Thursday, May 13, 2010

T-shirt Named Desire?

An example of an hypothesis cited enough times that it has become accepted lore suggests that one criterion upon which humans select mates is dissimilarity at the major histocompatibility complex (MHC) of genes. The MHC is a region of immune genes that vary greatly within species that have them (all jawed vertebrates), used in self/non-self recognition (and discovered because they're involved in tissue rejection in transplant surgery). The mate-selection idea is that offspring will be better equipped to combat infectious diseases with a broader spectrum of immune genes, and to ensure that, natural selection has built MHC dissimilarity into the way we choose mates.

The association between MHC dissimilarity and mate choice has been found at least in mice, as well as in humans. An apparent correlation between body odor and MHC genes (a correlation that is difficult to explain, but may have to do with the clustering of olfactory receptor genes near the MHC) has led to studies in which tests (in humans) were done by asking females to smell T-shirts worn by males for 2 nights, and to state which are more pleasant, and indeed which reminded them of current or former mates.

In one of the seminal studies (no pun intended?) on this subject, females not on oral contraceptives chose males with a very dissimilar MHC to their own, while women on the pill chose males that were much more similar to themselves. In another study, the preference for dissimilarity wasn't found in single women. Even so, this work has led to the founding of a company that will help you to find your perfect match -- based on science. (Though, if it can really lead you to your perfect mate though, why are they charging $1,995.95 for a lifetime membership -- shouldn't you need their services only once?)

But anyway a new paper in PLoS Genetics by Derti et al. refutes all this.
A recent study by Chaix et al. sought signs of mate selection in the genotypes of HapMap Phase 2 (Hap2) parents by comparing the genetic relatedness of mated and unmated opposite-sex couples, both at the MHC locus and overall (using all common autosomal variants). Yoruban mates (N = 27 couples) were reported to be slightly more similar than expected (nominal two-sided P<0.001) overall, but no significant difference in MHC relatedness was detected between mates and non-mates. By contrast, European-American mates (N = 28 couples) did not differ significantly from non-mates in autosomal relatedness, but were less similar at the MHC locus than were non-mates (P = 0.015).
The latter result was interpreted as supporting a role for the MHC locus in mate choice, and outliers were excluded as an explanation. However, a visual comparison of mate and non-mate pairs suggests a weak effect that may derive from a few extreme pairs. Furthermore, adjusting the significance threshold for the fact that multiple hypotheses were tested (two sets of SNPs in each of two populations) would have rendered the results insignificant.
Thus, with minor modifications of the methodology and analysis (excluding one extreme outlier couple, and correcting for multiple testing, e.g.), Chaix et al.'s findings become from marginally significant to non-significant. Derti et al. further suggest:
If mates were found to differ from non-mates in MHC relatedness, in these or other populations, we note that this phenomenon need not stem from mate selection alone, particularly if only couples with children are considered. If offspring with certain MHC allele combinations survive preferentially, exclusion of mated couples without children could yield a non-random MHC similarity distribution amongst the remaining couples. This idea is supported by the increase in MHC heterozygosity of mouse embryos following viral infection of the parents.
So, the seemingly solid relationship between sweat, MHC variation and mate choice is at least as rocky as the relationship between Blanche and Stan in Streetcar Named Desire. But why was it so seductive anyway? The idea that the kind of variation in MHC that is found in modern large out-bred populations existed and was a strong selective force when hominids lived in small bands is not even convincing on the face of it. And, if the results of different studies have been so equivocal that single women and women on oral contraceptives don't show the preference for men with dissimilar MHC, then what does that say about early hominids? During most of human evolution, adult females were not often cycling, because they were malnourished, pregnant or lactating -- meaning that it's unlikely that their hormonal profiles looked reliably like those of college students on which the T-shirt tests were carried out.

We have written a lot about the seductiveness of selective scenarios. But, this tenuous hypothesis has a questionable aroma about it, that of something too good (or too pat) to be true. Whatever the mechanism is, if it's actually there, and not just a manifestation of collegiate horniness, which is not at all certain, it may be incidental to the functions being tested, and may have evolved for entirely separate and unknown reasons.  (Of course, it's possible that this whole story has been misunderstood -- it could be that this sweat/attraction thing is a remnant of the ancient admixture with Neandertals that everyone's talking about.  That would explain why it's not found in Africans.)

Now whether nice arm muscles inside the T-shirt, or a tough-guy way of mumbling, can trigger your mating corpuscles, is a separate question. Brando does look pretty good in a t-shirt.


Holly Dunsworth said...


Anne Buchanan said...

Knew I wasn't the only one who felt that way about Brando.

Jennifer said...

but, evolution-wise, I'm thinking that t-shirts weren't considered at all along the human history, until just the last century or so.

And Brando just doesn't do it for me. 2 out of 3.

Anne Buchanan said...

That's true about the t-shirts, Jennifer, but mice apparently can detect MHC without any t-shirt tests, so probably humans can too. Or could, if that's what they were actually doing when they choose a mate.

Would you feel that way about Brando if he got himself a nice farmer's tan in that t-shirt?

amie said...

What about this for a great scientific study??

Ken Weiss said...

Well, there's a frenzy to geneticize behavior these days. Most of the findings have not borne out well in further studies, or the effects are smaller than first suggested.

Even if they were all true, and we're organic beings so have to be affected by our genetic variation, it still will not explain cultural patterns of mating. Further, other factors affect mating fidelity and bonding, including health, wealth, employment, children, and so on. These are cultural, as they have changed dramatically even just within European culture so fast that the gene pool--the underlying genetic variation, hasn't had time to change much at all.

Many might acknowledge that similar genetic variants might have different effects in different cultures, they might believe that within our own culture, the genetic variation will at least have predictive or even therapeutic value. But I think that won't be very effective, either, because our own cultural norms and circumstances change substantially even within a lifetime, and the predictive power of the genetic variations known so far is generally rather small.

It's a very anthropological story, but not very many anthropologists, at least not many who are committed to genetic explanations, seem to have digested. Making things genetic is too sexy these days. But it's a kind of argument that's been used ever since Darwin gave us the temptation to relate human traits to evolution, and hence to natural selection. And it's often done far more harm to people than good.

Still,I can't claim to know what makes for marital fidelity, one of the many complex social phenomena in our society.

Jason Hodgson said...

I think the evidence that species as distantly related as fish and mice mate dissasortatively with respect to the MHC is pretty good. I also think the MHC is variable enough that small populations would have enough variation to mate dissasortatively. So the idea that humans do it is not so far fetched. I actually think it would be surprising if we didn't do it to some extent. Of course there are a lot of other factors at play in human mate choice.

I did my masters in an MHC lab. When the silly t-shirt studies first came out they were generally thought ridiculous. However, some undergraduate did a research project to determine if women's preferences for t-shirt smell were repeatable. Do women prefer the same shirts? They repeated the study with women at the same point in their cycles in consecutive months. And they actually do choose the same smelly shirts!

Jason Hodgson said...

I also wonder about these HapMap studies. They are using SNPs that are located in the MHC region to assess MHC similarity. This is not ideal. The only MHC variation that matters for immune function is that which occurs in the antigen binding sites of the proteins. Most of the SNPs used in the study will not have any effect on the MHC proteins. The question then becomes how tightly linked are these SNPs to the different protein variants? If they aren't very tightly linked, then these SNPs will not be good proxies for actual protein variation.

This also raises the issue of what variation is being assessed by smell or otherwise? In other words, do animals make the same mistakes as the HapMap study sometimes?

Ken Weiss said...

It is interesting to wonder how the system works. I don't happen to know what the best knowledge is. But if it's olfactory, then one would want to know how olfactory receptors, unrelated to MHC genes (different gene family, for example) would be able to communicate MHC-specific reception to some system that matches that to the person's MHC antigens.

Humans evolved generally with prescriptive mating, usually of some form of cousin, but with clan or lineage exogamy. This is said to be so as to avoid inbreeding, but whether it's genetic or socio-economic in origin is debatable.

Other 'higher' primates may have some mate choice but even there there are some general determinants including dominance, male group exogamy, and so on.

These would seem to mitigate against mate choice by MHC specifically, though they might foster MHC correlations.

How and why axillary glands would specifically secrete MHC proteins, which are normally cell-surface-bound may be known, but it's not known by me.

There could be many other explanations. Also, the amount of MHC variation should be related to the strength of mate-choice effects, but I'm not aware of studies that have looked at that. Since MHC is involved in other kinds of functional selections, such as self-nonself recognition and interactions with the pathogenic environment, singling out mating (or, in mammals, uterine mother-fetus compatibility issues) isn't easy.

Whether the HapMap markers are suitable is a good question. The issue there could be that the HapMap markers yield limited sets of haplotype blocs, and the MHC variation may be too great for the markers to be effective in picking up linkage associations, as you say.

The final question I would ask, being no expert, is the strength of effect. If this is important, why not an easily detectable effect by research studies? Why, after so many years, is this still questioned? If the effect is weak, then what is its importance, and what fraction of MHC does it actually account for?

Anne Buchanan said...

I agree, Jason, it's perfectly possible that these studies do show that some kind of assortative mating is going on with respect to MHC. But, because I'm not sure that anyone has a good explanation for the mechanism -- if it is something about the immune system that's being picked up by olfaction, what is it and how? -- and because results are so variable, some finding that women detect variation but men don't, or that women on the pill or single women don't, that Yoruba don't, and so on, not to mention that the genetics are unclear, I am not yet ready to sign on. Indeed, if this is a trait that evolved long before humans, why wasn't it found in the Yoruba? Could be study design, sample selection explanations, but it could also be that the association just isn't strong enough or consistent enough to pick up. I am convinceable, just not yet convinced.

Jason Hodgson said...

I think it's worth remembering that who one pairs with and who one reproduces with aren't always the same person. The rate of non-paternity in "monogamous" species is generally pretty high and humans are probably no exception. risk of opening a can of worms, and related to a recent post, the untouchable Yanomami samples would be fantastic for testing the rate of non-paternity in a small human population. And anecdotally I've heard that as a biproduct of some of the analyses that have been done with these samples, the non-paternity rate has been observed to be significant.

The "good genes" sexual selectionists see no problem with the variation in women's preference across the cycle. The argument goes that women are looking for different things during different times in their cycle and whether pregnant, or lactating or cycling. During the most fertile phase of the cycle women are looking for good genes (hence the preference for MHC opposites in normally cycling women). When women aren't looking to get pregnant they are looking for resource investment from males. The story goes that men who have good genes may not be good providers. Therefore women may choose one man to provide genes and one poor sop to provide investment in the other man's child. ...and the sometimes observed preference for MHC similar males during times when conception is unlikely is conveniently explained as a preference for family during these times. Its all just so!

Jason Hodgson said...


I don't think the Yoruba present a huge problem either way. Aside from the potential problem of assessing MHC diversity that the HapMap data presents, there is some reason to believe that the null hypothesis that the study employed was inappropriate (at least for the Yoruba if not the mormons).

The null distribution that the study compared MHC similarity to was overall genomic similarity. The Yoruba are extremely genetically diverse relative to the Europeans used in the study (which consists solely of Utah Mormons). The Yoruba may simply be genetically diverse enough that you don't need the MHC to be more distantly related than the genomic average to create MHC heterozygous offspring.

Ken Weiss said...

If too many situations, conditions, and circumstances, overlain with all sorts of cultural constraints, perceptions, and misperceptions, and only a weak smelling ability apply, then the net selective effect on any one must be quite small.

In small human demes--our evolutionary heritage--drift will be as strong a player in what makes it over the generations. So that's one problem.

Another is the just-so element as you say. That removes almost all constraints.

Also, there is a temptation to try to make human-specific explanations, when these systems are clearly long pre-human. Since culture takes so many forms, the evolutionary constraints involved probably make very little difference, I'd say.

Anne Buchanan said...

There are so many arguments about which traits represent 'good genes' that if we were actually attending to them all we'd be paralyzed by indecision, since no potential mate has them all. And, the constraints of demography, ecology and culture have unpredictable effects that can overturn any selective argument.

Jason Hodgson said...

Well, the smelling ability isn't the only factor at play. For example it has been observed that macaques and humans have a higher rate of spontaneous abortion when the fetus is more MHC homozygous. So there may be pre and post fertilization mechanisms at play. In fact there may be some mechanisms at play during fertilization as there is some suggestion that MHC molecules expressed in sperm cells are involved.

Genetic drift cannot explain the variation seen in the MHC. If genetic drift was a strong force in explaining the evolution of the MHC then much of the variation in the complex would have been lost. It is just too variable to explain without some form or combination of balancing selection, or frequency dependent selection or dissasortative mating.

There may not be much of a role for sexual selection, but something is maintaining the variation in this most variable chunk of the genome. I think the MHC is the most variable genomic region in all vertebrates that have been looked at. That needs explanation.

Jason Hodgson said...


I find that unsatisfying. Do you think there no role for sexual selection in human mate choice? I think demography, ecology and culture certainly have a huge role to play. But for any selection argument, it is not enough to show there is no selection going on in some particular population in some particular slice of time. What matters for selection is that on average over many generations there is some excess reproduction. There is no reason to expect that you can observe this excess in any given population sample at any given chunk of time. Selection is tricky like that. Maybe any given expression of mate choice can only operate given a certain set of conditions. So long as it can operate sometimes and so long as these bouts are stronger than the drift that occurs between them then these genes will increase in frequency through time.

I'm skeptical of studies both in favor of and against some realization of sexual selection because the tests of the hypotheses are never quite appropriate to the actual predictions of the theory. However, I don't have any problem imagining that sexual selection plays a meaningful role in human mate choice at least sometimes.

Ken Weiss said...

If mate choice involves preferences that have any genetic basis, then sexual selection has to occur. But if the criteria are mainly environmental (including social but not genetic), its net effect will be small. Even if the criteria stay the same, in very small demes the probability of fixation of a slightly favored allele is hardly affected by the selective advantage.

If the selective effect is strong, which in human cultures I think is rarely true, or long-term consistent, then of course things will work as usually stated.

And if the criteria are always changing, for example because of fashions in clothing, body decoration, or more recently education, sports ability, etc., then the net impact of sexual selection, relative to other factors, will be slight

Anne Buchanan said...

No, Jason, my point is not that sexual selection never happens, but that selection scenarios are invoked far too often, when there are other equally possible explanations, most of which can't be tested or proven. Selection is very seductive, but people forget that it requires specific conditions, over very long periods of time. You yourself are careful to point this out, and to allow for the Just-So-ness of many selection stories. They are easy to think up, easy to rationalize and finesse and easy to sell, but to me, that doesn't make them good science.

Why is the MHC region so variable? Got me. But, so far, the sexual selection explanation doesn't convince me. The odor/immune connection and the representativeness of today's regularly cycling women to women throughout most of our evolutionary history are far too tenuous and the existence of many other constraints on mating and reproduction much too likely for me to buy it.

James Goetz said...

Does one of those studies mean that a woman could fall out of love with her partner if she stops or starts taking the pill?

Jason Hodgson said...


I totally agree that the science is abysmal. I don't think there is any good way to actually test these hypotheses. To me its like kin selection. Kin selection makes very specific predictions about when to help kin and when not to help kin (which is often forgotten). But it is impossible to test because there is no way to measure b and c. You can't know which way the prediction goes without knowing b and c so you can't ever test it (even if lots of researchers have claimed to have done just that). ...all the same, I'd be shocked if kin selection didnt operate sometimes.

Human mate choice is endlessly fascinating. It's just not possible to perform the controls necessary to study it properly. Too bad, because I have no problem imagining that these things are going on. So I think we mostly agree, though I am needlessly credulous!

Thanks for the interesting discussion and I enjoy your site!

Anne Buchanan said...

Jason, I think we mostly agree, yes. Glad you enjoy the site. We certainly enjoy your comments!

Anne Buchanan said...

Great question, Jim, but so many confounding variables....why did she go off the pill, did being on or off the pill affect her moods, does she love him during some times of her cycle but not others, and so on? And such a simple question!