No Two Alike

by Judith Rich Harris

  I conclude that within-the-family differences in environment, of which the most conspicuous are the differences associated with ordinal position in the family, are a red herring. They are not the perpetrator we are looking for. They cannot account for the personality differences between ordinary siblings, much less the differences between identical twins.

  That was my fourth red herring. Only one more to go, and I can eliminate it quickly—no need to give it a chapter of its own.

  The fifth red herring is gene-environment correlations. Don’t confuse them with gene-environment interactions; they’re not the same. A gene-environment correlation occurs when a person’s environment reflects, or is influenced by, the person’s genes. As the behavioral geneticist Sandra Scarr put it, “Individuals make their own environments, based on their own heritable characteristics.”81 People’s environments differ in part because their genes do.

  Examples abound. Children born pretty or charming are given special attention by their parents and are treated well by other people too. Those who are hungry for knowledge are provided with books and taken to museums. Those who have an aptitude for schoolwork tend to choose studious friends and intellectually challenging activities. Some people are more likely than others to experience stressful life events, perhaps because they have a knack for putting themselves into situations (jobs, romantic relationships) that are likely to end badly. Like prettiness and academic aptitude, the tendency to experience stressful life events shows significant heritability.82

  The correlation between genotype and environment sometimes comes about because of the way others react to an individual’s heritable characteristics, sometimes because of the active role the individual plays in picking and choosing among the available options. The upshot is that environments are not shuffled and dealt out randomly: they are to some extent predictable on the basis of genotype. Children with different genotypes therefore grow up in different environments.

  Gene-environment correlations produce what I call “indirect genetic effects”: the effects of the effects of the genes.83 Good-looking people and those who are big and strong get used to having their way and become more assertive. Children born bright and curious are attracted to intellectual activities and the result is that they get smarter. Those born with nasty dispositions evoke negative reactions from others and that makes them nastier still. Indirect effects of genes often, though not always, work in this vicious-circle fashion.

  Standard behavioral genetic methods cannot distinguish between direct and indirect effects of genes; the data analyses lump them together. Therefore, “heritability” includes indirect effects of genes, as well as direct effects. This is the source of the critics’ complaint that twin studies overestimate heritability: the observed similarities of identical twins may be partly due to indirect effects of genes. Identical twins are very similar in physical appearance, temperament, intelligence, aptitudes, and interests; and these similarities evoke similar reactions from other people and cause the twins themselves to make similar choices in deciding how to spend their time. If these reactions and choices have any systematic effects on the twins’ personalities, the effects will contribute to the portion of the variance attributed to heritability, not to the unexplained variance.

  Which means that, though gene-environment correlations may inflate heritability estimates, they can’t be the solution to the mystery of individuality. They make identical twins more alike, not more different. Robert Plomin made the same point in “Nonshared Environment a Decade Later.” He and his coauthors asked whether gene-environment correlations could account for nonshared environment effects. “No,” they said, “these can’t account for why identical twins are different.”84

  And yet there are theories of personality development based on gene-environment correlations, proposed by eminent behavioral geneticists such as Sandra Scarr and Thomas Bouchard.85 Another example is the theory proposed by the psychiatrist David Reiss in his book about the NEAD study. Reiss was unhappy with the results of the study and wanted to restore the family environment to what he felt was its rightful place. He thought maybe gene-environment correlations could do it. For example, he said,

  Heritable personality traits of adolescents may disrupt their relationships with their parents. These disrupted relationships may be a crucial step in the process whereby the personality traits evolve into serious antisocial behavior.86

  A more straightforward explanation would be that the disrupted relationships and the serious antisocial behavior both result from the heritable personality traits. This interpretation fits the data just as well as the one Reiss suggested. But it doesn’t matter for my purposes, because neither can account for the unexplained variance. The disrupted relationships are the indirect effects of genes, so they would contribute to genetic variance. To heritability.

  What Reiss was doing was trying to explain how genes work. That’s also what Sandra Scarr was trying to do, and Thomas Bouchard. Bouchard stated his purpose clearly in the subtitle of one of his articles: “How Genes Drive Experience and Shape Personality.”

  My purpose is different. I’m not interested in how genes shape personality: I’m interested in how the environment shapes personality. Here’s what Bouchard said in that article about explaining the variation in personality that is not genetic: “How non-traumatic environmental determinants influence the normal range of variance in adult personality remains largely a mystery.”87

  Now I will start putting together the clues that point to the solution of the mystery.

  5

  The Person and the Situation

  CLEARLY, THE PERPETRATOR I’m searching for is an elusive one. This perpetrator produces variation in personality above and beyond the variation produced by genes. It creates or widens personality differences even between identical twins who grow up together.

  The perpetrator I’m searching for is a shaper of personality. To solve the mystery of individuality, I have to figure out how personality gets shaped. I’m proceeding on the premise that the shaping serves some purpose—that evolution provided humans with a certain amount of plasticity in behavior so they can profit from their experiences. People differ in personality, not just because they have different genes, but also because they have different experiences.

  The way people profit from their experiences is through a process we call learning. Any theory that says that people’s personalities are modified by their experiences is, implicitly or explicitly, a theory of learning. Thus, in order to figure out how personality is shaped by experience, it is necessary to have a clear idea of how people learn. In particular, since most of the shaping occurs in the first couple of decades of life, it is necessary to have a clear idea of how young people learn.

  I’ve examined and eliminated five plausible suspects. Personality differences, as I showed in chapter 2, are not mainly due to differences in environment, nor to a combination of “nature” (genes) plus “nurture” (the part of the environment provided by parents). Nor, as I demonstrated in chapter 3, can they be explained in terms of gene-environment interactions. The remaining candidates were crossed off in chapter 4: environmental differences within the family and gene-environment correlations.

  Five red herrings. None of them can be the solution to my mystery. All make predictions that are inconsistent with the evidence; none can explain the differences between identical twins. But there are plenty of fish in the sea. What about all the other theories of personality development?

  Sorry, they won’t work either. Before you dismiss that statement as sheer chutzpah, listen to my reason for making it. I can eliminate all the currently popular theories of personality development with a single flick of my hand, because they all rest on the same basic assumption about learning. The assumption is that learned behaviors or learned associations transfer readily and automatically from one situation to another. What all these theories have in common is the idea that children learn something in one environmental setting (usually t
he home), or with one social partner (usually the mother), and that this learning subsequently affects the way they behave, and the emotions they feel, in other places and with other people.

  Those who make this assumption often place particular stress on very early experiences, the idea being that whatever is learned first will affect all subsequent experiences. Psychoanalytic theory is an obvious example and so is its grandchild, attachment theory. Modern attachment theorists believe that if you and your mother didn’t happen to hit it off while you were an infant, you are doomed to plod through life with the self-fulfilling expectation that all your relationships will be disappointing. The first prominent behaviorist, John B. Watson, also placed much importance on learning in infancy, though he would have attributed your later disappointments to the reflexes that got conditioned during that time. Even Jean Piaget, the famous Swiss developmentalist and cognitive scientist, believed in the importance of early learning and its subsequent generalization. Here’s a telling quote from Piaget:

  Day to day observation and psycho-analytic experience show that the first personal schemas are afterwards later generalised and applied to many other people. According as the first inter-individual experiences of the child who is just learning to speak are connected with a father who is understanding or dominating, loving or cruel, etc., the child will tend…to assimilate all other individuals to this father schema. On the other hand, the type of feelings he has for his mother will tend to make him love in a certain way, sometimes all through life, because here again he partially assimilates his successive loves to this first love which shapes his innermost feelings and behaviours.1

  I am not, of course, the first to question this view of the role of early learning and generalization in personality development. In 1968, Walter Mischel—then a professor of psychology at Stanford, now at Columbia University—published a book that knocked personality psychology on its ear. The resulting uproar became known as “the person-situation controversy.”

  It was Mischel who drew my attention to Piaget’s statement on the generalization of “first personal schemas.” Mischel quoted the part that had to do with the father schema and then cited evidence showing that people’s relationships with their fathers bore little or no resemblance to their relationships with other authority figures: “For example, attitude toward one’s father correlated .03 with attitude toward one’s boss.”2

  Other evidence that Mischel cited had to do with moral behavior and moral judgments. Children who learn not to lie or cheat at home may lie or cheat in the classroom or in games on the playground; children who are honest in the classroom may cheat on the playground. Whether children cheat when given the opportunity to do so bears little relation to what they say on paper-and-pencil tests of moral judgment. How they respond on a test of moral judgment depends in part on where they are when they take the test. “The data on moral behavior,” Mischel concluded, “provide no support for the widespread psychodynamic belief in a unitary intrapsychic moral agency like the superego.”3

  Mischel’s point can be stated more simply: people behave differently in different situations. This, he said, has serious implications for personality psychology.

  The intuitive conviction that persons do have consistent and widely generalized personality traits seems very compelling…. Obviously a person does not have to learn everything afresh in each new or slightly different situation, and his reactions to any novel situation are influenced by his prior experiences. The empirically established behavioral consistencies, however, do not seem large enough to warrant the belief in very broad generalized personality traits…. The conviction that highly generalized traits do exist may reflect in part (but not entirely) behavior consistencies that are constructed by observers, rather than actual consistency in the subject’s behavior.4

  Mischel was saying that the consistency of personality is an illusion. It’s in the eye of the beholder, not in the head of the beholdee. There is such an illusion; I mentioned it in chapter 1: the fundamental attribution error. I gave the example of the subjects who believed that the graduate student who was nasty to them in the laboratory would also be nasty if they ran into her in the supermarket. The fundamental attribution error, I said, is a human universal. It’s an illusion that everyone has, like the visual illusions you sometimes see on the back of cereal boxes—this one, for example:

  All five spots are the same shade of gray, but we see the one on the pale background as dark, the one on the dark background as pale. As Steven Pinker remarked in How the Mind Works, our visual systems do not provide this illusion so that cereal manufacturers will have something to print on the back of the box. This illusion, and the others he discussed in his book, are the manifestations of mental mechanisms that serve a purpose. In this case the purpose is to identify objects. Is this a lump of coal or a lump of snow? The problem is that the amount of light reflected by an object depends on the overall illumination: “A photographer’s light meter would show you that more light bounces off a lump of coal outdoors than off a snowball indoors.” The visual system cleverly compensates for the overall illumination, with the result that we see the coal as black and the snowball as white whether we are indoors or out. The illusion is a product of the visual mechanism responsible for what cognitive scientists call “lightness constancy.”

  Lightness constancy occasionally causes us to make mistakes, but what it does under ordinary circumstances is to help us identify objects correctly. Because information from the senses is incomplete and imperfect, “the brain,” Pinker explained, “must combine information from the senses with assumptions about the world to make an intelligent guess.” Occasionally the world violates our assumptions and the result is an illusion, but most of the time our guesses are correct. That’s why Pinker called them intelligent guesses.5

  The fundamental attribution error is an illusion that results from what might be called “personality constancy.” We are endowed with a mental mechanism that causes us to see a given individual’s personality as relatively constant—more constant than a mechanical measuring device would indicate—whether we see this individual indoors or out, in a laboratory room or in the supermarket. The same way we attribute blackness to a lump of coal and whiteness to a snowball, we attribute friendliness to one person and nastiness to another.

  The universality of the fundamental attribution error suggests that this bias helped, rather than hindered, our ancestors. Evidently it didn’t lead them astray more often than not. The reason it didn’t is that people are, at least to some extent, consistent in how they behave. There is such a thing as personality.

  Walter Mischel didn’t deny it. He just said that people are much less consistent than you have been led to believe.

  Mischel comes from a school of psychology called cognitive behaviorism: behaviorism with some added frills. In other words, he’s an environmentalist. His book—the one that shook up personality psychology—was published in 1968, a time when environmentalism was running rampant and there weren’t any behavioral geneticists to kick around (if there were, they were keeping awfully quiet). Mischel knew nothing about the behavioral genetic evidence I told you about in chapter 2 because the studies hadn’t been done yet.

  In denying that people have consistent personalities, Mischel was denying not only that early learning generalizes but also that people have inborn predispositions to behave in certain ways—in other words, that there are genetic influences on personality. He was arguing against the view that “the same underlying disposition (or ‘genotype’)” may affect behavior in a variety of situations.6

  As we now know, Mischel was wrong about there being no underlying dispositions. But he was right about the other things: that people behave differently in different situations and that “personal schemas” do not generalize to other persons.

  The fact that people behave differently in different situations is one of the things I talked about (I devoted an entire chapter to it) in The Nurture Assumption.7 But I
never claimed that there was no correlation at all between the way an individual behaves in two different situations. Nor did Mischel make that claim. Mischel’s point was simply that the correlation is low. My point was that it is surprisingly low, given the behavioral genetic evidence that genes account for about 45 percent of the variation in personality and social behavior.

  Your genes you always have with you. They go with you wherever you go. People born with a predisposition to be agreeable or aggressive or shy will take these predispositions with them into a variety of social contexts. I touched on this in the previous chapter when I mentioned the study that found a correlation of .19 between children’s obnoxious behavior with their parents and their obnoxious behavior with their peers.8 I pointed out that the researchers hadn’t considered the possibility that some children might inherit a predisposition to behave in a disagreeable or aggressive fashion wherever they go, and that this predisposition might account for the correlation between home behavior and playground behavior.

  Behavioral geneticists can now look into such possibilities, thanks to a technique, developed within the past dozen years, called “multivariate genetic analysis.” It enables them to assess the degree to which genetic and environmental influences each contribute to the correlations found between behavior in different contexts or at different times. Using this technique, researchers studying children’s shy or inhibited behavior found that genetic influences account for almost the entire correlation between behavior in the laboratory and in the home. When a child behaves in an inhibited (or uninhibited) fashion both in the laboratory and at home, it’s because he or she has a genetic predisposition to behave that way. The same is true of activity level—how physically active or inactive the child is. Genetic influences account for the entire correlation between activity level in the laboratory and in school.9 The correlation between activity level in the laboratory and in school, incidentally, was .19.