Saturday, 29 October 2011

Steven Pinker's account of the evidence: The surprisingly small long-term effects of parenting

This post is going to be mostly quotes from the chapter "Children" in Steven Pinker's 2001 book The Blank Slate. All the non-italicised text is taken from the book, with the italics being my commentary. The numbers in brackets refer to items on the alphabetic list of references at the end of this post.

The Blank Slate has been the starting point for my recent research on the long-term effects of parenting, which has led to radical shifts in my world view. I now think that I have allowed myself to be seriously misled by (probably) mostly well-meaning people for a number of years. I wish I had studied the evidence sooner.
I think this is very important, so I want to give you the account of the evidence as presented in the book, and I want to give you a bunch of references, because I think I am fortunate to have a few friends who are intent on scrutinising this research less than mercilessly.

All the quotes, except for the very last one, are from the first 9 pages of the chapter. I'm quoting these 9 pages very liberally, at the risk of pissing off the publisher. I sincerely hope that I am doing them a favour, though. The chapter is 28 pages long and has a lot more on this topic that should interest you, which I'm not going to cover here. And then there's the rest of the book… It's 434 pages long (I have the British Penguin paperback edition) and full of fascinating stuff (including an extremely interesting chapter on the psychology of politics). I also want it known that Steven Pinker has a new book out, called The Better Angels of Our Nature: The Decline of Violence in History and Its Causes, which I think should be of great interest to anyone interested in psychohistory. Also, if you want to get an idea of what I mean by "the science of psychology" which I've contrasted on this blog with the "psychotherapists talking from their experience" kind of "psychology", then there is no book I could recommend more highly than Steven Pinker's How the Mind Works.
I've been a fan of the man for many years.

Go Steve:

"THE NATURE-NURTURE DEBATE is over." So begins a recent article with a title – "Three Laws of Behavior Genetics and What They Mean" – as audacious as its opening sentence. (28) The nature nurture debate is, of course, far from over when it comes to identifying the endowment shared by all human beings and understanding how it allows us to learn, which is the main topic of the preceding chapters. But when it comes to the question of what makes people within the mainstream of society different from one another – whether they are smarter or duller, nicer or hastier, bolder or shyer – the nature-nurture debate, as it has been played out for millennia, really is over, or ought to be.


[In that article,] Turkheimer (…) was summarizing a body of empirical results that are unusually robust by the standards of psychology. They have been replicated in many studies, several countries, and over four decades. As the samples grew (often to many thousands), the tools were improved, and the objections were addressed, the results, like the Star-Spangled Banner, were still there.


Here are the three laws:
·         The First Law: All human behavioral traits are heritable.
·         The Second Law: The effect of being raised in the same family is smaller than the effect of the genes.
·         The Third Law: A substantial portion of the variation in complex human behavioral traits is not accounted for by the effects of genes or families.

As to the different kinds of tests used to measure such traits, he writes:

It sounds dodgy, but the tests have been amply validated: they give pretty much the same result each time a person is tested, and they statistically predict what they ought to predict reasonably well. IQ tests predict performance in school and on the job, and personality profiles correlate with other people's judgements of the person and with life outcomes such as psychiatric diagnoses, marriage stability, and brushes with the law. (4,7,15)

Here he gives some basic explanations of the methodology used in such studies:

Once the measurements are made, the variance of the sample may be calculated: the average squared deviation of each person's score from the group mean. The variance is a number that captures the degree to which the members of a group differ from one another. (…) It is mathematically meaningful to say that a certain percentage of the variance in a group overlaps with one factor (perhaps, though not necessarily, its cause), another percentage with a second factor, and so on, the percentages adding up to 100. The degree of overlap may be measured as a correlation coefficient, a number between -1 and +1 that captures the degree to which people who are high on one measurement are also high on another measurement. (8) (…)
Heritability is the proportion of variance in a trait that correlates with genetic differences. It can be measured in several ways. (21) The simplest is to take the correlation between identical twins who were separated at birth and reared apart. (…) Alternatively, one can compare identical twins reared together, who share all their genes and most of their environment, with fraternal twins reared together, who share half their genes and most of their environment. (…) The bigger the difference between the two correlations, the higher the heritability estimate. Yet another technique is to compare biological siblings, who share half their genes and most of their environment, with adoptive siblings, who share none of their genes (among those that vary) and most of their environment.


The results come out roughly the same no matter what is measured or how it is measured. Identical twins reared apart are highly similar; identical twins reared together are more similar than fraternal twins reared together; biological siblings are far more similar than adoptive siblings. (1,2,3,10,19,21) All this translates into substantial heritability values, generally between .25 and .75. A conventional summary is that about half the variation in intelligence, personality, and life outcomes is heritable – a correlate or an indirect product of the genes. It's hard to be much more precise than that, because heritability values vary within this range for a number of reasons. (21)

(Please refer to the book or some other source to learn about those reasons.)


The heritability of intelligence, for example, increases over the lifespan, and can be as high as .8 late in life. (14,22) Forget "As the twig is bent"; think "Omigod, I'm turning into my parents!"


"All traits are heritable" is a bit of an exaggeration, but not by much. (1,2,3,9,11,17,19,
26) Concrete behavioral traits that patently depend on content provided by the home or culture are, of course, not heritable at all: which language you speak, which religion you worship in, what political party you belong to. But behavioral traits that reflect the underlying talents and temperaments are heritable: how proficient with language you are, how religious, how liberal or conservative. General intelligence is heritable, and so are the five major ways in which personality can vary (summarized by the acronym OCEAN): openness to experience, conscientiousness, extroversion-introversion, antagonism-agreeableness, and neuroticism. And traits that are surprisingly specific turn out to be heritable, too, such as dependence on nicotine or alcohol, number of hours of television watched, and likelihood of divorcing. Finally, there are the Mallifert brothers in Chas Adam's patent office [this is a reference to a cartoon] and their real world counterparts: the identical twins separated at birth who both grew up to be captains of their volunteer fire departments, who both twirled their necklaces when answering questions, or who both told the researcher picking them up at the airport (separately) that a wheel bearing in his car needed to be replaced.


Behavioral genetics does have its critics, who have tried to find alternative explanations for the First Law. (…)
These possibilities have been tested, and though in some cases they may knock down the heritability estimate by a few points, they cannot reduce it by much. (21) The properties of adoptive parents and homes have been measured (…), and they are not homogeneous enough to force identical twins into the same personalities and temperaments. (3,18,19,20) Identical twins are not earmarked for homes that both encourage twirling necklaces or sneezing in elevators. More important, the homes of identical twins who were separated at birth are no more similar than the homes of fraternal twins who were separated at birth, yet the identical twins are far more similar. (3,16) And most important of all, differences in home environments do not produce differences in grown children's intelligence and personality anyway (as we shall see in examining the Second Law), so the argument is moot.
As to contact between separated twins, (…) the amount of contact turns out to have no correlation with the twins' degree of similarity. (2,3) What about the expectations of parents, friends, and peers? A neat test is provided by identical twins who are mistakenly thought to be fraternal until a genetic test shows otherwise. (…) the twins are as alike as when the parents know they are identical. (25) And direct measures of how similarly twins are treated by their parents do not correlate with measures of how similar they are in intelligence or personality. (10) Finally, sharing a placenta can make identical twins more different, not just more similar (since one twin can crowd out the other), which is why studies have shown little or no consistent effect of sharing a placenta. (2,5) But even if it were to make them more similar, the inflation of heritability would be modest. As the behavioral geneticist Matt McGue noted of a recent mathematical model that tried to use prenatal effects to push down heritability estimates as much as possible, "That the IQ debate now centers on whether IQ is 50% or 70% heritable is a remarkable indication of how the nature-nurture debate has shifted over the past two decades." (13) In any case, studies comparing adoptees with biological siblings don't look at twins at all, and they come to the same conclusions as the twin studies, so no peculiarity of twinhood is likely to overturn the First Law.


To say that the heritability of intelligence is .5, for example, does not imply that half of a person's intelligence is inherited (whatever that would mean); it implies that half of the variation among people is inherited.


By now you appreciate that our genes play a role in making us different from our neighbors, and that our environments play an equally important role. At this point everyone draws the same conclusion. We are shaped both by our genes and by our family upbringing: how our parents treated us and what kind of home we grew up in.

Not so fast. Behavioral genetics allows us to distinguish two very different ways in which our environments might affect us. (20,21) The shared environment is what impinges on us and our siblings alike (…) The nonshared or unique environment is everything else (…) including parental favoritism (…).

The effects of the shared environment can be measured in twin studies by subtracting the heritability value from the correlation between the identical twins. The rationale is that identical twins are alike (measured by the correlation) because of their shared genes (measured by the heritability) and their shared environment, so the effects of the shared environment can be estimated by subtracting the heritability from the correlation. Alternatively, the effects can be estimated in adoption studies by looking at the correlation between two adoptive siblings: they do not share genes, so any similarities (relative to the sample) must come from the experiences they shared growing up in the same home. A third technique is to compare the correlation between siblings reared together (who share genes and a home environment) with the correlation between siblings reared apart (who share only genes).
The effects of the unique environment can be measured by subtracting the correlation between identical twins (who share genes and an environment) from 1 (…). By the same reasoning, it can be measured in adoption studies by subtracting the heritability estimate and the shared-environment estimate from 1. In practice all these correlations are more complicated, because they may try to account for nonadditive effects, where the whole is not the sum of the parts, and for noise in measurements. But you now have the basic logic behind them.

So what do we find? The effects of shared environment are small (less than 10 percent of the variance), often not statistically significant, often not replicated in other studies, and often a big fat zero. (1,6,20,24,29) Turkheimer was cautious in saying that the effects are smaller than those of the genes. Many behavioral geneticists go farther and say that they are negligible, particularly in adulthood. (…)
[emphasis added]

Where do these conclusions come from? The actual findings are easy to understand. First, adult siblings are equally similar whether they grew up together or apart. Second, adoptive siblings are no more similar than two people plucked off the street at random. And third, identical twins are no more similar than one would expect from the effects of their shared genes. As with the First Law, the sheer consistency of the outcome across three completely different methods (…) emboldens one to conclude that the pattern is real.


The studies exclude cases of criminal neglect, physical and sexual abuse, and abandonment in a bleak orphanage, so they do not show that extreme cases fail to leave scars. Nor can they say anything about the differences between cultures (…). In general, if a sample comes from a restricted range of homes, it may underestimate effects of homes across a wider range. (27)


Behavioral geneticists have found that their samples of parents in fact span a full range of personality types. And even if adoptive parents are unrepresentative in some other way, the Second Law would survive because it emerges from large studies of twins as well. (3,20,23,24) Though samples of adoptive parents span a narrower (and higher) range of IQs than the population at large, they cannot explain why the IQs of their adult children are uncorrelated, because they were correlated when the children were young. (18,20,21)

About the Third Law:

Concretely, this means that identical twins reared together (…) are far from identical in their intellects and personalities. There must be causes that are neither genetic nor common to the family that make identical twins different and, more generally, make people what they are. (1,20,24,28,29)

A handy summary of the three laws is this: Genes 50 percent, Shared Environment 0 percent, Unique Environment 50 percent (or if you want to be charitable, Genes 40-50 percent, Shared Environment 0-10 percent, Unique Environment 50 percent). A simple way of remembering what we are trying to explain is this: Identical twins are 50 percent similar whether they grow up together or apart. Keep this in mind and watch what happens to your favorite ideas about the effects of upbringing in childhood.
[emphasis added]

Finally, about all the parenting advice you've heard according to which all sorts of features of parenting are essential in determining how the child will turn out:

But surely this advice is grounded in research on children's development? Yes, from the many useless studies that show a correlation between the behavior of parents and the behavior of their biological children and conclude that the parenting shaped the child, as if there were no such thing as heredity. [Note that this applies to the ACE Studies.] And in fact the studies are even worse than that. Even if there were no such thing as heredity, a correlation between parents and children would not imply that parenting practices shape children. It could imply that children shape parenting practices. (6,12)

1. Bouchard, T.J., Jr. 1994. Genes, environment, and personality. Science, 264, 1700-1701. What looks like the whole article is available for free here.
6. Harris , J.R. 1998a. The nurture assumption: Why children turn out the way they do. New York: Free Press. The link is to the 2009 revised edition of the book. 38 pages of the original book are available for free here.

1 comment:

  1. This meshes with my own experience - the rough, anecdotal comparison of me and my 10-years-older sister. There are definite similarities between us, but in general we are very unlike each other. The unique environment appears to have made her a socialdemocrat and me a pacifist market anarchist, go figure. Thinking further, we shared an environment in completely different stages of our lives, 0-8 years for me, 10-18 for her. But I have no idea how that plays out in the end.

    I'd like a definition of the variation between personality traits that's being talked about: the thing that is said to be caused in equal parts by genes and unique environment. In other words, short of my own familial experience I'm having a hard time relating these numbers to reality. What do the variations look like in real life? If part of a sample has an IQ of 100, and another 130, would the difference be explained 50/50 by genes and unique environment, or what? Statistics was my worst subject in uni.

    Sorry for the rambling comment.