Book Review: The Nurture Assumption, by Judith Rich Harris.

The Nurture Assumption isn’t quite the book I expected. Amazon tells me that customers who bought it also bought The Blank Slate, by Steven Pinker, and The Bell Curve, by Charles Murray, which is what I expected. Based on that, you might expect it’s a book about how human nature is influenced by genes rather than the environment, possibly with a racist chapter or two thrown in there. But Harris is less interested in the relative influences of nature and nurture than in changing how we look at the idea of “nurture.” As Steven Pinker says in the introduction, “We all take it for granted that what doesn’t come from the genes must come from the parents.” This book is about why that assumption is wrong.

This is not going to be a neutral review. I think that Harris is more or less correct, so I’m going to try to present her case as persuasively as possible. And I’ll start by talking about a paper that hadn’t been published when Harris wrote her book. You’ve probably heard that the nature versus nurture debate is over and that it’s a draw –  this article uses that phrase exactly – in that most human traits are about 50% heritable. Or,

human nature = 50% genes + 50% environment

But let’s take a few steps back – twin studies are a powerful tool, but it’s important to be extremely precise about what they do and don’t mean, because it doesn’t necessarily line up exactly with our intuitive sense of what’s “nature” and what’s “nurture.” Take a look at the abstract of the meta-study that article is based on: “The data are inconsistent with substantial influences from shared environment or non-additive genetic variation.” Come again?

Rounding things to nice, even numbers, the study says the following:

differences1 in traits between people in the study population
= 50% additive differences in genes + 0% non-additive2 differences in genes + 35% differences in non-shared3 environment + 15% differences in shared4 environment

That’s a bit different from our original equation, right?

  1. The word “differences” shows up a bunch of times. We can’t use twin studies to study the commonalities in human nature. And “differences” in this case specifically means “differences within the population that was studied.” So twin studies can’t tell us much about why humans in the United States are different from humans in China, or why humans in the 20th century are different from humans in the 18th century.That’s not especially relevant to this book review, but it’s worth keeping in mind.
  2. The difference between additive and non-additive genetic effects is completely irrelevant to this post so let’s ignore it.
  3. “Nonshared environment” means all the environmental things a pair of twins doesn’t share with each other – basically, their own idiosyncratic and unpredictable experiences of life, friendships, random events – and free will, if you believe in that sort of thing.
  4. “Shared environment” means all the things a pair of twins share with each other that they don’t share with society as a whole – their parents, their neighborhood, their school, and so on. The abstract says these factors don’t play a “substantial role”, but there’s a number close to 15% buried deep in the article.

So what does this tell us about “nature versus nurture”? It still seems like “nature” is about 50%, but whether the other 50% is “nurture” depends on precisely how we define that word. I think everyone agrees that parents count as “nurture”, but it’s less clear whether “unpredictable experiences of life” count as “nurture” – that sounds more like “dumb luck.” In fact, we’ve just learned that parents, schools, and neighborhoods taken together account for only 15% of individual differences – which means that the purest form of “nurture” – parenting – counts for even less on its own. And that’s basically the thesis of The Nurture Assumption – that parenting plays only a very small role in what kinds of people children grow up to be.

Harris’s argument can be roughly divided into two parts:

  1. The “parenting doesn’t matter” chapters.
  2. The “peer groups do matter and here’s how” chapters.

I. Parenting doesn’t matter very much.

There are, of course, many studies that seem to show that parenting matters – parents with criminal records are likely to raise children who go on to have criminal records, and so on. What almost all these studies have in common is that they do not control for genetics, so we can’t tell whether the similarities between parents and children are caused by parenting or by shared genes.

But there are many studies that control for genes. One typical example is a Danish adoption study of the criminal records of adopted children and their adopted and biological parents. For the sake of brevity, I’ll follow Harris’s lead and call families where at least one parent has a criminal record “criminal families”, however reductive that may be. I’ll also speak of children having criminal records even though I actually mean they later grow up to have criminal records.

  1. Children with non-criminal adopted families and non-criminal biological families were 14% likely to have a criminal record.
  2. Children with criminal adopted families and non-criminal biological families were 15% likely to have a criminal record.
  3. Children with non-criminal adopted families and criminal biological families were 20% likely to have a criminal record.
  4. Children with criminal adopted families and criminal biological families were 25% likely to have a criminal record.

Depending on exactly how you slice and dice this, it looks like having a criminal biological family makes a child about one-and-a-half times as likely to have a criminal record, whereas having a criminal adopted family makes a child somewhere between one-and-a-tenth and one-and-a-quarter times as likely to have a criminal record. This suggests that the genetic influence of biological families is much more important than the environmental influence of adopted families. It also suggests that Harris is probably overstating her case a little – she seems to literally believe that parenting has no influence, but this study and many others seem to show that parenting has a least a small influence – it’s just much smaller than the influence of genes, and – more on this in the next section – the influence of things in the environment that aren’t parents.

But first I’ll mention two important caveats, one that Harris acknowledges and one that she dismisses. The caveat she acknowledges is that parents can have a very large influence on what someone’s childhood is like – which is different from saying they have a large influence on what kind of adult someone grows up to be. Parents may not shape their children, but they do care for them, and so long as we think a happy childhood matters for its own sake – not just for the way it shapes adulthood – then parents should try to provide that.

The caveat Harris dismisses is…well…an article in Slate memorably calls it the “weird shit rule.”  Twin and adoption studies can tell us a lot about the effects (or lack thereof) of parenting within the normal spectrum of parenting behavior. “Weird shit”, like beating children severely, raising them in a cult, or whatever it is you want to call the way James Mill raised his son are rare enough that they simply won’t show up on a population-wide study. Harris doesn’t find this argument persuasive; she says it would be a strange coincidence if the parenting methods that are too rare to study happen to be the only ones that make a difference. But it’s not a coincidence, is it? Feeding your kid lead paint chips is rare in large part because everyone knows it can mess up their adult life.

(An aside: I have met plenty of people who were raised in cults, and most of them seemed just fine.)

II. Peer groups do matter and here’s how.

Harris believes that environment matters. For example, in the case of the Danish study, she points out that regardless of biological or adoptive parents, children raised in Copenhagen were more likely to have criminal records than children raised in rural areas. Harris is especially interested in one specific type of environmental influence that she believes has been neglected by researchers – socialization by peer groups. By “peer group”, Harris has in mind the following things:

  1. “Peers” are other children of the same age or slightly older.
  2. Same-sex peers are more influential than opposite-sex peers, especially before puberty. Children have a much stronger tendency to segregate by sex than adults do, and they are especially likely to identify with strict gender roles when they interact in mixed groups of boys and girls.
  3. Children are socialized mostly by peer groups, not by relationships with individual peers. Specifically, it is the sense of “groupness” – of identification with a certain group defined against others – that drives socialization.
  4. Children are socialized into “childrens’ culture”, not the culture of adults in their society. Sometimes the norms of childrens’ culture are different from the norms of adult culture; as noted before, one example is that childrens’ culture tends to enforce stricter gender roles.
  5. The peer groups that socialize children may belong to subcultures within the broader childrens’ culture, and some of these subcultures value behaviors that society sees as dysfunctional.

Unlike the “parenting doesn’t matter” chapters, which are built upon mountains of evidence using established research techniques like twin and adoption studies, the “peer groups matter” chapters are heavy on speculation. So I’m not willing to say “Harris is right” about this part – she presents a plausible story that draws on accepted psychological and sociological theories, but there’s not much empirical research out there that’s directly related to the ideas she’s talking about.

There is one area where Harris seems to be on firm ground: She’s especially interested in how the children of immigrants are socialized, because they’re the closet thing she has to a natural experiment – raised by parents of one culture and peers of another. The most straightforward example is language; I remember my astonishment as a child when I met some of my parents’ friends, Americans who lived in Australia – the parents spoke English with the same accent I did, but their children talked like Australians! Even at a young age, I had absorbed the “nurture assumption” to the point where finding a clear counterexample – children whose accents were influenced by their peers rather than their parents – shocked me.

She argues that it’s not just language that works this way, but socialization in general. Some immigrant parents try very hard to socialize their children in the ways of the old country, but they are usually only partially successful – their children typically learn “code-switching”; for example, speaking Spanish at home but speaking English with their peers. Indeed, the only cases where immigrant parents can be fully successful in passing down their culture is when there are enough other immigrants from the same country around that their children can form their own “childrens’ subculture”, defined in opposition to the mainstream childrens’ culture.

III. Should you believe this?

As I said before, I think Harris’s account is more or less correct. That’s not so much because of the strength of her arguments, but because I’m aware of evidence like the meta-study I mentioned near the beginning.  The “15% shared environment” figure seems to put a ceiling on how much influence parenting can have – with some caveats like the aforementioned “weird shit rule.”

It’s not clear to me exactly how controversial the book’s thesis is. Harris herself claims that it’s extremely controversial, but she seems to be the sort of person who wears controversy like a badge of honor; her website has a page filled with quotes about her theory, many of them negative – including this gem from Camille Paglia:

“I have indeed dutifully looked at Harris’ book several times in stores, but quite frankly, I found it too rambling, anecdotal and contradictory to purchase or take seriously. . . . My understanding of the formation of adult personae comes from wide-ranging sources, such as Sir James George Frazer’s survey of tribal rites of passage in ‘The Golden Bough’; Freud’s conflict-based analysis of the psychodynamics of ‘family romance’ in bourgeois society; and even Babylonian astrology.”

But it seems to me that the overall reception to her book has been positive. The original article she based the book on won a prestigious award from the American Psychological Association, and the press has covered her favorably – though many people note that she has probably overstated her case a bit. And the public in general seems receptive to the idea that our society is too hard on parents, and that they should relax and worry less about screwing up their kids – one comment, when I linked to a related article on Facebook, referred to “middle class handwringing” about childrearing.

Research on peoples’ beliefs about heredity seems to bear that out – that strong belief in the power of parenting is indeed most common among the white, educated, upper-middle class, and interestingly, least common among the parents of adopted children, who have the most direct relevant experience.  Also, overall belief in the importance of heredity is very high.

The only sustained criticism of Harris that I could find online is from developmental psychologists whose professional role involves giving parenting advice – see, for example, this exchange from Slate, between Harris and Jerome Kagan (Harris, futilely, wants the debate to be between the influence of parenting and peers, but Kagan shifts it to parenting and genes. Harris can’t really have the debate she wants while the issue of heredity is so controversial.)

I notice that almost all of Kagan’s counterarguments involve studies on the psychological traits of children who still live with their parents. The Dickens-Flynn model is built around the observation that IQ is much more heritable in adulthood than in childhood – which suggests that parenting has a fair amount of influence on children who are still living with their parents, but that influence declines when the children grow up and move away. It seems reasonable that the same thing would be true of other psychological traits as well, which could explain why the studies Kagan cites show more influence from parenting than most of the studies included in the meta-study. In general, it seems to me that Harris and Flynn share similar views of human development – that human traits are influenced a great deal by genetics and culture, but not very much by families.

Kagan also nitpicks that validity of twin studies, and here he makes an elementary mistake; he points out that some genetic effects are not additive, but rather, depend on interactions among multiple genes. That’s true, but it doesn’t help Kagan’s case; there is no reason to believe that incorporating non-additive genetic effects would increase our estimates of shared environmental effects – the additive term excludes non-additive effects, so if anything, adding non-additive effects to the equation is likely to lower our estimate of environmental influences.

Sloppiness of this kind is very common when people argue against heredity. I could complain about that, or I could assume that there is a rational basis for it – people fear that incorrectly believing in the influence of heredity, and disbelieving in the influence of parenting, could have harmful outcomes, so they tend to judge evidence for those views against extremely high standards, while letting weak arguments against heredity slide. This seems sort of reasonable, if the consequences are much worse for making a mistake in one direction rather than the other.

The usual fear regarding hereditarian theories is that their implications might be racist. That could be an issue for some theories that pit genes against the environment as a whole, but I think quite the opposite is true for a theory like Harris’s that pits genes and peer influence against parenting. The anti-hereditarian alternative risks arguing that people from groups that have worse social outcomes are bad parents.

On the flip side, I think people tend to underestimate how harmful it is to overestimate the influence of parenting, and I think that properly understood, rejecting the “nurture assumption” leads to a more humane and progressive understanding of human nature.

IV. Why the nurture assumption is harmful.

I mentioned before that the white, educated upper middle class are the group least likely to believe that heredity influences human behavior. Now, maybe that’s a good thing – these people are educated, after all, so perhaps their beliefs are based on better evidence. But it’s also possible that their beliefs are biased in ways that devalue the parenting practices of other cultures.

Different societies have had an extremely wide variety of child-rearing practices, without having obvious negative impact on their children (not so long ago, middle class norms entrusted children to nannies with magic umbrellas.)  And maybe we’re willing to say that we do things better than those old-timey folks did, but the comparisons become uncomfortable when we look closer to home.  For example, certain groups in our society, such as black parents or Chinese immigrants, are more likely to spank their children than white parents are. There’s a large body of research showing that childhood spanking is correlated with violent behavior in adulthood; anti-hereditarians tend to believe this means that childhood spanking is harmful to a child’s psychological development in a way that causes adult violent behavior.  This position leads to a straightforward, but awkward conclusion: Black parents are wrong and white parents are right.

Naturally, people are reluctant to accept this conclusion. But trying to reject that conclusion without rejecting its main premise leads to the most delicate of balancing acts – arguments that basically claim that it’s white peoples’ fault that black people spank their children, but nevertheless black people need to give up their cultural practices and adopt the cultural practices of upper-middle class whites.

There’s a much easier way to avoid blaming black parents, though.  Studies of spanking that account for genetics find that it has no measurable effect – positive or negative – on long-term outcomes (though “harsh physical punishment” does seem to cause long-term harm – see again the “weird shit rule.”)  Suddenly, spanking by black parents seems less like a blameworthy problem to explain away and more like a cultural difference that we should be extremely careful about judging.

The topic of “judging” brings me to another problem with the nurture assumption. I’m thinking of a scene from the But I’m A Cheerleader, in which a (conservative, religious) therapist tries to suss out why a girl is a lesbian:

Therapist: “Megan, it’s your turn to report out your root.”

Megan: “I think it might be my parents.”

Therapist: “Okay, go with that.”

Megan: “You know…we’ve kinda been like this greeting card family. And then there was that one year where…well…Dad was unemployed and Mom had to support us.”

Megan’s Dad: “Wait a minute, that was only for nine months. And then I was offered a much better job at the plant.”

Megan: “Maybe seeing Mom kind of being the Dad, maybe l maybe I got the wrong idea about the roles of men and women.”

Therapist: “Absolutely. I can’t believe that you didn’t mention this earlier. Your father was emasculated; your mother was domineering.”

Megan: “Well, not exactly.”

Therapist: “You wanted to emulate your mother. You have no respect for men, because you don’t respect your father.”

Likewise, Harris points to the case of Bruno Bettelheim, a psychologist (and apparent all-around asshole) who claimed that autism is caused by a mother’s coldness toward her child.

Nowadays, mainstream society doesn’t believe that parents influence their childrens’ sexual orientations or autism spectrum disorders, and we understand that believing otherwise is cruel and regressive. For a variety of reasons, it’s obvious to most of us that buying into the nurture assumption when it comes to these two things is not only incorrect but also harmful. Now, it’s possible that sexual orientation and autism are the exceptions, and that most other traits reakky are heavily influenced by parenting – but I think these examples show that such an assumption can be costly if it turns out to be wrong.

My final objection to nurture assumption is the most subtle, and to me – speaking as someone who is not a parent but one day hopes to be – the most moving. Harris says she’s often asked if her theory means we may as well treat our children callously. And she responds by asking, “Should we treat our romantic partners callously?” After all, we play only a very small role in shaping what kind of person our romantic partner becomes. And yet we know that we should not treat them callously, because a happy relationship with them is intrinsically valuable, here and now, not merely as a means for mutual improvement. Likewise, the relationship between a parent and a child, and the actions a parent takes to provide a happy childhood, are intrinsically valuable here and now, not merely a means for producing a happy and functional adult.

One formulation of Kant’s categorical imperative says that we must always act in such a way that we treat humanity, whether in our own person or in the person of any other, never merely as a means to an end, but always at the same time as an end. The nurture assumption can all too easily treat childhood, and especially the relationships between adults and children, as a means to an end – adulthood – rather than an end in itself.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s