis.  For instance, the author accounts for the observed hereditary element of homosexuality thusly:

 Homosexuality has high twin concordance, especially in monozygotic (identical) twins, where it approaches 50%. In fraternal twins it significantly lower, but still high, at 22%. Aside from genetic similarity, one of the factors that sets identical and fraternal twins apart is that most identical twin pregnancies are monochorionic, meaning that they share a placenta, whereas most fraternal twins are dichorionic, meaning that they each have their own placenta. Twins that share a placenta are naturally more likely to receive similar amounts of stem cells if any happen to be circulating through the placenta. Twins with separate placentas are still likely to be exposed to the cells, since they are both in the same uterus. However, each one has a distinct supply of nutrients and cells, so the ratio each twin receives would tend to vary more than in the case of identical twins.

Yes, “aside from genetic similarity.”  However, experience in other studies suggests that when two people with the same genes display the same behavior, the most parsimonious explanation is that the genes at least play a role.  I will also note that the presence of gay identical twins also offers a bit of a stumbling block for this theory as well, unless there was a third fraternal sister that was absorbed equally and similarly (i.e. replacing the same portion of the brain) by both twins, something that strikes me as highly unlikely.  His Darwinian logic is also a bit screwy too.  He argues that this arrangement would be evolutionarily advantageous because:

Fertility is obviously a genetically positive trait, and fertility is the result of eggs being both available and fertile. This means that more fertile eggs during ovulation would be genetically advantageous, but due to human child-rearing logistics the survival of only one fertilized embryo would also be advantageous. Therefore, women prone to having more than one egg fertilized, but whose pregnancies only resulted in only one live birth, would have the optimum level of fertility. A side effect of this could be an increased incidence of chimerism in human children, and in the case of multiple embryos with different genders, this could lead to physiological changes that result in homosexuality in a proportion of the population.

Generally speaking, for much of human history, higher child-per-mother rates period is considered the maximally effective strategy, since, given the vagaries of pre-industrial life, you never know which children will make it to reproduction and it is better to have more kids than to play the odds with fewer.  I certainly cannot conceive of a strategy that produces fewer children and does not provide them with any additional benefit in the process (instead saddling a few with some debilitating side effects) as being more effective than successful multiple pregnancies, when they do occur.  Instead, I would argue chimerism seems to simply be , from an evolutionary point of view, a neutral to mildly deleterious side effect of other processes, like the early elimination of unviable zygotes or the natural sloughing off of cells in the uterine environment.  This, especially when combined with the numerical issues involved in chimerical homosexuality, makes me somewhat dubious of this explanation.  However, since it does center on a little understood topic and is quite easily falsifiable, I would not dismiss it out of hand.  A quick experiment looking at slices of gay vs. straight hypothalmus neurons for double X’s or other signs of foreign female cells would settle the issue far better than supposition about numbers.

The final hypothesis comes from Gary Larden, who is, like Cochrane, a biological anthropologist.  He argues that early childhood stimuli and environment have a heavy influence in sexual orientation and gender, as much if not more so than prenatal or genetic factors.  This, he writes, leads to a wide spectrum of observed sexual orientations and genders:

If every single factor is thought of as a simple binary choice (and I use the word “choice” with no reference to human decision making) between two cannalized options, then the number of possible outcomes could be thought of as 2n where ‘n’ is the number of times a binary choice is encountered. So, if there are, say, three hormonal moments in utero, and one more after birth (puberty) and, say, three life stages that have major influences on gender (and I oversimplify) then the number of possible routes a person may take from conception to adulthood would be 27. That is 128. If these different paths lead to mostly different outcomes, wouldn’t there be over 100 “genders” among humans?

Amongst other things, Larden seems to neglect that, in development, upstream inputs have a far more significant influence on outcomes than those further downstream.  Thus, his conjectural use of the Fundamental Counting Principle** is highly misleading: even if all seven of the hypothetical events have an effect on sexual orientation (which is questionable), the genetic makeup and prenatal environment would slant things disproportionately one way or another and later events would likely simply add fuzziness or ambiguity, rather than shunting it to another category.  Larden’s hypothesis also seems disconnected from the evidentiary support he cites.  He, for example, notes differences in how parents treat infants or how school-age girls and boys socialize and self-segregate differently, but does not demonstrate that these two phenomena are correlated (not to mention causal).  Instead, he seems to retreat to a more nuanced version of blank slate-ism, which cases like David Reimer’s do not seem to corroborate, which only reinforces Razib’s original point about the creeping convergence between the political correct left and the religious right on these issues.

*I should note that almost all mentions of homosexuality here and on the linked to articles refers to male homosexuality, which is better understood and seems less transient and more heritable than female homosexuality.

**See if you can guess which Lurer teaches math…



  1. “it has been amply established that defects with beneficial secondary effects on heterozygotes could stabilize in a population at percentages as high as five or ten percent, especially if thrust that high initially through drift.”

    Name one example.

    • Gregory,

      First, thanks for commenting. Its always good to read pushback for things we launch into the void of the internet.

      What I specifically had in mind when I wrote that passage were various papers outlining how homosexuality could remain at stable levels, even under strong negative selective pressure, using a variety of genetic models. A quick Google search, for instance, brought up these two papers on the front page:

      Further, I do not think it necessarily all that controversial to assume this to be, at minimum, possible, especially if you assume:

      a) a modest heterozygote advantage
      b) a lack of expression of the negative phenotype in affected individuals due to deficiency in other triggers (like gene-gene or gene-environmental interactions), presumably retaining the pleiotropic benefits of having those genes
      c) a sexually skewed expression of the phenotype, relative to the allele frequencies (assuming only males display male homosexuality), which makes it even harder for selection to weed out the deleterious alleles
      d) a blunting of the negative effects of the phenotype due to either taboo (or other social factors) or marginal kin selection benefits (a la the gay uncle theory)

      However, this is, in a sense, purely speculative and you asked for examples. Again, I am not sure the idea of antagonistic pleiotropy is all that controversial. After all, its essential to understanding the evolution of everything from sex to senescense and those are not just defects that remained persistently resistant to natural selection, but ones that rapidly and early on rose to fixation and have become pervasive across most complex life.

      We see this in genetically influenced diseases as well. I know you noted, in your article, that, even in the most malaria-infected parts of Africa, only one out of thirty individuals is homozygous for sickle-celll anemia, despite having a hugely important heterozygote advantage. Thus it might seem unlikely that other genetically linked dysfunctions could affect as much as five or ten percent of the population. However, looking solely at current allele or phenotypic ratios gives an incomplete picture. You have to also incorporate some sense of change from an initial state. If homosexuality is at 5 percent currently, but its frequency has been slowly chipped away at for hundreds of generations by relentless selection while the 10/3 percent with sickle cell represents the influence of a recent mutant on the rise, I think a different picture is drawn

      More importantly, though, the sickle cell allele only has one thing going for it: the heterozygote advantage (or “a” from above). However, we have seen deleterious genes (or at least their detrimental phenotypes) reach quite high frequencies in human populations due to some combination of b-d. Many maladies, from autism to asthma to various cancers, seem strongly hereditary (and, thus, likely genetically-linked) but, like homosexuality and unlike sickle cell anemia, are quite non-Mendelian in their expression. They are all speculated to involve a complex genetic origin , influenced by polymorphisms at a number of locii (whereas sickle cell results from a single base pair mutation in a single gene, and thus its expression can be modelled by any eigth-grader with a Punnett square). They also are presumed to involve environmental triggers of varying number and influence, which sickle cell also lacks. All this aids the negative alleles in escaping selection’s clumsy grasp, and, as a result, all are expressed at ratios equal to or greater than sickle cell in the malarial thicket, despite being clearly deleterious and having no known advantage for carriers. Disadvantageous alleles need not even have any masking to penetrate deeply into a population. For instance, red-green color-blindness and left-handedness are expressed by roughly 10% of the population, due in part to having a blunted and mildly negative effect and a skewed sex-ratio (c-d, from above). Finally, this all supposes a specious view of selection that assumes it is so overwhelming that it trims off all fitness-reducing excess with the upmost accuracy, ignoring the role that drift, bottlenecks, and pure chance probably played in the history of these genes.

      Thus, based on both modelling specific to homosexuality and evidence from other genetically-linked disadvantageous conditions, I think my original, quoted statement retains some validity. However, since you’ve responded to the post. I would be remiss not to ask what your opinion of other objections that either my post or others have raised against the hypothesis you floated? I actually found your thesis to be quite interesting, though not overwhelmingly convincing, and I do not dismiss it out of hand (as I hope my post made clear). Actually, it seems immenently falsifiable, which always a great attribute for theories to have, especially in the typically highly speculative world of human ethology. Would you agree, for instance, that. if your hypothesis were true, that you would see a greater than chance clustering of cases of homosexuality of those that share an age cohort and either a place of birth/neighborhood/school (depending on your preferred model for the viral vector)? Or would there be a better way to quickly test the thesis?

      This is of course the great thing about scientific debates: they can have some semblance of progress and certain explanations elevated over others (with some Quine-Duhem related caveats). Homosexuality remains an intriuging Darwinian puzzle and probably will for some time, but a continued investigation will only shed more light.

  2. Red-green color blindness is unusual: the genes are adjacent, and illegitimate recombination is frequent. Effectively, they have an enormous mutation rate. Moreover, the effect of red-green color blindness is not severe, at least not in farming peoples (as opposed to hunter-gatherers).
    It is of course much rarer in hunter-gatherers – around 1% in Pygmies.

    When I say “not severe” i mean that there is no sign that the fitness disadvantage of colorblindness in that situation is even as large as 1%. Whereas that of homosexuality is more like 80%.

    You have to consider both the frequency and the negative selective effect of homosexuality. In fact, the simple way of taking them both into account is just multiplying them together – which produces the ‘fitness hit’, a measure of the selective pressure against that syndrome. The fitness hit of homosexuality is far higher than that of any narrow syndrome: it can only be compared with the sum total of all known genetic disease due to mutational pressure. Not counting malaria defenses in some regions, the sum total of all serious Mendelian disease only hits something like 1% of the population. A genetically based syndrome like that would not take hundreds of generations to become much rarer – more like 5.

    Read up on quantitative selection. Run some simulations. Read about the results of selection experiments. Whatever gave you the idea that selection is clumsy and slow? More exactly, _whoever_ gave you that impression?

    Consider cystic fibrosis, the most common Mendelian genetic disease in the general European population. It has a gene frequency of about 2%, and about 1 in 2500 children are born with cystic fibrosis, a condition that uniformly lethal in the old days and very unpleasant even today. You might think that is an example of the clumsiness of natural selection. Not so: without a noticeable advantage to carriers, that result is _impossible_. Simple simulations show that without a heterozygote advantage (of about 2%), CF cannot possibly exist its current frequency. If you back-project the frequency over time, which is wholly deterministic because of Europe’s large population over that period, the gene frequency would have been 10% 1000 years ago, while _everybody_ would have been a CF carrier 1250 years ago. You can’t invoke bottlenecks: to have the kind of effect you need a bottleneck that took the population down to _tens_ of people, and I hardly think that happened over the last 1500 years in Europe.

    I think you’re thinking of marmosets, rather than marmots.

    Identical concordance for homosexuality is more like 25%, not 50%.

    The chimera theory is unlikely because microchimerism has been around since the start of placental mammals, maybe earlier. Crazy adaptationist that I am, I figure that’s long enough for natural selection to find a solution. It makes no sense in the same way that the old explanations of duodenal ulcers (all that acid in the stomach is bound to cause problems) made no sense.

    Human homosexuality is like 5% of a male population refusing to eat, or furiously poking themselves in the eye with a sharp stick. it is an anomaly.

  3. Gregory,

    Apologies for the delay.

    First, I think you vastly overestimate the historical fitness cost of homosexuality by quite a lot. Yoiu keep citing the 80% number (which I agree, would be essentially impossible to counteract), but that number comes from this section of your original article:

    “Disinterest in the opposite sex reduces reproduction quite a bit – around 80% in American conditions.”

    I think we both can agree that current American conditions are somewhat of a deviation from the norm. But by how much? I am not aware of any statistics that tracked how effectively the taboo against open homosexuality kept gays reproducing (and we can’t exactly go back and measure), so I think we are stuck using anecdotal evidence. For instance, Oscar Wilde had three children, and Walt Whitman claimed to have six. Thus, I would not be surprised, especially if you factor in some marginal positive considerations (like the gay uncle theory) that the actual fitness cost of homosexuality is as much as an order of magnitude lower than you suggest.

    Tangentially, this would suggest that the period between the lifting of the taboo until surrogacy becomes more widely available should severely lessen the prevalence of homosexuality. We shall see if that bears out.

    As for the identical twin numbers, I am not sure where you got your numbers, but mine came from here:

    The 50% MZ concordance also conforms to that seen with left-handedness and CCW hair whorls (which in turn track with male homosexuality) so I am inclined to give it extra credence.

    However, I think the biggest problem seems to be that you are still conceiving homosexuality in simplistic Mendelian terms. As I have previously stated, sickle cell anemia (and cystic fibrosis for that matter) are caused by a single mutation at a single locus, expressed recessively. Thus, if you have two copies of the defective gene, you will almost certainly express the genetic disorder. This makes the alleles associated with the defect quite easy for selection to “spot” and to select against, because their presence tracks regularly with the defect’s expression

    Homosexuality, on the other hand, does not seem to be expressed similarly, otherwise we would observe near 100% twin concordance. This is also why it is unlikely we could clone a gay sheep (because, though the genes may be the same, the prenatal environment would be different). Thus, one must assume (if it is genetically-linked) that homosexuality is caused by both being homozygous for the recessive, defective allele (or alleles, if it is influenced by multiple loci) and having some epigenetic or environmental trigger. The sum effect, from an evolutionary perspective, of requiring either gene-gene or gene-environment interactions to produce phenotypic homosexuality is a dilution of the fitness cost of the defect to the responsible genes, causing selection to be somewhat less than ruthless in their elimination. I gave some examples of this phenomenon in my original response and, as far as I can tell, its prevalence is fairly widely accepted amongst geneticists and has been for sometime. I am not quite sure who/what gave you the idea that all or even most traits are expressed purely as Mendel described.

    Now, granted, even in this case, as generations -> infinity, the expectation is that the responsible allele would still be removed from the gene pool, unless there was another effect weakly buoying its fitness. This is where the gay-gene proponents’ necessary epicycle, either a heterozygote advantage or some pleiotropic benefit (remember, those who are homozygous recessive but whose expression is masked for a variety of reasons can still benefit from other effects of the gene), comes in. Some theories have been floated and some preliminary results returned (like the recent increased female fertility observed in an Italian study), but I have yet to encounter something wholly convincing. However, as a sum total, the gay gene plus theory holds together pretty well, especially in comparison to its competitors.

    This brings me to my final point. I hate to be a pedant here, but scientific hypotheses are not judged in a vacuum. Arguments in favor of the genetic underpinnings of homosexuality certainly have some stumbling blocks and make suppositions about ambiguous or poorly understood topics, but they also account for the totality of the evidence, atleast in my view, far better than the gay germ hypothesis (or the chimera or nuture ones, for that matter). Which was why I ended my previous post asking if you had responses to the criticisms of the gay germ theory, either those related to issues of chronology or how it would explain observed twin concordance or the birth order effect or the linkage between homosexuality and many physical traits (which would presumably be locked into place prior to any purported infection). Arguments (or, rather, argument) from incredulity only gets one so far in science; eventually one has to defend his ideas as well.

  4. […] finding might also shed light on the abnormal prevalence of color-blindedness in humans: a mix of unusual inheritance and carrier effects washing out the […]

  5. […] details that proponents of pleiotropic explanations usually cannot (and is certainly superior to the germ of the gaps and other fringier explanations).  I am really interested to see what the researchers and others […]

  6. Bailey, Dunne, and Martin found a 20% concordance of homosexuality in MZ twins in their Australian study. From Wiki:

    “A number of twin studies have attempted to compare the relative importance of genetics and environment in the determination of sexual orientation. In a 1991 study, Bailey and Pillard found that 52% of monozygotic (MZ) brothers and 22% of the dizygotic (DZ) twins were concordant for homosexuality.[4] ‘MZ’ indicates identical twins with the same sets of genes and ‘DZ’ indicates fraternal twins where genes are mixed to a similar extent as non-twin siblings. In 2000, Bailey, Dunne and Martin found similar results from a larger sample of 4,901 Australian twins.[5] Self reported zygosity, sexual attraction, fantasy and behaviours were assessed by questionnaire and zygosity was serologically checked when in doubt. They found 20% concordance in the male identical or MZ twins and 24% concordance for the female identical or MZ twins. A meta-study by Hershberger (2001)[6]compares the results of eight different twin studies: among those, all but two showed MZ twins having much higher concordance of sexual orientation than DZ twins, suggesting a non-negligible genetic component.”

    Bailey himself has said the Australian study should be considered the more reliable study (compared to the earlier Bailey/Pillard ’91 sample.)

  7. […] to take up his challenge to rebut the previous deBoer.  I know we’ve been down this road before here, but I want to state the case as clearly and as jargon-free as possible, so that future […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s


%d bloggers like this: