Message #1: Oct. 28, 1998
From: Judith Rich Harris
To: Jerome Kagan

Dear Jerry,

You wouldn't know it from the way it has been depicted in the media, but my book The Nurture Assumption covers a diversity of topics. I talk about how cultures are passed on and what makes them change, and why girls and boys behave differently in some circumstances, and how adolescents turn into adults and lose their annoying ways, and why humans have such a strong tendency to identify with a group and to hate the members of other groups, and why the "me" that looks out at the world from behind my eyes still feels 20 even though the eyes are 60 years old. But none of this gets mentioned in the media storm surrounding my book. All the attention is focused on one idea, which journalists insist on overstating as "Parents Don't Matter."

Yes, I'm the terrible grandmother from New Jersey -- the gadfly who has taken on the academic establishment and has been ferociously attacked by some of its members (few of whom have actually read the book). I'm the one who says that parents have no power to mold their child's personality -- that parents have no important long-term effects on the way their children behave when they're not at home. And I should tell you right off, Jerry, I'm not here to repent of my sins. My head is bloody but unbowed.

In your attacks on me on National Public Radio and in the Boston Globe, you accused me of ignoring all the evidence that indicates that parents do have important long-term effects on their children. You implied that I must be ignorant of that evidence. But I've spent almost 20 years studying it. As a writer of textbooks in developmental psychology (the senior author of a text that went through three editions), I spent many years telling credulous college students about that evidence. I told them about your work, too, Jerry!

But now I've looked more closely at the evidence -- looked at it under a microscope, to turn one of your metaphors around -- and I've seen that it is full of holes. It isn't just that there is ample evidence against the nurture assumption, it's that the evidence supporting the nurture assumption is so embarrassingly weak. If it weren't for the fact that researchers take the idea of parental influence as a "given," they would never accept evidence of this sort as proof of anything. Yes, there is an awful lot of it, but no matter how high you pile it, horse manure never turns into gold.

Developmental psychology rests upon data that come almost entirely in the form of correlations, and correlations are ambiguous. Epidemiology has the same problem. When epidemiologists find a correlation between exercise and good health, they like to think it shows that exercising causes good health. They know there is a possibility that the connection works the other way -- that healthy people may be more likely to exercise -- but they assume that at least some of the correlation must be due to what they're looking for: an effect of exercise on health.

But assuming just isn't good enough. Ambiguous results must be backed up with evidence from other kinds of studies. When all you've got is correlations, you have to home in on the truth from different directions, using a variety of techniques.

In developmental psychology, all you've got is correlations. And when I look at other kinds of studies, they do not back up your interpretation of the correlations. Take, for example, the correlation you mentioned in the Boston Globe: "that the best predictor of a child's verbal talent is the frequency with which the parents talk and read to the child."

It is perfectly true that talkative, bookish parents tend to have children who score high on vocabulary tests. But this is an ambiguous result, because correlations don't come with labels saying, "X caused Y." Parents and children tend to be alike partly for biological reasons -- children inherit some of their characteristics from their parents. Although most developmental psychologists admit this, they nevertheless make the assumption that some of the correlation between verbal parents and verbal children must be due to what the children learned at home.

The problem is, if we do a different kind of study -- look at adopted children, for example -- we find no support for this assumption. On the contrary. By the time they reach late adolescence, adopted children do not resemble their adoptive parents in any measure of intelligence. The adolescent reared by talkative adoptive parents is no more verbal, on the average, than the one reared by taciturn parents.

The problem with your evidence, Jerry, is that a whole bunch of causes and effects are tangled together, and the methods of developmental psychology provide no way to tease them apart. Before you can conclude that the parents are influencing the children, you have to separate parental influence from other possible explanations of your results. Children resemble, for genetic reasons, their biological parents. Parents are more likely to read to children who want to be read to. Children and parents are usually members of the same culture or subculture, and the culture influences both the parents and the kids.

Like you, I believe that kids are influenced by their culture. But you assume they get their culture from their parents, and that's an assumption I don't buy. I hope to take this up with you in the next round of our debate.
Till then,
Judy Harris

Dear Judy:

This reply to your essay deals directly with your claim that parents have no power to mold their child's personality and that they exert no important long-term effects on the child.

First, let us clear up the meaning of correlations. Astrophysics deals only with correlations, yet it has been able to infer deep principles. Moreover, watch out or you will be damaging your argument. Your claim that parents have little effect is based on correlations, albeit low ones. If correlations are no basis for inferring anything important then you, too, are on weak epistemological grounds.

I begin the rebuttal with the August 1998 issue of Child Development, our premier journal. The National Institute of Child Health and Human Development research group reports the results of a study of over 1,000 children from 10 different cities who were exposed to very different forms of surrogate care, along with children raised only at home. When the children were assessed at age 3, there was very little effect of form of surrogate care, while family influences were the major cause of variation among the children. The authors wrote, "What transpires in the family appears to be more important in explaining children's early social and emotional development than ... quality, quantity or stability of type of care."

You claim that the reason children of parents who talk frequently to their child are more verbal is attributable to genes. However, this argument is vitiated by a study published in Science in 1978. Schiff, et al., studied working-class children who were adopted early into upper-middle-class homes and compared the intellectual talents of these adopted children with that of their siblings who remained with their working-class parents. The adopted children had significantly higher intelligence scores than the nonadopted children who were genetically related to them. The differences in cognitive skills had to be due to the family practices and not to genes. (See Schiff, et al., Science, 1978, 200: 1503-1504.)

Perhaps the most serious source of vulnerability in your position comes from the fact that children from different cultures behave very differently even before peers have had a chance to have a serious effect. The work of the Whitings has proved this point to the satisfaction of most scientists. I trust that you will not claim that rural Mexican children are more nurturant and less aggressive than New England children because of heredity.

A third flaw in your argument comes from studies of young children orphaned by war -- World War II and the Korean War. These children were adopted by nurturant families and years later their intense anxiety and retarded cognitive abilities were muted in a serious way. There is no other way to account for this result without attributing power to family practices. (See Rathbun, et al., American Journal of Orthopsychiatry, 1958, 28: 408-415; Winick, et al., Science, 1975, 190: 1173-1175.)

Finally, I believe you ignore one of the most important influences on children; namely, the identification with the parents. I could find no reference in your book to any study that dealt with this process. A child with an incompetent, unemployed, impulsive parent is ashamed of that parent and his or her image of self is affected by that shame. On the other hand, the child with a competent, talented, reflective parent feels a certain pride that has a benevolent effect on personality development. Most memoirs ascribe extraordinary power to these processes. (See J.S. Mill; C. Darwin; F. McCourt.)

I know of no culture that claims that parental character and practices have a minimal effect on children. I take that universal fact to mean that a truth is contained there. The fact that your review of the literature led you to a very different conclusion suggests that your review was selective, indifferent to data that were not consistent with your position (see Schiff, for example), and perhaps colored by the a priori bias you arrived at for reasons that remain obscure.

Jerry

Dear Jerry,

Let me begin with your assertion that I reviewed the literature selectively on the basis of my a priori bias.

Five years ago, my bias was the same as yours: I too was a believer in the nurture assumption. Up to that point I had focused on the literature of developmental psychology. The change in my thinking came after I read more widely in psychology and anthropology, in preparation for writing another textbook. It gave me a bird's-eye view of the field. Sometimes things can be seen from a middle distance that aren't visible from close up. I've noticed that psychologists who are not focused on a single area but who are able to see the bigger picture -- writers of introductory textbooks, for instance -- tend to be more receptive to my ideas.

Yes, there are behavioral differences between Mexican children and New England children. No, I wasn't going to claim that they are due to heredity. But surely you're not implying that you would object to such a claim, in view of the fact that you've attributed the behavioral differences between Chinese babies and Irish babies to differences in heredity!

You're way ahead of most developmentalists, Jerry, in your willingness to admit that genes play a role in how children behave. But even developmentalists who are willing to admit it are strangely reluctant to admit that, since children inherit their genes from their parents, genes can produce resemblances between children and their parents.

That study you mentioned from the August 1998 issue of Child Development? There are hundreds like it. The method provides no way of controlling for the effects of heredity. When researchers do use the appropriate controls, heredity is always found to make a contribution. Therefore it is scientifically indefensible to go on doing research that provides no way of controlling for these effects. The study also provided no way of separating the effects of the parents' child-rearing style on the child from the effects of the child on the parents' child-rearing style, even though we know that the parents' behavior is partly a response to the child's behavior.

Nevertheless, this study did produce an interesting result: No differences were found between children who entered day care in early infancy and those who spent their first three years at home. Which brings me to what you said about correlations. I never claimed that correlations are "no basis for inferring anything important": I said they are ambiguous. If you find that two things tend to go together, you don't know what caused that correlation. You need additional information to explain it. But if you discover that two things are not correlated, you've learned something. If the amount of time children are cared for by people other than their parents is unrelated to the children's behavior and adjustment, that tells us something important, doesn't it? Doesn't it tell us that a lot of the advice being given to parents is hogwash?

The differences between adopted children reared in middle-class homes and those reared in working-class homes do show up in studies that control for heredity (see Capron and Duyme in Nature, Aug. 17, 1989). But you were going well beyond the data when you said, "The differences in cognitive skills had to be due to the family practices and not to genes." There is a third alternative you haven't considered: They could be cultural differences, like the difference between the Mexican kids and the ones from New England.

When children reared by middle-class parents live in middle-class neighborhoods and go to middle-class schools -- which is almost always the case -- you can't tell whether the effect on their cognitive skills is due to the home environment or to the environment outside the home. The problem is that the two environments match. You have to look at cases where they don't match to see what's really going on. You have to look at cases where the parents' culture differs from that of most of the other families in the neighborhood.

I describe many such cases in my book. The children of immigrants: They adopt the language and culture of their peers. The hearing children of deaf parents: They adopt the language and social behavior of the hearing culture, not those of the deaf culture. Kids from fatherless, low-income families who live in middle-class neighborhoods are as unaggressive as their middle-class peers, whereas kids from similar families living in low-income neighborhoods are highly aggressive (see Child Development, April 1995). When there is a cultural discrepancy between the home and the neighborhood, the child starts out with the home culture and ends up with the neighborhood culture. But in most cases there is no discrepancy between them, and no need to change. Almost all your data come from those uninformative cases.

The idea that parents have important and lasting effects on their children -- is it universal? No. See Chapter 5 of my book. But even if it were universal, would that make it true? Didn't everyone used to believe that the Earth was flat?

One of our problems is that you seem to think of the study of children as art or literature, and I think of it as a science. You talk about what people say in their autobiographies and what they tell pollsters who ask, "Do parents matter?" I happen not to believe that we can answer scientific questions by means of public opinion polls.

Judy

Dear Judy:

Your last reply makes two points. The first is that a child's heredity always makes a contribution to his or her behavior and mood. The second claim is that similar neighborhoods usually produce similar psychological profiles because of interactions with peers. Both of these claims are broad and remain premises rather than proven facts once one specifies particular profiles. Each is an idea to be tested empirically. I now deal with each of the claims in turn.

The bold proposition that heredity always makes a contribution to a child's psychological profile is as general and, therefore, as empty of significant meaning as the statement that environments always make a contribution to a child's profile. Therefore, your demand that all investigators who study the role of the environment must control for a child's heredity invites the equally reasonable suggestion that all investigators who study the role of heredity must measure the child's environments. As you know, behavioral geneticists do not do so. No report describing the influence of heredity on any psychological trait, whether it be IQ or a personality characteristic, has directly measured the psychological environments of the subjects over the first dozen years of life. The complex influences of the environment are inferred from an equation called the heritability equation. But these inferences are not accurate for, as you know, the equation that estimates the heritability of a trait contains terms not only for genes and environment but also for the interaction of genes with environment and the influence of all the background genes possessed by a particular individual. This process is given the technical name epistasis. Neither of the latter two influences is awarded a reasonable value in the current heritability equations.

Thus, we have a situation in which the equation that estimates the heritability of a trait has four separate terms: genes, environment, interaction of genes and environment, and epistasis, but the latter three are never directly measured, leaving us with unknowns for three of the four terms. For that reason, all current estimates of the heritability of any psychological trait must be suspect.

Equally relevant is the fact that most studies of the heritability of personality traits in children are based on parental report, and in adults on information from self-report using questionnaires. In the few cases where actual observations of children were made, a paradoxical result occurred. When mothers describe their children's degree of shyness on a questionnaire, estimates of the heritability of shyness increase with age. However, when actual observations of children are made, estimates of the heritability of shyness go down with age. Every conclusion concerning the magnitude of heredity's influence is limited in significance for it depends always on the source of the evidence.

One finding on fruit flies should motivate reflection among those who claim that genes exert profound influences on personality. A pair of alleles in fruit flies produce abnormal wings in offspring if the two parents possess the genes for this anatomical anomaly. But this result only occurs if the flies are reared in a laboratory where the temperature is about 20 degrees Centigrade. If the temperature of the laboratory is raised by only 10 degrees, the offspring develop almost normal wings. If a difference of 10 degrees in room temperature can influence a significant anatomical feature of a fruit fly, surely children's personality traits, which are controlled by a much larger number of genes, must be influenced, in a major way, by the environments in which they are reared. (See W.J. Keaton and J.L. Gould, 1993, Biological Science.) Listen to a leading behavioral biologist, Gilbert Gottlieb who writes in the recent edition of the Handbook of Child Psychology, "When there is heredity-environment interaction, the degree of apparent heritability of some characteristic depends on the specific rearing environment."

This means that every study you cite and interpret as showing the influence of genes is seriously dependent on the rearing environments of the individuals. The vast majority of studies of the heritability of personality are conducted on middle-class, white children and adolescents, and we have no idea of the magnitude of heritability of any trait in other groups, for example, economically disadvantaged Vietnamese immigrant children or children from elite families living in Kuwait. Your book is written as if you do not appreciate that the heritability of any trait can never be a fixed quality; it is always at the mercy of the rearing environment.

Finally, the estimates of heritability you quote assume that the effects of genes and environments are additive. For readers who are not familiar with this idea, it means that the distribution of a particular trait, say children's verbal ability, can be predicted by assuming that the influence of the genes for verbal skill are added to the influence of environmental supports for verbal ability.

This assumption raises a serious problem because most biological characteristics are not additive. For example, the electroencephalogram profiles of identical and fraternal twins are best explained as the result of an interaction among the person's genes (epistasis) rather than by assuming that genetic and environmental forces combine additively. (See J.C. Christian, et al., 1996, Psychophysiology, 33: 584-591.) Even the simpler phenomena we call "size of a snowfall" is not predicted very well by adding the influence of the temperature to the influence of humidity.

Thus, your evaluation of the power of genes to influence a child's personality is permissive and not sufficiently self-critical.

Your second claim is that neighborhoods play an important role in children's development. I found this conclusion surprising for I know of no study that has shown, for example, that half of children living in the same neighborhood are essentially similar in their perceptual, memorial, and verbal abilities or in the constellation of personality traits. Consider as a thought experiment a single block on East 75th Street in Manhattan, where children from different cultural backgrounds live in the same small area. If you sampled 100, 15-year-olds on that block, you would find extraordinary variability in any psychological trait you decided to measure.

There is a study that supports this hypothetical thought experiment. Most of Warsaw was destroyed by bombs during World War II. When the city was rebuilt after the war, authorities decided that families with different educational and vocational backgrounds should live in the same apartment houses so that their children would play on the same playgrounds and attend the same schools. Your position suggests that these children should be very similar in their abilities. But the data revealed a great deal of variability in cognitive qualities, and this variability could be attributed to the values of their families. The children's intellectual talents were a function of their family's influence, not of the peers with whom they interacted everyday.

You state on Page 317 of your book that we do not know why children who are abused by their parents often grow up with psychological problems. You note correctly that this outcome could be due to parental treatment, parental personality, genes, or peers. Your conclusion is correct not only for the outcomes of abuse but also for every trait and talent that psychologists study. We simply do not know the differential contribution of the varied influences that can affect a particular profile. The proper posture to this frustrating state of affairs is reflection and reserve, and not your opening statement in the preface that claims boldly that you wished to dissuade readers of the notion that a child's psychological profile is shaped by parents. It is hard for me to understand how you can be so certain of the minimal influence of parents when you acknowledge on Page 317 that scientists don't know why parental abuse is likely to produce problems in adolescence. Let me be clear: No one, whether a scientist or an educated citizen, understands the complex ways in which genes, prenatal effects, family practices, parental personality, children's interpretations of their environments, historical era, size of city and, yes, even peers come together to produce any psychological trait. Under these conditions, every scholar should resist temptation to cry, "Fire!"

Dear Jerry,

OK, you've made your point -- I'm sorry I accused you of being unscientific. But is this the place to talk about epistasis and the wing anomalies of fruit flies? Why don't we save the technicalities for a specialized journal, where I'd be happy to go on debating you till the last fruit fly wears out its telomeres.

Better still, I could find you a behavioral geneticist to argue with. This is not my quarrel. My interest is not primarily in heredity. I'm interested in the environment -- in how the child's environment shapes the child. It was you, not I, who wrote an entire book on inborn differences in temperament. "Thus," you said in it, "inhibited and uninhibited behaviors are heritable" (Galen's Prophecy, Page 168). The reason I didn't bother explaining that heritability depends upon the population measured is that it's not relevant: My book is not about heritability.

Heredity concerns me mainly in a negative way: I want to control or eliminate its effects, the better to judge the effects of the environment. My point was that the studies you use to support your position, including the one from the August 1998 issue of Child Development that you mentioned in your first posting, use methods that cannot disentangle the effects of heredity and environment. You don't have to take my word for it: The authors of that study admitted it themselves! (See Page 1,168 of that article.)

Nor is it enough to disentangle genetic effects from environmental ones: We also need to disentangle environmental influences from each other. That often isn't done either. You needled me for admitting, on Page 317 of my book, that we don't know why children who are abused by their parents grow up with psychological problems. But the reason we don't know is that researchers haven't done the work: They haven't separated the effects of parental abuse from other environmental factors. For example, children who are physically abused by their parents are often treated badly by their peers as well, because risk factors that increase the chances that a child will become a victim in the home -- slow development, a difficult disposition -- also increase the chances that he or she will become a victim on the playground. When this unfortunate individual ends up with psychological problems, psychologists assume that the parental abuse was to blame. No one has bothered to find out whether being picked on or rejected by peers might actually be a better predictor. It's simply taken for granted that what happens at home must matter more.

What really disappoints me is the evidence that, even though you've been attacking my book for two months now, you still haven't looked at more than a few odd pages here and there. People who've read the book tell me they found it very interesting -- even fun. The Nurture Assumption is about how evolution shaped the human mind and how the child's mind, as a result of that evolutionary history, is predisposed to conceptually divide people into categories -- male/female, kid/grown-up, good reader/bad reader -- and to put the self into one of the categories. "Self-categorization," Australian social psychologist John Turner calls it. Children identify not with their parents but with a group of people they see as being "like me." Under ordinary conditions they don't see a grown-up as being "like me" because grown-ups are in a different social category, with different rules of behavior. Identification with a parent is real but comes much later -- it doesn't happen until the offspring can plausibly be included in the same social category as the parent. The embarrassment children feel when their parents don't measure up is not due to identification, at least not in the sense you mean. It's the same embarrassment they feel when they have to go to school carrying a Mickey Mouse lunchbox when everyone else has a Lion King lunchbox.

I never said that peer groups homogenize their members. Groups make their members more alike in some ways but also act to widen the differences among them. Of course living in the same neighborhood doesn't turn children into little clones -- neither does living in the same family!

My theory of child development is new, speculative, and largely untested. But it does a good job of explaining a lot of annoying little things that don't fit into traditional theories. It explains why only children, who bear the full brunt of their parents' attention, do not differ in personality from children with siblings. Why generations of upper-class British boys, reared by nannies and sent to boarding schools at the age of 8, turned out so much like their fathers. Why girls and boys behave as differently in egalitarian societies as in sexist ones, and why these differences show up in some situations and not in others. Why well-behaved children become rebels in their teens and then solid citizens in adulthood. And why the children of immigrants end up with the language and accent of their peers, not of their parents -- and not something in between, either.

Yes, my theory is still speculative: It has not been proved. But what you are attacking me for is not my theory. What you are attacking me for -- the assertion that parents have little or no long-term influence on the personality, intelligence, or mental health of their children -- is not new, is not mine, and is not speculative. It is firmly grounded in evidence. The evidence is of many different kinds and has been accumulating for nearly 20 years. It indicates that, across a wide range of households, variations from one household to another -- including variations in how the parents rear their children -- have no measurable effects on how the children turn out.

I am by no means the first to point out the implications of these findings -- see David C. Rowe's excellent book, The Limits of Family Influence -- nor will I be the last. David Cohen, a University of Texas professor with a background in clinical psychology, is currently putting the finishing touches on a book called Stranger in the Nest: Do Parents Really Shape Their Child's Personality, Intelligence, or Character?

Stopping me -- telling people that I don't know what I'm talking about -- will not help your cause. This is not a leak that can be plugged: It's the beginning of a deluge. I know this must be upsetting to someone who has put in 45 years laboring under what has turned out to be a false premise. I don't blame you for being mad at me. But killing the messenger will not stop the message.

What I'm really shouting is not "Fire!" -- it's "Hey, the emperor isn't wearing any clothes!" If that message sounds to you like "Fire!" it's not surprising. After all, Jerry, you're one of the tailors.

Judy

Dear Judy:

I was a bit saddened by the ad hominem tone of your last reply. I accepted the invitation from Slate because I believed you came to a mistaken conclusion, despite honest intentions.

I have only three brief points to make in this final reply. First, I continue to believe that you selected the evidence and ignored the data inconsistent with your premise. Each year's literature provides ample support for the effect of family experience. The November 1998 issue of Developmental Psychology has two papers (Pages 1,233 to 1,245 and 1,450 to 1,458) pointing to the significant influence of parental behavior on children and youth.

Second, you argued that peers play a major role in sculpting significant personality traits (e.g., introversion, sociability, impulsivity, conscientiousness, and others). I know of no study that has shown this claim to be true. Peers influence dress code, musical tastes, and dialect but not major personality profiles. And as I indicated in an early reply, peers do not exert power until the child is 6 or 7 years old.

Finally, you acknowledge on Page 361 of your book a premise held by a majority of developmental scholars; namely, most of the information children process and store is unconscious. The problem for you is that almost all the studies you cite did not measure these unconscious structures, which include the shame and pride children carry with them because of the emotional identification with parents. You write, "The bond between parent and child lasts a lifetime." You add, to my surprise, the suggestion that readers not blame their parents for their personality profile. I agree. But it does not follow from a restraint on ascribing blame to family that the family had no influence. Of course, adults should not blame their parents because, in most cases, parents did the best they could. The ethical imperative to accept oneself is independent of the scientific fact that families sculpt part of the adult profile.

Happy Thanksgiving.
Jerry