No Two Alike

by Judith Rich Harris

  If you think my theory is unnecessarily complex, just wait till you see what the theories will be like fifty years from now.

  “Scientists do not conduct research to find things whose existence they don’t suspect,” the evolutionary psychologists John Tooby and Leda Cosmides noted.26 It is my hope that this book will inspire scientists to conduct research on three mental systems whose existence they might not have suspected, or suspected only vaguely.

  Perhaps I might make some suggestions. Brain-imaging studies have proved very useful in research on how the mind works: the activation of different mental processes lights up the brain in different ways. This technique provides one way to test my theory. What happens in the brain when, for example, people are given different kinds of feedback regarding social success or failure? In a recent experiment, subjects were given information that made them think they were being excluded from a group. Brain scans done immediately afterwards showed activation of two areas in the brain (the anterior cingulate cortex and the right ventral prefrontal cortex) that are also activated by physical pain; the conclusion was that being excluded hurts.27 Well, a demotion in status also hurts, as does finding out that someone you like doesn’t like you. But perhaps these things hurt in different ways. My theory predicts that feedback on group acceptance, on status, and on success or failure in a relationship should produce somewhat different responses in the brain.

  A second method that has been useful in the search for mental mechanisms is the examination of individuals whose brains aren’t working in the usual fashion, due to inborn abnormalities or to damage. The subjects who gave Baron-Cohen the idea for his model of a mindreading mechanism were children with autism. Such children lack the ability to read people’s minds.

  Unfortunately, the disabilities of children with autism cannot be used to distinguish the three systems I have proposed because the deficits are too pervasive; all three systems appear to be disrupted. Autism affects every aspect of social life. Children with this disorder aren’t interested in people and don’t form normal relationships. They don’t conform and they don’t compete. The neurological abnormalities that produce such devastating effects must be widespread.

  What about more limited social disabilities? Are there people with malfunctioning relationship systems whose socialization and status systems are working okay? People who don’t get socialized but who are good at competing? Yes, certainly. Researchers have found and reported such disparities and I’ve mentioned them here and there: for instance, the fact that children who are poorly accepted by their peer group may nonetheless be good at making and keeping friends, and vice versa. Folklore supports such distinctions. Sayings like “There’s honor among thieves” imply that people who feel no need to conform to the norms of their society may nonetheless have successful relationships.

  My theory doesn’t predict, however, that these three areas of social life will be completely uncorrelated. Some weaknesses—a problem with language, say, or below-average skill at mindreading—can adversely affect the functioning of two or more systems at once. Other characteristics, such as physical appearance, can have widespread effects on social success or failure. Good-looking people are likely to be high in status and also high in group acceptance.

  Because there are many ways to go about it, I think it will be relatively easy to obtain evidence of the existence of the three systems I have described. What will be more difficult is finding ways to test my solution to the mystery this book is about—the mystery of the unexplained variance in personality, exemplified by the personality differences between reared-together identical twins. To the extent that these differences are environmental in origin, and not the result of developmental noise, I attribute them to the workings of the status system.

  The difficulties involved in testing this theory are illustrated by a true story about a pair of identical twins. Conrad and Perry McKinney, age fifty-six, were featured in an article in the Boston Globe titled “Two Lives, Two Paths.” Born and reared in New Hampshire, the twins did everything together in their early years. They attended the same schools, sat in the same classrooms. Academically they were average students, but they were troublemakers. Eventually their teachers got fed up with their shenanigans and the twins were separated: Perry was held back in fifth grade, Conrad was promoted to sixth. That, according to the Globe reporter, was where their paths diverged. Conrad went on to graduate from high school; Perry dropped out in eleventh grade. Today, Conrad is a successful businessman—as it happens, he runs a private detective agency. Perry…well, Perry is a homeless alcoholic “who sleeps amid trash under a bridge,” by the Piscataqua River in New Hampshire.28

  This story shows, first of all, why ethical considerations make it impossible to test my theory by doing an experiment. We cannot tinker with the lives and futures of human beings. Testing my theory will therefore have to rely on “natural” experiments, provided by the world. The cruel experiment the world performed on Conrad and Perry produced results that were consistent with my theory. The theory predicts that if you change the way the community sees a child, and the change is substantial and persistent, the result will be a change in the child’s personality. When a child is made to repeat a grade in school, it’s a public event that changes the way the community sees him. Thereafter, when the child looks into his classmates’ minds to find out what they think of him, he reads things like “dummy” and “loser.”

  But natural experiments tend to lack scientific rigor. When the decision was made to hold one twin back, did the teachers flip a coin and it came up Perry? Unlikely. More likely, they chose Perry because he was already doing a little less well in his schoolwork than Conrad, or acting a little more troublesome. It’s even possible that Perry was already showing early signs of a mental illness that would worsen over the years.

  Though identical twins are genetically identical (or nearly so), they have slightly different brains due to developmental noise or to minor injuries or infections. These little neurophysiological differences can result in differences in behavior, which means that if we find a behavioral difference between identical twins, it’s usually impossible to tell whether it’s due to the twins’ experiences or to a preexisting difference in their brains. If people treat them differently and they behave differently, is the difference in how they’re treated a cause or an effect of the difference in behavior? Even longitudinal studies can’t answer this question. Researchers may find that differential treatment by teachers or peers at Time 1 is correlated with a difference in the twins’ behavior at Time 2. But the teachers or peers might be reacting to a difference in behavior that already existed at Time 1, though it might not yet have been apparent to the researchers. The fact that some behavioral problems and mental illnesses show up in a mild form in the early years and worsen as time goes on doesn’t prove that the worsening is due to the child’s experiences: neurophysiology can work that way too.

  Another problem in interpreting the story of Conrad and Perry is that being left back in school affects many aspects of a child’s life, not just one. The status system, which keeps its owner informed about what others think of him, might not have been the only perpetrator involved in Perry’s long, slow slide: the socialization system could also have played a role. Left-back kids, though they’re bigger than most of their classmates and hence rank high in the pecking order, tend to be poorly accepted. Kids who are rejected by their peers often get together and form groups of their own—in many cases, antisocial groups. The Globe reporter mentioned that after Perry was left back he “began to accumulate a circle of friends who were rowdier and more risk-taking” than Conrad’s friends.29 This means that socialization by a delinquent peer group could be one of the reasons why Perry ended up the way he did. As it happened, the poor guy didn’t have any luck in relationships, either. His wife died suddenly at the age of twenty-nine from an aneurysm. So all three systems may have conspired to put Perry under the bridge.

  In chapter 4, I poin
ted out that birth order studies are a handy way to test the theory that within-the-family environmental differences are responsible for sibling differences in personality. The virtue of this method is that (as far as we know) there are no systematic biological differences, due either to genes or to developmental noise, between firstborns and laterborns. Thus, birth order studies give us a way of controlling for biological differences while varying within-the-family environment. The cause-or-effect problem is solved, in this case, by a very common natural experiment: siblings born one at a time, in random order. If the birth order theorists had turned out to be right—if systematic differences in adult personality between firstborns and laterborns had been found—we would know that these differences must be an effect and not a cause of the environmental differences.

  Now I’m looking for a way to control for biological differences while varying outside-the-family environment, and it’s a good deal harder to find one. Environmental differences that are unrelated to biological differences tend to be idiosyncratic and nonsystematic, and thus not very useful to researchers. So far, the best I’ve been able to do was to find characteristics such as height and strength that are linked to systematic differences in outside-the-family experiences but that presumably are not genetically associated with particular personality characteristics. “Presumably” is the caveat here. Though theoretically unlikely,30 it is not impossible that people who inherit genes that make them tall or strong may also tend to inherit genes that make them aggressive. Another possibility is that developmental noise may affect these physical and psychological characteristics in similar ways.

  Though my theory is, in principle, testable, these methodological and ethical problems are not going to be easy to solve. Researchers are not likely to come up with a quick and easy way to test the theory. But perhaps it can offer them something better: a new approach, a new way of looking at personality development and social behavior. A viewpoint that can make existing research more understandable and suggest more productive strategies for future research.

  You know by now, because I’ve belabored the point, that research on human behavior is useless if the method doesn’t provide a way of controlling for the effects of genes. The behavioral geneticists have been saying that for years. I agree with them completely, but it’s only a start. The research method also has to be sensitive to the effects of context, because people behave differently in different contexts and with different social partners. Ideally, the research method should not only control for the effects of genes but also assess the role they play in the carryover of behavior from one context to another.

  But controlling for genes and for context is still not enough. The confusion between the effects of the socialization system and those of the status system—between the processes that enable children to adapt to their culture and those that cause each child to develop a unique personality—has long been an impediment to progress. So has the belief that successful relationships are the be-all and end-all of the child’s life. I predict that researchers will make more progress if they attempt to distinguish the effects of the three systems I’ve described, rather than use methods that jumble them together. It was only when I started thinking of three separate systems, providing different motives and collecting different kinds of information from the environment, that I began to understand how evolution could have produced a bunch of conforming individualists like us. What a piece of work is the human being!

  One of my goals in writing this book was to give you a healthy skepticism in regard to research. Researchers are human; they make mistakes; they have their own hopes and needs, their own beliefs. Doing research is a lot of work and is seldom carried out for the sake of pure curiosity. The researcher is earning a living, burnishing a reputation, trying to prove a pet theory (or disprove a competing one), or all of the above.

  Some years ago, students taking a college course in experimental psychology were given the assignment of testing maze learning in rats. The students were told that the rats they would be testing were the products of selective breeding: some of the rats were bred to be quick at learning mazes, others were bred to be slow. Half the students were each given five “maze-bright” rats to test; the others received five “maze-dull” ones.

  The students ran their rats in the maze, recorded the data, and tallied their results. According to the reports the students handed in, the “maze-bright” rats learned the maze significantly faster than the “maze-dull” ones. These rats made fewer errors, ran at a speedier pace, and were less likely to stall at the starting gate.

  The truth is that the rats all came from the same genetic stock and were handed out randomly to the students. The “maze-bright” rats did better because the students who ran them in the maze expected them to do better. Since the rats had no way of knowing how well they were expected to do, the expectation must have exerted its effects on the students. Without realizing it, they must somehow have tipped the scales in favor of the results they were expecting to get. It’s called “unconscious experimenter bias.”31

  How does it work? One possibility in this case is that the students might have treated the rats differently, depending on their expectations for them. Perhaps the students were more favorably inclined toward the “maze-bright” rats and handled them more gently, with the result that these rats were less scared. Bias could also have been introduced in the process of recording data and even in correcting errors. If a student inadvertently wrote down a wrong number, she might have been more likely to notice the error and correct it if it disagreed with her expectations.

  There’s no end of ways in which a researcher’s expectations, or hopes, can influence the outcome of a study. As medical researchers have discovered, the only sure way to avoid experimenter bias is to use strict double-or triple-blinding. The physicians who examine the patients, and the statisticians who analyze the results, aren’t supposed to know whether a given patient received the drug or the placebo.

  Someone who thinks up a new theory is the last person who should be trusted with the job of testing it. A new theory should be tested by independent researchers who aren’t cronies of the theorist and who don’t have an axe to grind.32 It’s division of labor again: proposing theories and doing research to test them are jobs that should be carried out by different entities.

  “A good theory should go in advance of the evidence,” the evolutionary psychologist Geoffrey Miller said in a recent interview. “It should stick its neck out and say, this is how I think the world is, and leave it to other people to test it.”33

  Making a virtue of necessity, I will leave it to other people to test my theory.

  Notes

  CHAPTER 1: AN APPRECIATION OF DIFFERENCES

  1. Reuters, December 12, 2002: “Adult conjoined twins may get surgery” (http://www.cnn.com/); J. Wong and K. Espina, July 8, 2003: “Iranian twins die after historic surgery” (http://dailynews.att.net/); W. Arnold, July 8, 2003: “Iranian twins die after separation surgery” (http://www.nytimes.com/).

  2. Lykken and Tellegen, 1993.

  3. Dyson, 2002.

  4. Arthur Conan Doyle, “The Beryl Coronet” (1892), in Doyle, 1994, p. 274.

  5. Grafton, 1989, p. 3.

  6. Tey, 1951/1977, p. 7.

  7. Tey, 1951/1977, p. 88. I know it’s presumptuous to compare myself to fictional detectives such as Alan Grant and Kinsey Millhone. But after I finished writing this book, I discovered that someone else had made a similar comparison: I’ve been called “the Miss Marple of developmental psychology.” See Wesseling, 2004.

  8. Arthur Conan Doyle, “A Scandal in Bohemia” (1892), in Doyle, 1994, p. 11.

  9. Pinker, 1997.

  10. Pinker, 2002, p. 1.

  11. Napolitan and Goethals, 1979.

  12. Gilbert and Malone, 1995.

  13. Myers, 2002.

  14. Buss, 1995.

  15. Hamilton, 1964. See Dawkins, 1989.

  16. Searby and Jouventin, 2003.

 
17. Trivers, 1971.

  18. Wilkinson, 1990.

  19. Ridley, 1996.

  20. Garcia and Koelling, 1966; Garcia, McGowan, and Green, 1972.

  21. Quoted in Seligman and Hager, 1972, p. 15.

  22. Pinker, 1994, p. 18.

  23. Pinker, 1997, p. 403.

  24. Carpenter, 1975; DeCasper and Fifer, 1980.

  25. Dunbar, 1996, p. 62.

  26. Dunbar, 1996.

  27. Murphy, 1976, cited in Lykken, 1995.

  28. Sacks, 1985.

  29. Farah, 1992.

  30. Pliny the Elder (1st century A.D.), Natural History, Book VII.

  31. Tey, 1951/1977, p. 83.

  32. Bellew, ca. 1955, p. 16.

  33. Macbeth, act 1, scene 4.

  34. Gladwell, 2000, p. 39.

  35. Dunbar, 1996, pp. 4, 79.

  36. Pinker, 2003.

  37. Allport and Odbert, 1936.

  38. D. E. Brown, 1991.

  39. Muhle, Trentacoste, and Rapin (2004) give a concordance rate of 60 percent for classical autism in identical twins. If autism is defined more broadly, to include related syndromes such as Asperger’s, concordance increases to 92 percent.

  40. Dawson, Carver, Meltzoff, Panagiotides, McPartland, and Webb, 2002.

  41. Baron-Cohen, 1995; see also Frith and Frith, 2001.

  42. Barkow, 1992.

  43. Gray, 1999, p. 563.

  44. Caspi and Roberts, 2001.

  45. McCrae and Costa, 1999.

  46. Kagan, 1998b, p. 36.

  47. Personality correlations: Bouchard, Lykken, McGue, Segal, and Tellegen, 1990; depression: Plomin, DeFries, McClearn, and Rutter, 1997; schizophrenia: Gottesman, 2001.