Wednesday, January 25, 2006

10 questions for Judith Rich Harris   posted by Razib @ 1/25/2006 12:42:00 PM

Judith Rich Harris is author of The Nurture Assumption and the forthcoming No Two Alike: Human Nature and Human Individuality. My questions are in bold.

1) One criticism some of my readers made about 'The Nurture Assumption' is that it did not take evolution into account enough, will we see more evolutionary-historical considerations at play in 'No Two Alike'?

Yes, there is quite a lot about evolution and evolutionary history in No Two Alike.

2) Do you believe cognitive psychology has any insights into why people seem to have a strong bias in asserting the overwhelming role of family in the character of a child? Or do you believe that this is a cultural innovation?

Cultural factors are certainly involved. Americans didn't always have this strong belief in the role of the family - in particular, the role of the parents - in shaping a child's personality and behavior. That belief became popular around the middle of the 20th century. Prior to that, when children were troublesome or otherwise disappointing, the general consensus was that they were "born that way."

But there may be a cognitive component as well. There is a cognitive bias that makes people overestimate their own importance and their own ability to influence how things turn out - not just in child-rearing but in everything they do.

3) In 'The Nurture Assumption' you argue that children's peer groups are more influential on their behavior than their parents. One of your key illustrations of this is the fact that children of immigrants quickly acquire the language and accent of their non-immigrant peers. But it might be objected that this is a special case, as children have a specific 'language instinct', in Pinker's sense, which governs their language acquisition. What would you reply to this objection, and do you have any equally good alternative examples of peer-groups prevailing over parents?

The language instinct can explain why the child of English-speaking parents learns to speak English, but it cannot explain why, if this child goes outside and discovers that the people out there are speaking a different language, he not only acquires that new language but comes to favor it over the language his parents taught him - a language he still speaks at home.

But I can give you some examples that don't involve language. Robert McCrae found that there are personality differences between people reared in different cultures. For example, North Americans are somewhat more outgoing and less agreeable, on average, than Asians. McCrae gave personality tests to Asian-Canadian college students, the children of immigrants from Hong Kong. He found that the students who had recently arrived in Canada had personality profiles similar to those of the people back in Hong Kong, but the Asian-Canadians who were born in Canada were similar to other Canadians. Those who had arrived in childhood were somewhere in between. So the culture of the home - the culture the parents brought with them from Hong Kong - wasn't what determined the offsprings' personality. The children who were raised in Canada became Canadians.

My second example has to do with neighborhood effects on behavior. Researchers studied two groups of African-American school-age boys. These children all came from the same kind of home: low-income, headed by single parents. But some homes were located in black, poverty-level neighborhoods, and others were in neighborhoods that were predominantly white and middle-class. The researchers found that the African-American boys living in poverty-level neighborhoods were highly aggressive, but that those living in middle-class neighborhoods were no more aggressive than their white,middle-class peers. In both cases, these children had adapted their behavior to the local norms.

4) In your 2005 response to the Edge Question, "What Do You Believe Is True Even Though You Cannot Prove It?," you alluded to two things, 1) selection for light skin 2) hairlessness by parents in infants. When you pointed to these facts, did you do so in light of recent genetic work which suggests that dark skin might have evolved in humans as a response to loss of body hair? In other words, one trait would never been selected for if not for the other.

No, I hadn't heard of that work. But it doesn't matter. All humans have more or less hairless bodies, so I assume that the characteristic of hairlessness is at least as old as our species - at least 100,000 to 200,000 years old. Racial differences in skin color, on the other hand, are no more than 50,000 years old. If humans turned dark-skinned as a response to hairlessness (a theory I find dubious), then an explanation is still needed for why their skin turned white again so quickly when they inhabited Northern Europe, thousands of years later. My response to the 2005 Edge question offered a possible explanation.

By the way, I've expanded that essay into an article for a journal called Medical Hypotheses. It will be published in a few weeks.

5) Research that compares correlations of adoptive/biological families (mostly done by a handful of American behavior geneticists) typically finds low shared family influence, but research that compares means of adoptive/biological families (mostly done by a handful of French sociologists) typically finds big roles for genetics and shared family. Is correlation a reliable method for saying there is no shared family influence, might means need to be given more weight by behavior geneticists?

You're talking now about the effects of adoption on IQ. First, let me make it clear that all these studies showed a big role for genetics. Second, I agree that American behavioral geneticists might have underestimated the influence of "shared environment" (the environment that siblings raised in the same family have in common) - not because they've ignored means but because the adoptive homes these researchers looked at tended to come from a narrowed range: adoptive parents are generally middle- or upper-middle class. The French researchers, on the other hand, made a special effort to include lower-class families in their studies, and hence found a larger influence of shared environment. These results, by the way, are consistent with those from the behavioral genetic study of reared-apart identical twins: no influence of shared environment on personality (the correlation between the reared-apart twins was the same as that between the reared-together twins), but a small influence of shared environment on IQ (the IQ correlation was higher for the reared-together twins).

But I have a quarrel with the way you phrased your question: you said that correlational studies typically find "low shared family influence." What the researchers actually find is low influence of the shared environment. The environment shared by reared-together siblings doesn't just include the family: it includes the neighborhood, the school, the ethnic group, and the socioeconomic class. Sometimes siblings even belong to the same peer group. In other words, reared-together siblings share a culture or subculture.

My interpretation of the IQ data can explain both the means and the correlations. Here's how it goes. The family does have an effect on IQ during childhood. If the parents use big words or do various other things that increase a child's vocabulary, the child will score higher on IQ tests. But research has shown that this advantage - measured as an effect of shared environment - is temporary: it gradually fades away and is gone by late adolescence.

What isn't temporary is the advantage given by a culture (or subculture) that fosters intellectual activity. At higher socioeconomic levels, there tends to be a greater awareness that things like reading and going to science museums are good things to do and might pay off in the long run. So socioeconomic class does have a long-term effect on IQ. This is a cultural (or subcultural) effect and results in a difference in means: adoption tends to raise a child's IQ because most adopted children are raised in middle- or upper-middle-class neighborhoods.

A similar cultural effect can explain the gradual increase in average IQ scores that has occurred in the last 75 years all over the world. All over the world, socioeconomic levels have gone up and people are more aware than they used to be that intellectual activities might pay off in the long run.

6) Has behavior genetics declared the death of shared environment prematurely without considering levels of "shared environment" that occur above family - neigborhood, city, state, country, etc? Also if these things matter (which seems indisputable) and are mediated by shared family (which seems indisputable), again are the correlations hiding important details of parental influence?

No, not at all. Most behavioral geneticists are well aware that "shared environment" can mean the environment siblings share outside the home, rather than (or in addition to) the family environment. For example, behavioral geneticists have found an effect of shared environment on teenage delinquency. But, as behavioral geneticist David Rowe showed, the evidence suggests that the relevant environment is the neighborhood or school shared by teenage siblings. Siblings close in age may belong to the same peer group, and Rowe found that the shared environment effect on delinquency is larger for siblings close in age.

I see no justification for saying that the effects of shared environment are "mediated by the shared family." There are things that may in fact be mediated by the shared family - cooking styles and religious denomination spring to mind - but for most of the things that behavioral geneticists have studied, the shared environment should not be equated with the family environment.

You ask if correlations might be "hiding important details of parental influence." Perhaps what you're getting at here is the notion that parents might influence one of their children one way and another child in a different way. For example, the parents' child-rearing style might cause one sibling to become more outgoing and bold, the other to become more timid. If the direction of the effect depends on the preexisting (genetic) characteristics of the child, then what you've got is a gene-environment interaction. There's a whole chapter (Chapter 3)in No Two Alike devoted to gene-environment interactions. I show why they can't account for twin and sibling differences in personality.

But perhaps when you ask whether correlations might be "hiding important details of parental influence," you are talking about sheer unpredictability: the notion that parents do have an effect, but there's no way to predict in advance what the direction of the effect will be. Developmental psychologist Ellen Winner used this notion to explain away the behavioral geneticists' findings, in her response to the 2005 Edge question. "To demonstrate parents' effects on their children," Winner said, "we will need to recognize that parents may influence their children to become like them or to become unlike them." Winner suggested that researchers should study adult adoptees "and look at the extent to which these children either share their adoptive parents' values or have reacted against those values. Either way (sharing or reacting against), there is a powerful parental influence."

It's a heroic attempt to preserve the faith in parental influence, but a futile one. What does it mean to say that parents do have a powerful influence but that the direction of the influence is unpredictable? Is there any way to prove or disprove that statement? Does it have any scientific value? For that matter, does it have any practical value? Would parents be satisfied to be told, "Yes, your parenting will have an effect on your children, but we can't tell you what that effect will be"? It would mean that books of child-rearing advice would have to begin with a disclaimer: "If you follow this advice, your children might turn into happy, successful people; on the other hand, they are just as likely to turn into miserable failures."

7) OK, to something serious, east coast vs. west coast, is there any comparison in weather?

Not according to my older daughter, who lives in Berkeley. Whenever I complain about the snow, ice, or cold here in New Jersey, she points out that where she lives, the weather is "sensible."

8) How far do you go with 'modularity' in 'No Two Alike.' I ask because one of the questions of interest in behavior genetics is variation within a population. On the other hand evolutionary psychologists tend to emphasize human universals and the 'psychic unity of mankind,' often rooted in a paradigm of massive mental modularity which assumes that cognitive organs are fixed genetically (monomorphic) and not subject to non-pathological variation.

I go pretty far with modularity. I don't think it's possible to give a satisfactory description of social and personality development in childhood without thinking in terms of a modular mind. Simple theories of social development don't work because the human mind isn't simple!

You're right that the behavioral geneticists are mainly interested in human differences, whereas the evolutionary psychologists are mainly interested in human universals. But that distinction is starting to crumble. In his book The Blank Slate, evolutionary psychologist Steven Pinker has a chapter (Chapter 19) devoted to individual differences.

As for the idea that "cognitive organs are fixed genetically...and not subject to non-pathological variation," I think it's nonsense. There is variation in all our essential organs. Why should my language acquisition module be identical to yours if there are differences in our hearts, lungs, kidneys, arms, and legs?

9) If you had to pick one thing, what do you think has been the most important finding from cognitive neuroscience which psychologists have had to take into account in formulating their theories?

In cognitive science, I would definitely pick modularity. But in psychology in general, I think the most important finding is the behavioral geneticists' discovery that the environment doesn't work the way everyone expected it to. Shared genes, as expected, make people more alike; but shared environment, to everyone's surprise, hardly ever makes people more alike. To put it another way, having different environments - growing up in different homes, being reared by different parents - isn't what makes people differ from one another. So what does make them differ? That's the mystery I try to solve in No Two Alike.

10) If you could have your full genome sequenced for $1000, would you do it? (assume privacy concerns are obviated)

I'd jump at the chance, and I wouldn't give a damn about privacy concerns - I'd want the information to be made freely available. My father spent his adult life crippled by an autoimmune disorder called ankylosing spondylitis. His father died young of an autoimmune disorder called pernicious anemia. And I have been ill most of my adult life with an autoimmune disorder that has launched attacks on several different body systems. So I think my genes might have something interesting to tell medical researchers.