Maps can tell surprising stories. About a year ago, Northwestern University psychologist Joan Chiao pondered a set of global maps that confounded conventional notions of what depression is, why we get it, and how genes — the so-called “depression gene” in particular — interact with environment and culture.
Chiao had run across data suggesting that many East Asians seemed to carry the “depression gene” — shorter variants, that is, of a mood-regulating gene known as the serotonin transporter gene, or SERT — at unusually high rates. Yet though dozens of studies over the prior 15 years had shown these short SERT genes made people more prone to react to trouble by becoming depressed or anxious,* it was not Chiao’s impression that this association held for most Asians. Then again, no one had gathered the data.
So she gathered it. Chiao and one of her grad students, Katherine Blizinsky, found all the papers they could that studied serotonin or depression in East Asian populations. These papers, along with similar studies in other countries and some World Health Organization data on mental health, painted a pretty good picture of short-SERT variant and depression rates not just in North American and Europe, but in East Asia. A pretty good picture — but seemingly twisted in the middle. The eastern half was upside down. For while East Asians carried the short-SERT “depression gene” variants at almost twice the rate (70-80%) that white westerners did (40-45%), they suffered less than half the rates of anxiety and depression.
You can see it in the maps. Below, the first map shows prevalence of the short-SERT ‘depression gene,” and the second shows prevalence of depression. Their colors should line up, but instead they conflict.
Fig 1. Known prevalence of S-S and S-L serotonin transporter gene variants worldwide. Yellow denotes low rates, orange middling rates (around 40-50%, and red high, around 80%. From Chiao and Blizinsky 2009.
Fig 2. Percentages of populace diagnosed with mood disorders at some time in lifetime. Again, yellow is low, in the single digits, while red is high, around 20%, and gray areas lack sufficient data. If the prevalance of the “depression gene” predicted the prevalence of depression, then this map should look much like the one above it. But — especially if you look at North American and Asia, which are the areas in interest here — it doesn’t. It looks almost ass-backwards. From Chiao and Blizinsky 2009. Gray areas lack sufficient data.
You can chart the data in other ways too, and it still looks weird. A well-established gene variant that is supposed to predict depression seems to predict just the opposite in East Asia.
Squaring two maps with a third
Why did fewer East Asians get depressed even though more of them carried the depression risk gene? It wasn’t as if life in East Asia was stress free. The gene seemed to generate vulnerability in one culture and resilience in another.
As Chiao recognized, several possibilities offered themselves. Might depression be underdiagnosed in East Asians and overdiagnosed in westerners? It might — but probably not enough to account for a complete reversal of the risk dynamic. Perhaps most East Asians carried some other gene that canceled the SERT gene’s depression risk? Again, could be, but it seemed an awfully strong effect.
To Chiao, these sorts of explanations couldn’t reconcile the two maps. The maps did start to make sense, however, when Chiao considered them in light of gene-culture evolutionary theory (aka dual inheritance theory). This is the notion that genes and culture influence each other, and that culture can shape the way genes express themselves and even how they evolve. To Chiao, the mismatch between the SERT map and the depression map smelled of gene-culture effects. The gene in question was obviously SERT. So what was the cultural suspect? What cultural difference between western whites and East Asians might affect both the prevalence and apparent effect of the so-called “depression gene”?
And what jumped out of that question, both to Chiao and Blizinsky and to Baldwin Way and Matthew Lieberman, a pair of UCLA researchers who happened to be asking the same questions in California, was the difference between individualism and collectivism.
This individualism – collectivism distinction comes not from Mao, but from a Dutch organizational sociologist named Geert Hofstede. Back in the 1970s, Hofstede did a massive study for IBM of several hundred thousand of the company’s workers in 70 countries. Hofstede found several cultural factors that shaped business practices differently in IBM offices around the globe, the most famous of which became the spectrum between individualistic cultures, which emphasize a person’s independence, and collectivist cultures, which emphasize a person’s interpersonal, social, and civic connections. The study wielded enormous influence and made the collectivism individualism spectrum a staple of certain strains of sociological studies. (For other echoes, see here.) And as another map from Chiao’s paper shows, the white west generally leans toward individualism while East leans toward collectivism.
Fig 3. Collectivism in world cultures. Yellow is low in collectivism, red is high. From Chiao and Blizinsky 2009.
So how does individualism-v-collectivism relate to depression and depression genes? Here Chiao and Blizinsky, as well as Way and Lieberman (these connections were apparently ripe) turned to another emerging idea: That the short SERT gene seems to sensitize people not just to bad experience, but to all experience, good or bad. (I explored this “sensitivity gene” or “differential susceptibility” hypothesis at length in an Atlantic article last December and am now working on a book about it.) Both Chiao & Blizinsky and Way & Lieberman published papers within the last year laying all this out: Chiao and Blizinsky last December (abstract; pdf), Way and Baldwin this June (abstract; pdf download; Replicated Typo has a good write-up here). And both pairs assert that these short SERT variants make people sensitive to social experience in particular.
Way and Lieberman, for instance, note several studies in which the short, or S/S variant, seems to magnify both the negative and positive effect of social support.
In a study of depressive symptomatology, when short/short individuals had experienced more positive than negative events over the last 6 months, they had the lowest levels of depressive symptomatology in the sample (Taylor et al., 2006), indicating that short/short individuals are more sensitive to positive life events as well as negative ones. Subsequent research has shown that this relationship between life events and affect for individuals with the short/short genotype was primarily driven by the social events, as the nonsocial events were not significantly related to affect (Way and Taylor, 2010). Other groups have found heightened sensitivity to positive social influences amongst short allele carriers as well, which has even been documented using neurochemical measures (Manuck et al., 2004). Thus, these results suggest that the 5-HTTLPR moderates sensitivity to social influence regardless of its valence [that is, whether the experience is positive or negative].
Because short/short individuals are more sensitive to the social realm, social support appears to be more important for maintaining their well-being. In support of this claim, short/short individuals exposed to a natural disaster (a hurricane) were at no higher risk for depression than long/long individuals provided they perceived that they had good social support (Kilpatrick et al., 2007). However, if short/short individuals exposed to this disaster perceived that they did not have good social support they had a 4.5 times greater risk for depression. Similarly, a randomized control trial designed to improve nurturant and involved parenting reduced adolescent risky behavior, but only amongst those with the short allele (Brody et al., 2009b). A similar differential sensitivity was seen among adolescents in foster care. If the short/short individuals had a reliable mentor present in their life they were at no higher risk for depression than adolescents with the other genotypes. However, if they did not have such support they were at a high risk for depression (Kaufman et al., 2004). Thus being embedded in a richly interconnected social network, as is present in collectivistic cultures, might be particularly im-portant for maintaining the well-being of short/short individuals.
This starts to explain the purported interplay of the S/S allele and a collectivist culture: If short-SERT people get more out of social support, a more supportive culture could buffer them against depression, easing any selective pressure against the gene. Meanwhile the gene’s growing prevalence would make the culture increasingly supportive, since those who carry it might be more empathetic. Studies have shown, for instance, that short-SERT people more readily recognize and react to others’ emotional states. In one still-unpublished study — a favorite of mine — marriage partners with S/S SERT alleles more accurately read and predicted their spouses’s emotional states than did people (sometimes those same partners) with L/L variants. This could make for some interesting dynamics at the breakfast table over the years.
A conversation between genes and culture
One major piece of the puzzle remains: How did the short SERT variant, which has generally been painted as bad news, become so prevalent in East Asia in the first place? Good question. The short-SERT variant appeared in humans only in the last 100,000 years. It was during this same period that humans moved out of Africa and spread around the globe. And it was during this time that the this S/S variant thrived in particular in people who moved east and took up residence in East Asia. Why did it blossom so spectacularly? And what came first, the high S/S rates or the collectivist culture?
Here the gene-culture dynamic must walk on tiptoes, as the sketchy evidence forces caution. Yet it can offer some speculative hypotheses. Drawing on work by Corey Fincher and Randy Thornhill, for instance, Chiao speculates that both a collectivist culture and the socially sensitive S/S allele gained ground when high pathogen loads along human migration routes from Africa to East Asia rewarded socially sensitive, collectivist behaviors that defended against pathogens. (The high pathogen loads in turn rose from the warm, most climates and abundant bird and mammal life in those regions.) The heightened danger of infection, that is, may have selected for a more group-oriented mindset, such as more attention to group rules regarding sanitation, food preparation, and whatever elemental medical care (such as stopping to rest) might have helped people avoid or survive infection. The adjustment would have been partially cultural: Those who followed these practices would suffer less infection. But (the argument goes) the adjustment would also have been genetic, as selection favored an S/S SERT variant that made carriers more likely to observe the rules.
I’m not quite sure what to think of this idea. A paper exploring the link between high pathogen loads and lower IQ recently came under fire, and this may too; yet Chiao cites a strong correlation. Meantime, Replicated Typo offers an alternative but compatible mechanism for this gene-culture evolution, based more directly on migration routes. In any case, as Chiao notes, pathogen loads offer just one among several possible environmental or cultural factors, not mutually exclusive, that might have selected for collectivist behavior and socially sensitive genotypes, creating a feedback loop increasingly friendly to behavior, gene, and a particular culture.
The new math
This is a lot to wrap your mind around. If you consider yourself of hard-nosed empirical bent, you might, after you take a deep breath or a long walk, cast about about for a “western-type” study that runs along more classic gene-environment lines. If you did, you would soon reel in a 2004 study of badly abused children. This study, by Joan Kaufman and others at the Yale genetic psychiatry lab of Joel Gelernter, looked at 57 school-age children who were so badly abused they were moved to foster homes.
First the researchers crossed the kids’ depression histories with their SERT genotypes. They found the expected: maltreated kids with the short SERT gene — the double whammy — suffered mood disorders at almost twice the rate as did maltreated kids who had the L-S or L-L variants or, for that matter, short-SERT kids with no maltreatment.
So far, so predictable. Then Kaufman laid both the depression scores and the SERT types across the kids’ level of “social support.” She defined social support quite narrowly: contact at least monthly with a trusted adult/mentor figure outside the home. This modest, closely defined social support, however, eliminated about 80% of the combined risk of the risk gene and the maltreatment. It virtually inoculated kids against extreme maltreatment and a proven genetic vulnerability.
It makes you wonder: What’s the real toxin in situations like this? We tend to view bad experience — abuse, violence, extreme stress, family strife — as toxic, and risk genes as semi-immunological weaknesses that let the toxin take hold. And maltreatment is clearly toxic. Yet if social support can almost completely block the effects of a severe toxin in a vulnerable individual, isn’t a lack of social support almost as toxic as the severe maltreatment? Even this clever study’s design and language frame “social support” as a protective add-on. But this framing implies that humanity’s default state is isolation. it’s not. Our default state is connection. To be unconnected — to feel alone — is to endure a trial almost as noxious as regular beatings and sharp neglect.
The University of Chicago psychologist John Cacioppo and William Patrick explore this beautifully in their book Loneliness, And Michael Lewis’s hysterically funny article about the Greek credit crisis, published just a few days ago, suggests that a hyper-individualistic default state doesn’t serve the world economy too well, either. Lewis describes how the Greek credit crisis, which currently threatens to spread to the European and perhaps the global economy, arose partly because a break in the social contract created an every-man-for-himself ethic in Greece, since everyone assumes everyone else cheats and that no one pays taxes. He signs off with this:
Will Greece default? There’s a school of thought that says they have no choice…. On the face of it, defaulting on their debts and walking away would seem a mad act: all Greek banks would instantly go bankrupt, the country would have no ability to pay for the many necessities it imports (oil, for instance), and the country would be punished for many years in the form of much higher interest rates, if and when it was allowed to borrow again. But the place does not behave as a collective … It behaves as a collection of atomized particles, each of which has grown accustomed to pursuing its own interest at the expense of the common good. There’s no question that the government is resolved to at least try to re-create Greek civic life. The only question is: Can such a thing, once lost, ever be re-created?
If Greece doesn’t do some fast gene-culture evolution toward collectivism, the whole world may get depressed.
As gene-culture theory gets hold of the kind of data that allows for papers like Chiao’s, I suspect we’ll see a growing stream of studies showing that genes have different effects in different cultures. A few weeks back, for instance, Ed Yong wrote up a fascinating paper by Heejung Kim and colleagues demonstrating that a particular variant of an oxytocin receptor made Americans, but not Koreans, more likely to seek emotional social support in times of distress. As Yong noted, these studies all but insist that we may need to expand our definition of environment when we consider gene-environment interactions.
Many studies have looked at how nature and nurture work together but in most cases, the “nurture” bit involves something social that’s either harsh or kind, such as loving or abusive parenting. Kim’s study stands out because it looks as cultural conventions instead, and Ebstein says that it “provides an interesting new avenue for researching gene-environment interactions.”
In a sense, these studies are looking not at gene-x-environment interactions, or GxE, but at genes x (immediate) environment x culture — GxExC. The third variable can make all the difference. Gene-by-environment studies over the last 20 years have contributed enormously to our understanding of mood and behavior. Without them we would not have studies, like these led by Chiao and Way and Kim, that suggest broader and deeper dimensions to what makes us struggle, thrive, or just act differently in different situations. GxE is clearly important. But when we leave out variations in culture, we risk profoundly misunderstanding how these genes — and the people who carry them — actually operate in the big wide world.
_________
*This so-called “depression gene” is, to most researchers, either of two “short” versions of the serotonin transporter gene —SLC6A4, and known by some as SERT. SERT appears to regulate levels of the neurotransmitter serotonin and to be crucial to mood, among other things. Because we effectively get one half of this gene from each parent — either a “long” or a “short” — each of us carries a version that is either long-long (L/L), long-short (L/S), or short-short (S/S). As many have pointed out, depression is far more complicated than one gene , which is probably why even the depression gene is so clearly probabilistic rather than predictive. But as we’ll see shortly, it’s far more complicated than that.
Yet others may recall that this depression-risk view of the gene was aggressively challenged by Risch et alia. That view got a lot of attention. Less attention went to the several strong rebuttals and critiques noting that Risch’s challenge was a) was rather selective in its choice of studies to include in its meta-analysis, b) weighted the studies in odd ways, so that the small ones least likely to see a genetic association were weighed as heavily as the large ones that were more likely to detect such associations; and c) ignored altogether a huge body of physiological work that details mechanisms through which SERT variants could affect one’s sensitivity to environment. As I noted in an earlier post, that leaves intact the SERT-depression link’s main framework.
