No Two Alike
Page 9
In her talk, Eleanor Maccoby presented evidence for sensitivity interactions. Much of her evidence had to do with IQ. She also mentioned criminal behavior and schizophrenia. Adopted children whose biological parents were criminals are more likely to become criminals themselves if reared in an environment where crime is a viable career option; adopted children whose biological parents were law-abiding are unlikely to become criminals regardless of the circumstances in which they are reared. The biological children of schizophrenic parents are more likely to develop a mental illness if reared in adoptive families in which the parents aren’t functioning well themselves.11
Statistical interactions tend to be evanescent; those found in one study often fail to turn up in the next. But Maccoby had hit upon the three areas—IQ, criminal behavior, and mental illness—where reliable interactions have been demonstrated. None of these interactions, however, are found in the absence of main effects. There are main effects of genes on all three outcomes: the biological children of intelligent, schizophrenic or criminal parents are more likely to be intelligent, develop a mental illness, or commit crimes no matter where they are raised. There are also main effects of environment on all three outcomes. Socioeconomic status affects IQ. Criminal behavior and schizophrenia both occur more commonly in crowded urban settings. For schizophrenia, this finding fits with the hypothesis that virus infections may trigger the disorder in susceptible individuals. For criminal behavior, the evidence suggests that it’s the neighborhood, not the home, that determines whether or not a child will become a lawbreaker. There is nothing newsworthy about the observation that crime rates are higher in some neighborhoods than in others.12
In fact, nothing in Maccoby’s talk came as news to me. Nothing she said at the NICHD conference on parenting provided convincing evidence that parents can influence a child’s personality by the way they bring up the child. The most convincing evidence presented at the conference had to do, not with human parents and children, but with rhesus monkeys. It was Stephen Suomi who provided what appeared to be rock-solid evidence of an effect of parenting on child outcomes. That is, on monkey outcomes.
Suomi started his talk13 by contrasting “peer-reared” monkeys with monkeys that were reared in the normal way by their mothers (all these monkeys were born and raised in laboratory cages). The peer-reared monkeys were taken from their mothers at birth and bottle-fed, but from an early age they were caged together with three or four other monkeys of the same age. Peer-reared monkeys develop strong attachments to one another; they cling to each other when frightened. Although monkeys reared in isolation turn into socially abnormal adults, monkeys reared with peers are reasonably normal. However, statistical differences between peer-reared and mother-reared monkeys do show up in certain circumstances.
Does it surprise you to hear that I would expect such differences? The headline DO PARENTS MATTER? appeared above so many of the newspaper and magazine articles about The Nurture Assumption that many readers got the impression that my answer was no. But I’ve never said that parents don’t matter; certainly they matter! That’s why evolution provided parents with the motivation to take care of their children. And it’s not just a question of keeping the baby alive: having a mother (or mother substitute) may be necessary for optimal development of the baby’s brain, the parts of the brain involved with social behavior. “In order to complete its development,” I explained in The Nurture Assumption, “the brain requires certain inputs from the environment…. You might say that the developing brain ‘expects’ certain stimuli to be present in the world outside the womb and relies on them in producing the finished product.”14
Since an infant mammal cannot survive in the wild without a mother, its developing brain “expects” a mother to be present. The fact that peer-reared baby monkeys turn out nearly normal is testimony to the adaptability of the primate brain. The fact that subtle statistical differences can be found, if you look for them, between peer-reared and mother-reared monkeys didn’t worry me—it’s what I would predict.
But Suomi went on to tell his audience about a second study with baby rhesus monkeys. He described a “cross-fostering” experiment in which infant rhesus monkeys were reared by adoptive monkey mothers. Some of these infants came from strains known to produce nervous, high-strung animals: “high-reactive,” he called them. Other infants came from calmer “low-reactive” strains. Infants from both strains were reared either by good foster mothers or by bad ones. Evidently the baby monkeys from the calmer strains turned out okay regardless of which kind of mother reared them, but—here’s the interaction—the high-reactive baby monkeys were sensitive to the way they were mothered. High-reactive monkeys reared by bad mothers turned into social failures who couldn’t handle stress, but those reared by good mothers did very well. The effects of good mothering persisted even after they were separated from their foster mothers and caged with same-age peers. They ranked high in the dominance hierarchy of their peer group, Suomi reported.
These results were clearly at odds with the theory presented in The Nurture Assumption. What I was trying to do in that book was not to argue that parents don’t matter but to answer a question about differences in parenting: Are differences in the kind of care parents provide, or in the way they provide it, responsible for the differences in how the children turn out? My conclusion, based on a good deal of evidence, was no, as long as the kind of care the parents provide is within the normal range for our species. But the normal range for our species, I pointed out, is very wide. Baby-care and child-rearing practices vary tremendously across the globe and through history. You don’t have to go back very far to see this. At the time I was born, American babies were fed on a strict schedule and mothers were warned against “spoiling” them by picking them up too often. Children were routinely spanked for minor infractions of household rules. Household rules were made for the grownups’ convenience, without regard for the children’s wishes. This was only sixty-seven years ago!
Of course, differences in child-care practices are going to have visible effects on the way the child behaves in the parents’ presence. What I didn’t expect were differences that persisted into adulthood and that could be measured even in social settings that were not associated with the parents. Researchers have found no systematic differences in adult personality between people born in the first half of the twentieth century and those born in the second half, despite the sweeping changes in child-care practices and parent-child relationships that occurred during that period. Despite the reduction in corporal punishment, adults born in the latter part of the century are not less aggressive. Despite the increase in emotional closeness between children and their parents—in particular between children and their fathers—they are not happier or less neurotic. Despite the fact that they received more praise and less criticism, they are not more self-assured.15
So I was taken aback by the results of Suomi’s cross-fostering experiment. What surprised me was not that the monkeys with good foster mothers did well while they were with their mothers, but that the effects of the good mothering persisted even after they were separated from them. Suomi was describing long-term effects of differences in parenting on baby rhesus monkeys and I didn’t expect them. I couldn’t explain the results of his cross-fostering study. All I could do was to shrug my shoulders and mutter, “Well, it’s monkeys, not humans.”
But it wasn’t so easy to sweep those monkeys under a rug. They popped up again six months after the NICHD conference, when an article by five prominent developmentalists, including Eleanor Maccoby, appeared in the most widely read journal in psychology, the American Psychologist. The article was titled “Contemporary Research on Parenting: The Case for Nature and Nurture.” It was the establishment’s response to the charges I had made in The Nurture Assumption and that David Rowe—my sole ally at the conference—had made in his book The Limits of Family Influence.16
Despite the “contemporary” in the title, most of the article was the same old same old. Alm
ost a third of the references cited dated from the 1980s or earlier; fewer than half had been published after 1993. I had already explained what was wrong with most of that evidence.17 But not all of it. Stephen Suomi’s study of cross-fostered baby monkeys figured as prominently in the American Psychologist article as it had at the NICHD conference. I could no longer ignore his research.
My notes on Suomi’s NICHD talk were sketchy, so I had to rely on the description of the cross-fostering experiment given in the American Psychologist article. Baby monkeys from a genetically high-reactive strain were reared, the article reported, either by high-reactive foster mothers or by calm foster mothers. Here’s how the results of Suomi’s experiment were summarized:
Genetically reactive young animals that are reared by calm mothers for the first six months of their lives and then placed in large social groups made up of peers and nonrelated older adults develop normally and indeed rise to the top of their dominance hierarchy…. By contrast, genetically reactive infants who are reared by reactive mothers typically are socially incompetent when placed in the larger living group at the age of six months and are particularly vulnerable to stress.18
The reference cited for this experiment was Suomi, 1997—a chapter in an edited book with a mouthful of a title: Neurodevelopment and Adult Psychopathology.19 It was time I learned more about Suomi’s research—the details of his procedure and results. I can request books like Neurodevelopment and Adult Psychopathology from my local library (without even having to pronounce their titles) by filling out a form. But it usually takes two or three weeks to obtain a book on interlibrary loan. So I got in touch with Joan.
In The Daughter of Time, the mystery novel about Richard III and the murdered Little Princes, detective Alan Grant manages to do his investigating despite being flat on his back in the hospital. He does it with the help of a research assistant named Brent Carradine, a young American who has access to historical documents stored in the British Museum. There is a Brent Carradine in my story, too. Her name is Joan Friebely and she has access to the riches of the Harvard libraries.
Brent Carradine is described in The Daughter of Time as “a tall boy, hatless, with soft fair curls crowning a high forehead.”20 I can’t give you a physical description of Joan because we have never met in person (I’ve met few of my friends in person), but she has an upbeat personality and an off-beat sense of humor. Joan and I met in what I have come to think of as the conventional way: in e-mail. She came across my Psychological Review article in 1997 and wrote to me. Her doctoral dissertation, she explained, had been based on the work of developmentalist Diana Baumrind, whose work I had deconstructed in that article (and would later criticize in greater detail in The Nurture Assumption). Joan had ended up with serious doubts about Baumrind’s work and was glad to have someone to share them with. Thus began our e-mail friendship.
Joan is the one I turn to when I need access to a university library. I can make it to my local library on my own, and when I’m not in a hurry I can use their interlibrary loan service to borrow books from university libraries. But some books aren’t loaned out, and I’m often in a hurry, and much of the reference material I need is published in journals, not books. Those are the times when a Brent Carradine comes in handy.
In e-mail, I asked Joan to send me a copy of Suomi’s chapter in Neurodevelopment and Adult Psychopathology, the source cited in the American Psychologist article for the cross-fostering experiment. I read it the minute it arrived and was puzzled to discover that it didn’t say anything at all about the cross-fostered baby monkeys—they weren’t even mentioned. The chapter was about the differences between mother-reared and peer-reared monkeys.
The first author of the American Psychologist article was Andrew Collins, a developmentalist at the University of Minnesota. On Valentine’s Day 2000, I wrote to Collins and told him that the reference he had cited for the Suomi work didn’t contain the research mentioned in his article. “Can you give me the proper reference for the Suomi study you described?” I asked.
Collins replied in a few days, saying that Eleanor Maccoby had drafted that portion of their coauthored article. After he received my e-mail he asked her about the Suomi research, he said, and she told him that her sources were a phone conversation with Suomi, some unpublished papers by him, and a 1999 chapter in an edited book.21 But the chapter, Collins warned, contained only a brief summary of the experiment—no data. “We suggest that you contact Steve directly for the data,” Collins told me. “He has been slower than many of us would like in publishing the actual data, so you can goad him a bit perhaps!” He ended by thanking me for alerting him to the problem of the incorrect reference.22
What I had been hoping for was a journal article describing Suomi’s cross-fostering experiment. A chapter in an edited book isn’t good enough because there is no requirement to meet the standards of journal publication. To get a research article into a peer-reviewed journal, an author has to give a detailed account of the procedure. The number of subjects who contributed data to the study—what researchers call “the N”—has to be specified, along with pertinent information about the subjects, such as their sex. The results have to be given precisely; this almost always means that they have to be expressed in numbers, not just words. Tests of statistical significance also have to be reported.
This was the kind of information I needed for the cross-fostering experiment, so I followed Collins’ advice and wrote directly to Stephen Suomi. “Can you tell me if the cross-fostering experiment has been published,” I asked him, “and if so, where? If unpublished, can you give me more details about the procedure and results? For example, how many high-reactive and low-reactive baby monkeys were there, and how many mothers of each type?”
Suomi’s reply was lengthy and packed with details about the cross-fostering experiment. He explained that cross-fostering is a difficult procedure because two monkey mothers have to give birth at around the same time in order to make it possible to switch their infants. It took four years to complete the experiment. “In all,” he told me, “36 infants (18 high-reactive, 18 low-reactive) were cross-fostered, 2 of which died during their first year of life. Half of the high-reactive and half of the low-reactive were foster-reared by ‘nurturant’ females, etc.”23
I read it twice before it dawned on me that Suomi’s e-mail contained a lot about the procedure but nothing about the results. Collins had advised me to “contact Steve directly for the data,” but there were no data in Suomi’s e-mail. However, he did mention three papers that he said contained information about the cross-fostering experiment. He also said that he was currently preparing a monograph “that will describe the results of the entire study in comprehensive fashion” and that he expected to complete by the end of 2000. “I’d be happy to send you a copy of it when it’s done,” he said, “if you so desire.”24 “Yes, please,” I replied.
There was plenty of other stuff to read in the meantime. Joan was finding and sending me a steady stream of papers by Suomi and his colleagues. By the time she was done she had located more than three dozen book chapters and journal articles, including the three he had recommended to me, on the effects of rearing conditions on rhesus monkeys. Writer’s block does not appear to be an occupational hazard for monkey researchers.
The cross-fostered monkeys, Joan and I discovered, made their first appearance in print in a 1987 chapter in an edited book. This chapter contained preliminary results from the “first cohort of rhesus monkey infants in a major cross-fostering longitudinal study.” How many monkeys were in this first cohort? No N was given in the chapter (see why I was hoping for a journal article?), but clues dropped here and there strongly suggested that there were fewer than eight. The foster mothers in the experiment varied in two ways: high-versus low-reactive, and nurturant versus punitive. In this first cohort, Suomi admitted, “there were not enough infants and foster mothers to fill all cells of the 2 × 2 × 2 (Infant Reactivity × Foster Mother Reactivity × Foster Mot
her Maternal Style) design matrix.”25
It would take eight baby monkeys to fill a 2 × 2 × 2 design matrix, and only half of them—four—would be of the high-reactive type that Suomi claimed were affected by the type of mothering they got. These four baby monkeys would be reared by four different kinds of foster mothers. Suomi’s description of the results of this first cohort, I realized, was based on individual monkeys, such as “the one high-reactive subject who also had a punitive and high-reactive foster mother” and who ended up “at the very bottom of the dominancy hierarchy of its group.”26
Any number of things can affect the status of an individual monkey. The monkey’s size, for instance: bigger, heavier animals dominate smaller ones. Punitive monkey mothers push their babies away and don’t let them nurse as much as they want to. Were the monkeys with nurturant mothers heavier than those with punitive mothers, and is that why they were dominant? With such a small study, it would be impossible to control for body weight.
Okay, Suomi had told me in e-mail that it took four years to cross-foster thirty-six baby monkeys. So I turned to a paper he published four years later, in 1991—another chapter in an edited book—and found another summary of the cross-fostering experiment, now described as “recently completed.”27 But there were no new data and again no N was given. There were four panels of numerical graphs, showing behavioral measures of the cross-fostered monkeys in the first months of life, before they were permanently separated from their foster mothers. But these were the same graphs that had appeared in the 1987 chapter: every line was identical.28 Suomi had had four years to update his graphs; why hadn’t he done so? In 1987, he had sounded apologetic about not having enough subjects to complete a 2 × 2 × 2 experimental design. If his N was larger in 1991, why didn’t he say so?
Joan and I spent weeks searching for traces of the cross-fostered baby monkeys in published papers by Suomi and his colleagues. There were some journal articles that did contain an N, but they were about other things—not about the social behavior or dominance status of high-reactive monkeys reared by calm or nurturant foster mothers. When an N was given in a paper by Suomi and his colleagues, it wasn’t for the cross-fostering experiment described by Andrew Collins and Eleanor Maccoby in their American Psychologist article. When the cross-fostering experiment was mentioned, no N was given, just a reference to the 1987 or 1991 chapter.