Evolution has provided the members of our species with a large, intricately wired brain and a lengthy period of immaturity: the prerequisites for an education. Natural selection favors attributes that contribute to their owner’s reproductive success. Costly attributes—the ability to be modified by experience is a costly attribute—are unlikely to be selected if the modifications are made on a random basis, because random modifications would have random effects on an individual’s chances of surviving and reproducing. Making personality plastic but leaving it at the mercy of chance, to be buffeted this way and that by unpredictable events, doesn’t make sense from an evolutionary standpoint. Natural selection would favor any alternative to randomness that produced results that were better than chance.
That statement isn’t something I’ve proved: it’s a hypothesis, derived from Darwinian reasoning. I am hypothesizing that evolution made personality plastic so that children can profit from their experiences—so that they can learn ways of behaving that will serve them well in adulthood. I am hypothesizing an adaptability that gives children some advantage, or ameliorates a disadvantage, in the competition to survive and reproduce. That hypothesis has to be backed up with evidence, which I will do my best to provide.
I have given randomness an arm and a leg: I’ve admitted that some of the unexplained variance in personality is due to biological noise and some to environmental noise. Now it is following me around like the crocodile that followed Captain Hook: it wants all the unexplained variance. But I’m not going to give in to it that easily. Randomness is the explanation, or the excuse, that a scientist resorts to when all else fails. It is a shrug of the shoulders, an admission of defeat.
Let us proceed on the premise that there is a real solution, a satisfying solution, to this mystery.
3
Monkey Business
I BECAME A WRITER of college textbooks because it was something I could do at home. I gave up writing textbooks when it suddenly occurred to me that what I had been telling my undergraduate readers was wrong. But the first thing I did after that realization struck was not to write a book of a different sort: first I wrote an article, laying out some of the evidence against the prevailing theories of child development, and submitted it to the Psychological Review. Despite my lack of the usual credentials—I have neither a Ph.D. nor a university affiliation—the journal accepted my article and published it.1
It caused very little stir, at least among the people who taught the courses for which I used to write textbooks, the professors of developmental psychology. But not everybody in the academic world ignored my article. In fact, the American Psychological Association gave me an award for it: the George A. Miller Award, named in honor of an eminent cognitive psychologist and presented each year to the writer of an outstanding article in psychology. The award came with a check for five hundred dollars, a bronze plaque (hanging right here on the wall of my office), and an invitation to give a fifty-minute address to a large audience of APA members at their annual convention, held in San Francisco that year.
The year was 1998. Although I hadn’t reached the point of being unable to travel, California is a long way from New Jersey and the trip would be exhausting and somewhat risky for me. And I had had very little experience speaking in public. Nevertheless, I accepted the invitation. I arrived in San Francisco a day and a half before my talk and spent that time resting in my hotel room.
I began my address2 by reading aloud the letter I had received, thirty-eight years earlier, from the Harvard Department of Psychology. The letter explained that the department had decided not to let me stay on for the Ph.D. degree. They had decided that I wasn’t worthy of a Ph.D., the letter said, because they didn’t think I had the “originality and independence” to live up to Harvard’s standards. The punch line came when I told my audience who signed that letter: the acting chairman of the Harvard Department of Psychology, George A. Miller.
The audience’s reaction to my revelation—the fact that I was receiving an award named after the very man who had kicked me out of graduate school—was just as I’d hoped: they burst into laughter. Perhaps that moment of shared merriment helped to diffuse some of the hostility I might otherwise have received; they were pretty gentle with me during the question period. The flak came afterwards, when the APA was publicly criticized for giving me the award and angry letters were published in the organization’s monthly magazine.3
But the real cause of the angry letters was not the article: it was my book, The Nurture Assumption, which—as luck would have it—appeared in print just a few days after my talk in San Francisco. The developmentalists had paid no attention to my article but it wasn’t so easy to ignore the book. I had gone over their heads, as it were, and taken my case directly to the general public.
That got the developmentalists’ attention, all right. It was like giving them a wedgie. What I was saying—that parents have no power to mold their children’s personalities—was heresy to their ears. Many of these people had spent their entire professional lives doing research designed to show how parents mold their children’s personalities. Not whether parents mold their children’s personalities but how they do it. That parents have this power was something the developmentalists simply took for granted. That’s what I meant by “the nurture assumption.”
With the publication of the book, my life changed in an eyeblink. For years I had worked quietly at home, seldom seeing anyone outside my family. Suddenly journalists were clamoring for interviews and TV producers were asking if they could come to my home to film me. Though I was unable to go on book tours or appear on talk shows, essays about the book were printed in newspapers and news magazines around the country.
The media attention hadn’t yet died down when I got a phone call from Marie Bristol-Power, of the National Institute of Child Health and Human Development (abbreviated NICHD), part of the National Institutes of Health. She wanted to organize a conference on parenting, she told me, at the NICHD campus in Bethesda, Maryland. The purpose of the conference would be to discuss some of the issues I raised in my book. Would I give the keynote address? I said yes.
It wasn’t until much later that I heard about the dispute that erupted after Marie proposed the conference. “Why would you want to invite her?” she was asked by her colleagues. “Hasn’t she gotten enough publicity already?” One prominent developmentalist who was invited to the conference said she wouldn’t come if I was going to be one of the speakers. Marie didn’t waver. “What, are you afraid of a little woman from New Jersey?” she asked. “But there’s so much evidence against what she’s saying,” protested the developmentalist. “Fine,” said Marie. “Tell the audience about it.”4
“I had to fight to get that conference,” Marie told me. “I had to fight to get you on the program.” Marie won those fights: the conference was held and I was invited to it. But she had had to make some concessions. The program was packed with developmentalists who were opposed to me, and there were two keynote speakers rather than one. The other, who got top billing, was Eleanor Maccoby, the grande dame of developmental psychology. Though officially retired from a long teaching career at Stanford, she was still quite active and mentally sharp.
Being an amateur at public speaking, I am always very nervous beforehand. At the NICHD conference I would be facing an audience that had already taken up arms against me. Only one of my allies—David Rowe, a behavioral geneticist at the University of Arizona—had been invited. My talk was scheduled right after Maccoby’s, and Maccoby was a pro who had been speaking in public for years.
To my surprise, she stepped up to the podium, pulled out a thick sheaf of paper, and began to read from it. That’s how Maccoby gave her talk: she read it from a typescript, rapidly and dryly, hardly ever looking up at her audience. She said I was wrong, but the evidence she cited didn’t worry me and her humdrum performance gave me courage. Nevertheless, the audience was friendly to Maccoby and not to me. They sat there stony-faced while I tal
ked. My little jokes evoked no more than scattered titters.
Afterwards Maccoby and I took questions. All the questions were addressed to me and most of them were hostile. But the one I remember—the most overtly hostile of them all—came from a heavyset fellow at the back of the room. He stood up and told me, in a scornful voice, that I was ignoring interactions. Naturally the behavioral geneticists hadn’t found any effects of the shared home environment: they had forgotten about interactions. Nature and nurture don’t act directly to produce developmental outcomes: they interact.
The heavyset fellow turned out to be Stephen Suomi, head of the Laboratory of Comparative Ethology at NICHD. He does research on rhesus monkeys, not on human children. His comment—it wasn’t really a question because he clearly wasn’t expecting an answer—had to do with interactions between genes and environment. Gene-environment interactions are my third red herring.
Stephen Suomi wasn’t afraid of a little woman from New Jersey. Maybe he should have been.
If you took a biology course in college you may have seen a drawing, similar to the one in the diagram here, of a bunch of plants.5 Each of the seven plants in the top row is the “identical twin” of the one directly below it in the bottom row: they were both grown from cuttings from the same parent plant, which means the two in each pair have identical genotypes (the biologists’ way of saying they have exactly the same genes). So what we have here are seven different plant genotypes, all from the same species, grown under two different environmental conditions: the top ones, let’s say, at sea level, the bottom ones at a higher elevation. The point of the picture is that some genotypes grew taller in one condition, others grew taller in the other condition. If this result is obtained reliably—if certain genotypes consistently grow taller in one environmental condition and other genotypes consistently grow taller in the other—then we have evidence of a gene-environment interaction.
But there is a curious thing about this diagram: it shows no “main effects,” to use the statisticians’ term, either of genotype or of environment. A main effect of genotype would mean that some genotypes tended to grow taller than others in every environment. A main effect of environment would mean that some environments are more favorable to plant growth than others. The second diagram illustrates both of these effects; this is in fact a far more likely outcome than the one shown in the first diagram. In nature, main effects of genes and of environment are common. When interactions occur, they usually are superimposed on main effects. Often they are not apparent to the naked eye and special statistics must be used to tease them out.
At the NICHD conference it soon became clear that cutting-edge developmentalists, shaken by the revelations of the behavioral geneticists, had regrouped. The most knowledgeable among them, including Eleanor Maccoby and Stephen Suomi, had pretty much given up hope of finding main effects of the home environment on children and were pinning their hopes on gene-environment interactions.
In casual use, the word interaction just means that two things have something to do with one another, as in “Ginger Rogers interacted with Fred Astaire.” But the people at the NICHD conference were using the term in its technical, statistical sense. Here’s how Maccoby put it:
Nowadays, interactions between parenting styles and possible genetic attributes are a matter of active interest in developmental psychology…. [One group of researchers] found that a given parenting style has a different effect on a child who is temperamentally bold and adventurous than it does on a timid, shy child…. If we were to aggregate the two kinds of children together, some of the parenting effects would wash out.6
There was a good reason for the developmentalists to pin their hopes on interactions: the behavioral geneticists’ data didn’t rule them out. Any effects of gene-environment interactions would contribute to the unexplained portion of the variance in personality, not to the portion attributed to the shared home environment or to the portion attributed to heritability. So this was a way for the developmentalists to nod their heads at the behavioral genetic evidence while retaining their faith in the importance of the home. The home environment might be important—they were certain it was important!—but it doesn’t make children more alike. Quoting Maccoby again,
If a given kind of household or a given style of parenting has different effects for children with different predispositions, this means that parenting often functions to make children in the same family different rather than alike.7
It was Maccoby’s answer to the question posed a dozen years earlier by behavioral geneticists Robert Plomin and Denise Daniels: “Why are children in the same family so different from one another?”8 Maccoby had interpreted “children in the same family” to mean siblings, ordinary siblings. Ordinary siblings have different genotypes, and people with different genotypes might indeed react differently to the same environmental conditions. But Plomin and Daniels’ question wasn’t just about ordinary siblings: much of the data they summarized in their groundbreaking article came from twins. Identical twins have the same genotype. A gene-environment interaction cannot explain why genetically identical individuals reared in the same household turn out differently. Genetically identical individuals are presumed to be born with the same predispositions, and therefore should react in the same way to a given kind of household or a given style of parenting.
Perhaps the reason why it never occurred to the developmentalists that they might be called upon to explain the personality differences between identical twins is that so little has been said about their differences. On the contrary, we’re always hearing about how amazingly alike they are. You’ve heard, I’m sure, the stories about the pairs who were separated at birth, reared in different homes, and reunited in adulthood. There were the twins who both enjoyed startling people by sneezing in elevators. The twins who, at the beach, were afraid to confront the waves straight on and always went into the water backwards. The twins who both married women named Dorothy Jane Sheckelburger. Okay, I made the last one up, but it’s only a little more bizarre than some of the actual reports.
That there are some striking similarities in personality between identical twins is beyond dispute. These similarities, however, are entirely due to their matching genotypes. As I explained in the previous chapter, once you take into account their identical genes, the differences between identical twins—the nongenetic differences—are as wide as the nongenetic differences between ordinary siblings. So an environmental mechanism that produces differences between ordinary siblings, but doesn’t work for identical twins, is not going to be the answer to the developmentalists’ prayer. Nor is it the perpetrator I’m looking for.
But let me put the question of identical twins aside for the moment—I have other bones to pick in this chapter—and just consider whether gene-environment interactions can account for the differences between ordinary siblings.
What the behavioral genetic data showed is that the home environment has no main effects. The plants grown in one environmental condition were, on average, as tall as the plants grown in another. The net effect of the home environment was essentially zero. Can a gene-environment interaction explain how the net effect of environmental conditions can average out to zero, or near zero? Yes, if it’s the crossover kind of interaction, illustrated in the first of the two sets of plant diagrams. Some genotypes—plants or children—do better in one environmental condition, others do better in another. Perhaps having parents who fight all the time causes some children to become more outgoing, less agreeable, and less conscientious, while others in the same household react by becoming less outgoing, more agreeable, and more conscientious.
Crossover interactions, however, are rare in nature. The authors of biology textbooks have to grope for genuine examples. One involves the eyes of fruit flies. It seems that normal (wild-type) fruit flies grow smaller eyes if they’re kept at temperatures above 25 degrees Celsius, but a mutant variety, called “infra-bar,” grow larger eyes at higher temperatures.9 So if you had
an even mix of wild-type and infra-bar fruit flies and didn’t know you had two different kinds, you might think that growing them at high temperatures had no effect on eye size.
The kind of interaction that is far more common in nature doesn’t involve a crossover but only what might be called “sensitivity.” Some genotypes are more sensitive than others to certain environmental conditions. Some genotypes respond, positively or negatively, to conditions that have little or no effect on other genotypes. A sensitive genotype might not develop properly if exposed to an environmental condition that wouldn’t bother a tougher sort. On the other hand, a sensitive genotype might profit from an environmental condition that would provide no benefit for an insensitive one.
It’s easy to find real-life examples of the sensitivity kind of interaction and we don’t have to call on mutant fruit flies. Put a child who was born with musical talent into a musical environment and you might get a Mozart. Put a child who was born tone deaf into the same environment and what you will get is an adult who can’t carry a tune. Put a child who has a predisposition to become depressed into a stressful environment and you might get a depressed adult. Put one who lacks this predisposition into the same environment and what you will get is an adult who tells you how stressful her life is but who sounds more annoyed than despondent.10 Your results will vary because what you get depends on what you start with.
Unlike crossover interactions, sensitivity interactions don’t average out to zero. If some genotypes respond to a certain environmental condition and others do not, then what you would get on average is a weaker response—a main effect, though a smaller one.
No Two Alike Page 8