Orchids & Dandelions Abloom (Repost)


Can Genes Send You High or Low? The Orchid Hypothesis A-bloom

by David Dobbs

Originally posted March 2012*


A few years ago, Arial Knafo, a psychologist at Jerusalem University, wanted to see if three-year-olds would share their bonbons. Snack time would come amid a bunch of other activity at Knafo’s lab — drawing, games, doll-making — whose real purpose was to disguise tests of prosocial behavior in these toddlers. The researcher, saying it was high time for a snack, would bring out two packages of Bambas, peanut-butter-flavored corn puffs much coveted in Israel. The child’s pack, like every pack, would hold 24 of the little treats. But when the researcher opened her pack, she would cry out in dismay, dump the bag out on her plate, and say, “Ohhh, mine has only three!” Which it did, because the researcher had earlier removed the rest. Would the child share her bigger treasure without being asked?

Most did not. This was expected. “The average child,” says Knafo, “will help or share without being asked less than one in three times. Self-initiated sharing is a difficult task — they have to detect the need, then decide to do it.” A few 3-year-olds, however, will do it far more often than their counterparts. And in Knafo’s study, the ones who tended to share more were kids carrying what is generally considered a risk gene for antisocial behavior: DRD4-7R, a variant of a dopamine-processing gene called DRD4. In a pile of previous studies, 7R kids, if they had harsh or distant parents, were far more likely to develop attention, social, conduct, and school problems. These studies had given the DRD4‘s 7R variant a reputation as a “vulnerability gene” — bad news. People had dubbed it the ADHD gene, the drinking gene, the bully gene, even the slut gene. Now Knafo, in effect, was calling it the Bamba-sharing gene.

Why the difference? Simply that Knafo, unlike most prior investigators, looked at the interaction between 7R and all parenting, good and bad, rather than focusing on the bad. And when he did, he found that the gene didn’t just create greater vulnerability to problematic parenting; it created greater response to any parenting.

Knafo’s study supported a bold reconception of the vulnerability-gene hypothesis — the view, dominant for more than a decade, that certain heavily researched genes create risk for mood and behavioral problems. These include not just DRD4 but the serotonin transporter gene (also known as SERT or 5HTTLPR), the short variant of which is often blamed for depression and anxiety disorders, and a variant of MAOA, the monoamine oxidase A gene, that some studies associated with aggression or violence. In the conventional vulnerability- or risk-gene model (also known as the diathesis-stress model), these genes create specific vulnerabilities that stress can reveal. Knafo’s research supported a view recasting those problems as merely the downside of a genetically shaped higher sensitivity to experience — the upside being enhanced function and happiness. The evidence for this revision lies not only in Knafo’s studies but, embedded but overlooked, in the data of the very studies that created the risk-gene view of these variants.

This revisionist hypothesis is known variously as the sensitivity hypothesis, the differential susceptibility hypothesis, or the orchid-dandelion hypothesis — a term that Thomas Boyce and Bruce Ellis coined based on the vernacular Swedish term “dandelion children,” who seem to grow up okay in almost any environment; to that they added “orchid children,” who thrive under good care but wilt under bad. It is a young hypothesis, hatched 15 years ago and obscure for most of that time. But in the last two or three years it has gained enormous traction, spreading through behavioral genetics, child development, and anthropology.

“This thing is just exploding,” says Jay Belsky, a developmental psychologist at the University of California, Davis, who helped pioneer the idea with a seminal paper in 1996.

In a special issue of Development and Psychopathology this past February, Belsky and researchers W. Thomas Boyce and Bruce Ellis, who had independently forged a parallel view rising from physiological reactivity studies, reconciled their approaches into a more comprehensive framework. A dozen papers tallied the growing evidence. The orchid hypothesis is now a staple at child development conferences, and the upside/downside paradigm is becoming common in behavioral genetic studies. It offers to radically rework our conceptions of mental illness — and of how genes shape our behavior, our moods, and even our evolution.

Yet the orchid hypothesis faces new obstacles as the vulnerabilty-gene paradigm out of which it grew comes under closer scrutiny. With genetics moving toward sophisticated whole-genome studies using data from thousands of people, researchers can no longer rely solely on the kind of small-scale, individual-gene studies that led to the orchid hypothesis. (Such studies are called candidate-gene studies, since they study a gene already suspected of having an effect.) If the orchid-gene hypothesis is to hold up, researchers must broaden and deepen their evidence. Will it prove just a pretty story — or will the orchid hypothesis show the flexibility and strength to adapt among changing conditions of proof?


To Elaine Aron, a psychiatrist and researcher who has written since the mid-1990s about what she calls highly sensitive persons, or HSPs, the sensitivity granted by “orchid genes” starts with a heightened attention, followed by deeper processing and, ultimately, response.

In a review paper now in press1, she draws on several lines of animal and human research to describe a sensitive creature who typically pauses in the face of a new or ambiguous situation. “This can look like anxious inaction,” says Aron. “But it’s actually a time to gather information and take things in.” When her own child, for instance, hung back quietly on the first day of preschool years ago, the teacher asked Aron if the child was scared. “But he was just taking it in,” says Aron. “Soon enough he engaged quite fully.”

This pause suggests a third mode supplementing the classic behavioral choice between approach and withdrawal: a pause to read cues and await opportunity. Aron cites studies showing this mode in animals from fruit flies to a freshwater fish called a pumpkinseed2. Some animals measure a new context more carefully before either exploiting the new environment more fully or, if the new situation looks unpromising, moving on to a different arena. The more sensitive creature’s heightened emotional state, says Aron, burns the experience more deeply into psyche and memory, creating valuable lessons for future choices.

Belsky, meanwhile, is trying to broaden the empirical base with both larger studies and multigenic ones that gauge additive effects of purported plasticity genes. Late in 2010 he and Kevin Beaver published a study of 1,586 adolescents* representative of the 26,000 in a major national U.S. longitudinal study. The environmental variable was maternal involvement in the children’s lives, such as how emotionally engaged they were and how much time they spent with their boys. The behavioral outcome measured, by both interview and testing several times over a 12-year period, was self-regulation of attention, emotions, and behavior, such as whether the youths had trouble paying attention, whether they used an organized way of making decisions, and how well they controlled their temper. Once tabulated, the results were analyzed according to how they were influenced by five genes that have variants previously associated with effects on behavior or mood: three genes affecting dopamine availability and processing (DAT1, DRD2, and DRD4); the serotonin transporter gene, 5HTTLPR (or SERT); and the two variants of the MAOA gene.

Strangely, the researchers found no significant effects on the girls in the study. “We’ve no idea why,” says Belsky, “and can only guess, really: Maybe girls self-regulate better than boys. Maybe they’re less sensitive to maternal engagement at that age. But these are only wild guesses.”

The boys did react, however, with additive effects showing in boys who carried more than one of the plasticity variants. Those with just one variant reacted about the same to maternal involvement as did the boys with no variants, which is to say they reacted very little: To boys with no or just one variant, it seemed to make little difference whether their mothers were engaged — a true dandelion effect.

Boys with two or three plasticity variants, however, showed a steeply sloped sensitivity, with self-regulation scores dropping sharply from the mean among those with distant mothers and rising sharply among those with more engaged mothers. To these kids, mothering style was a big deal. The effect was even greater for boys with four or five plasticity variants.

This is just one study. But the additive effect seems to argue well for the orchid hypothesis. And the absence of significant effect in boys with just one plasticity variant may suggest why studies on individual candidate genes produce such varying results: The effect of any one plasticity gene may depend heavily on whether a person also carries a second.


Individual studies like Knafo’s, meanwhile, are revealing some interesting wrinkles.

Knafo, as earlier described, tested 3-year-olds and, confirming the orchid hypothesis, found that kids with the more plastic DRD4-7R variant and warm, engaged parenting were more prosocial than even warmly parented kids with the gene’s more common 4R form. These 7R kids with good parents not only shared their Bambas; they were more likely to pick up pencils a researcher had “accidentally” knocked onto the floor, express sympathy when a researcher bumped her knee on the table, or help find and then comfort a missing doll.

But a layer down, Knafo found a surprise. He had rated the kids’ mothers not only on warmth and engagement but on whether they sometimes punished without explanation. Earlier studies — studies that didn’t look at genotype — generally found that unexplained punishment produces little prosocial effect or creates a contrarian pushback. This held true for most of Knafo’s kids. But in a subset of his kids, unexplained punishment created sharply higher self-initiated prosocial behavior — and that subset, paradoxically, were those carrying the 7R plasticity variant.

What on earth was that about? Possibly it’s a fluke that other studies won’t replicate. But Knafo thinks it might be a sort of inverted consequence of sensitivity, an overreaction. He thinks these 7R kids, cue-alert and eager to make an unpredictable world less so, may exercise what psychologists call pathological altruism: they try to stave off disapproval or harsh treatment by being proactively helpful.

One final finding that impressed Knafo was that the prosocial edge held by 7R kids with good parents grew over the three years he followed these children. Possibly it simply reflected a cumulative effect. It may also refflect a positive feedback loop, as a sensitive, responsive child and a sensitive, engaged parent each responds to the growing good chemistry developing between them. Some of these mothers were almost surely 7R themselves, since at least one parent had to be. Knafo hadn’t genotyped them, so he can’t say. Studies that genotype both parent and child are on the short list for additional ways to test these effects.


One of the strengths of gene-environment studies is that they can find specific responses to particular environmental influences. Yet critics of these studies have seized on exactly this sort of response-specific focus because it can lead researchers to unintentionally manipulate responses to show a result— or, if they measure enough responses, focus on those that show the greatest result, even if that result came by chance or artifact.

“It’s not necessarily that anyone is trying to game a study,” says Daniel MacArthur, who just opened a lab at Massachusetts General Hospital that will use whole-genome studies to investigate rare and complex diseases. MacArthur prefers whole-genome studies because they avoid some of the bias problems that can creep into candidate-gene studies. Rather than start with a hypothesis and a candidate gene, as gene-by-environment researchers do, whole-genome researchers choose a disease, then examine the genomes of hundreds or thousands of people who have the disease to see if they have genes or gene variants that don’t appear in people without the disease. “There’s no massaging of data or decisions about what environmental conditions to look at,” says MacArthur. Associated genes either show up or not.

In candidate-gene studies of the sort used in orchid-gene studies, however, MacArthur worries that “understandably, someone convinced that a dynamic exists might throw a bunch of different filters at the data — different definitions of environment, different measures of behavior — until they find the relation they’re convinced is there.” Parse enough measures of environment and behavior, in other words, and you’ll eventually find a pairing that shows that some environment seems to lead to some change in behavior. In addition, such studies are often fairly small, since they require a lot of data collection, and this increases the odds that an effect that is actually a statistical fluke will be snatched up as proof of principle.

Such criticisms were central in a withering attack on behavioral candidate-gene studies in 2009 from a team led by whole-genome researcher Neil Risch, of the University of California, San Francisco. The team essentially conducted a brisk strip-search of a 2003 study by Avshalom Caspi and Terrie Moffitt that established the short SERT variant as a risk gene for depression —perhaps the single most foundational paper in gene-environment studies of mood and behavior. Risch collected a selection of papers that purported to replicate the Caspi and Moffitt findings and concluded they essentially proved it false. A feud ensued as Caspi allies accused Risch and colleagues of gaming their meta-analysis by doing some cherry-picking of their own.

“They picked just the papers you’d pick to disprove Caspi and Moffitt and left out the ones that would support it,” says Belsky. Another researcher said, simply, “That Risch paper is bullshit.”

Caspi and Moffitt responded a bit more diplomatically in a major 2010 paper in the American Journal of Psychiatry. Analyzing what they said were all the replicating papers rather than a selection, they argued that not only did those studies confirm a gene-by-environment interaction for the serotonin transporter gene and depression, but that human studies showed that the short SERT produced distinctive sensitivities in studies of brain waves, brain imaging, hormonal response, and inflammation, while tightly controlled animal studies showed that short SERT variants created similar physiological sensitivities as well as a more sensitive temperament.

This debate will continue. MacArthur, meanwhile, acknowledges that the agnostic, a-theoretical approach that whole-genome researchers prefer can’t readily test the orchid hypothesis. For one thing, if an orchid gene has both upside and downside effects — which is the orchid hypothesis’s central assertion — those effects might cancel each other out, leaving no significant “disease” effect visible. In addition, whole-genome approaches don’t generally compare genetic effects by environment — and even in conventional candidate-gene studies, the candidate genes such as DRD4-7R or the short SERT rarely produce measurable effects unless the study group is split according to environmental measures. Standard whole-genome surveys built for detecting disease, in other words, simply aren’t likely to reveal the bidirectional sensitivity that orchid genes purportedly create, since positive and negative measures of, say, depression, would cancel one another out.

To remedy this, Caspi and Moffitt, in their 2010 paper, call for whole-genome studies[1] that include environmental measures. If done well, such studies might go a long way toward proving or disproving the hypothesis. Meanwhile, some orchid-hypothesis researchers are looking for physiological effects of orchid variants, including heightened gene expression and other epigenetic changes.

There is also, finally, an evolutionary argument for the importance of these polymorphisms: These variants, Belsky and others note, appear to have emerged and then rapidly expanded through humankind over the last 50,000 to 100,000 years. Of the leading orchid-gene variants — the short SERT, the 7R DRD4, the more plastic version of the MAOA gene — none existed in humans 80,000 years ago. But since emerging through mutation (or, possibly, through interbreeding with other hominids), they have spread into 20 to 35% of the population.

“That’s not random drift,” says evolutionary anthropologist John Hawks. “They’re being selected for.”

They may not spread much further. For most of these variants, the adaptive edge gained would likely hold only as long as the trait didn’t spread too wide in a population — a dynamic known as negative frequency-dependency. The cost of aggression, for instance (such as someone hitting back), rises with the trait’s frequency in a population. Likewise, an extra taste for novelty or exploration — both of which are “expensive” in evolutionary terms, requiring time, energy, and risk — would become nonadaptive if everyone had it. This makes such traits self-regulating in a population, for if they become too common they become disadvantages and are selected out.

Still, such plasticity genes serve a group well if they remain in a minority, since — to stereotype a bit — they create a populace of steady, reliable do-ers (dandelions) mixed with individuals of more volatile temperament but greater range of behavior (orchids). Thus some evolutionary anthropologists have argued that some of the traits associated with these variants, such as the high sensation-seeking and risk-taking found in 7R carriers in many DRD4 studies, may have helped drive human expansion around the globe;[2].and in fact the 7R variant appears at its highest rates in populations5 that migrated fastest and furthest from Africa and Arabia. The sensitivity to cues and flexibility of response that orchid genes provide, at both the individual and group levels, may have been essential to our human success.

It may seem odd to link such communal achievements to bits of behavior such as a toddler’s willingness to share Bambas. Yet if the orchid hypothesis is right, the genes and genetic dynamics that help create some of our most grievous frailties and foibles — anxiety and aggression, melancholy and murder — may also underlie our greatest strengths and successes. Something to ponder next time you’re offered a sweet.


Note: The piece above is an expansion of a feature I published a few weeks ago in New Scientist. This version adds material that wouldn’t fit in the New Scientist space, most notably on caveats, complications, and other conundra, and I’ve moved a few things around to make room for those. It draws from research for a book I’m now writing, The Orchid and the Dandelion (Crown; ETA 2013). I originally explored this subject, at more length (and with monkeys) in a November 2009 Atlantic article, “Orchid Children.”

Photo by Bitman, via Creative Commons license at flickr.

Figure from Belsky and Beaver 2011, cited below.


1 Elaine Aron, Arthur aron, and Jadzia Jagiellowicz, “Sensory Processing Sensitivity: A Reivew in the Light of the Evolution of Biological Responsivity,” Personality and Social Psychology Review, in press [as of 12/5/11]

2 Summarized on pp 201-203 of Wilson, DS. 1998. “Adaptive individual differences within single populations.” … Transactions of the Royal Society of …. http://rstb.royalsocietypublishing.org/content/353/1366/199.short.

3 As related by Belsky in interview, 12/5/11. Study is Drury, S.S. et al. (submitted). Genetic sensitivity to the caregiving context: The influence of 5httplor and BDNF val66 met on indiscriminant social behavior.

4 Caspi, A, a R Hariri, A Holmes, R. Uher, and T E Moffitt. 2010. “Genetic Sensitivity to the Environment: the Case of the Serotonin Transporter Gene and Its Implications for Studying Complex Diseases and Traits.” American Journal of Psychiatry 167 (5) (May 3): 509–527. doi:10.1176/appi.ajp.2010.09101452. 4

5 Matthews, LJ. “Novelty‐seeking DRD4 polymorphisms are associated with human migration distance out‐of‐Africa after controlling for neutral population gene structure.” … journal of physical anthropology (2011).

*Belsky, Jay, and Kevin Beaver. 2011. “Cumulative Genetic Plasticity, Parenting and Adolescent Self-Regulation” (April 6): 1–8. doi:10.1111/j.1469-7610.2010.02327.x.

[1] See note 137 in Caspi and Moffitt 2010

[2] See Harpending and Cochran, The 10,000-Year Explosion.



This post is #10 in my Best of Neuron Culture Moving Party —  a run of 10 of my favorite posts from the blog’s stay at WIRED, posted on the eve of my departure to move the blog here on June 8 , 2013. The Party spotlighted each day a post from the past that I feel embodies the best of Neuron Culture’s WIRED tenure.

Why leave Wired? So I can focus more steadily for a time on finishing my book, tentatively titled The Orchid and the Dandelion, that I’ve often mentioned here. I know some people manage it, but I’ve found it hard to reconcile the demands of blogging at a venue like Wired and of writing a serious book that requires deep immersion: a matter of not just the time needed for each venture, but of the mindset and what you might call the focal length of one’s mental lens. A venue like this requires, methinks, either an unrelenting focus on a particular beat or a fairly steady tour through many fields; I can’t seem to mesh either with the sort of time and focus needed for a book. The move also frees me up to experiment a bit more with the blog now that it’s self-hosted.

5 responses

  1. I find this research to be fascinating and I wonder about this current pop culture fascination with sociopathy and narcissism… To me it seems like both likely have a positive aspect as well… To your knowledge have any of the gene variants been implicated in research on sociopaths?

  2. David, I am absolutely fascinated by this research. I believe the possibility strongly exists that I am an orchid. Do you have any suggestions as to a method of getting the SERT genotype in the private sector?
    Thanks a bunch. Enjoy your articles very much.

Leave a Reply

Your email address will not be published. Required fields are marked *