This is the story of how I circuitously journeyed into my life’s work. I began as a music major, then thought I would be a music therapist, then decided on psychology as the major because I was better at it. A graduate school sojourn in Michigan, a fateful conversation with John Hagen, nabbing a summer research position with Eleanor Maccoby and spending the summer trying to scare babies with a cymbal-clapping monkey. Chutzpah! Telling Eleanor I would help with the next phase of their study if she made me a graduate student. From scaring babies to giving them control over a scary toy to realizing I was studying the key factor in stress regulation to cortisol to a seminar offered by Seymour Levine and the discovery of the field of developmental psychoneuroendocrinology. I was in my fourth year of graduate school before I knew where I was going. Take heart students. You will figure it out if you keep asking questions.
I wish I could say that when I was young I had a grand vision of where I was going. Honestly, for a long time I just followed my nose. I started at Mills College in Oakland, California as a music major, then taking a psychology course led me to change majors. However, it was not until I took experimental psychology that I began to catch fire. I was fascinated that there were sex differences in basic perceptual processes. It was the early 1970s, the height of the women’s movement – sex differences were learned, weren’t they? When I began thinking about graduate school, I drove across the bay to Stanford University and talked with Sandra Bem. She told me I had little chance of getting in because I hadn’t done research. No one did research at Mills! To improve my prospects, I decided to write a review of sex differences and submit the abstract to an undergraduate conference. To my amazement my paper was accepted, and when I presented the paper, I met undergraduates working with Eleanor Maccoby on her sex difference longitudinal study. Though I did not know it, meeting Eleanor’s students presaged my future.
I got into several graduate schools and chose developmental psychology at the University of Michigan because I wanted to work with Lois Hoffman on her research on women’s employment; but the year I got there she was on sabbatical. I was miserable. I was in love with a Stanford student who was still in California, and it was cold and dark in Michigan. When I told one of my professors, John Hagen, that I was thinking of not doing summer research but just going to California for three months, he looked alarmed. Then he picked up the phone, called his old advisor, Eleanor Maccoby, and before I knew it, I had snagged a job working for her. At the end of the summer, Maccoby and her colleague, Carol Jacklin, were planning the next assessment in their longitudinal work and were talking as if I was staying. I said, “I have to go back to Michigan, but I’ll stay if you make me a graduate student.” (What chutzpah.) Eleanor pulled strings and got me in that fall of 1974.
Now you would think that my path would finally straighten out, but there were a few more twists. Maccoby had me testing twelve-month-olds for sex differences in fearfulness by scaring them with a cymbal-clapping toy monkey. It amazed me that they were fearful of this toy as given a spoon and a saucepan babies made more noise themselves. I began wondering if they were frightened of the noisy toy because they were not in control. I read the literature on what causes fear in infants, and no one was studying the infants’ control of potentially frightening stimuli. With Eleanor’s encouragement, I designed and ran a study in which I randomized some twelve-month-olds to turn the toy on themselves and others to a control condition in which the toy came on according to a scheduled generated by another infant. I found that allowing the babies to turn the toy on themselves transformed the cymbal-clapping monkey from scary to funny.
In my reading about fear and control in research on animals and adults, I soon ran into a large literature on learned helplessness, control, predictability and stress hormones. In this literature I read again about the work of Seymour Levine, whose research I first encountered in learning about the role of testosterone in masculinizing the infant rat. But this time I was reading his work on early handling in rat pups, the hypothalamic-pituitary-adrenocortical (HPA) axis,Footnote 1 and fearfulness. Then the defining moment of my career came when, walking down a hall in the psychology department, I came across a sign saying that Levine from the medical schools was to offer a seminar on developmental psychoneuroendocrinology. I literally ran to Eleanor with this news. She wasn’t thrilled about me taking the course, but I persisted. I was the only psychology student in the seminar. The rest were his students, and the seminar became “Teach Megan about everything we do.”
Soon I was testing infants for my dissertation on control, predictability and fear, and Levine’s student, Joanne Weinberg, was teaching me how to take gonads out of adult rats and give them injections of testosterone in a study of flavor toxicosis. I was already doing my postdoctoral work in Levine’s lab as the ink dried on my PhD diploma in 1978. It was one of those magical times in a lab where a group of stellar students comes together and amazing science happens.
With this group, I did research on both rats and monkeys. However, I did not want to continue as an animal researcher, and thus I was excited to learn that the assays for cortisol were being adapted for use in saliva. After one postdoctoral year, I went on the job market as this interesting case of a human developmental psychologist who was beginning to do psychoendocrine research. This put me on the leading edge of a research wave (biopsychosocial research) that wasn’t quite discernable yet, but which interested enough people to get me five offers. I chose the Institute of Child Development (ICD) at the University of Minnesota because I would not only have colleagues who were stellar researchers, but also ones who were remarkably supportive. Indeed, from the moment that Andy Collins, the chair of the search committee, picked me up at the airport I felt I was home.
During my early years at ICD my research was guided by a practical issue: I did not have any money. It was 1979 and start-up packages were not really offered then. I needed cheap-to-run studies to keep my publications coming. Thus, I continued my behavioral research on fear and control, while trying to figure out how to get started with the more expensive psychoendocrine research. Knowing what I needed, Andy Collins, who lived in a University of Minnesota faculty housing area near campus, introduced me to his neighbor, Michael Steffes, who was the head of lab medicine at the U’s medical school. He assayed my first plasma cortisol study samples without charge. Andy also introduced me to another neighbor, a pediatrician, Robert Fisch, who got me into the newborn nursery and generously served as the phlebotomist in my first cortisol study. This first study was on the effects of circumcision on stress responses in newborns. Fisch was a Hungarian and Holocaust survivor. I can still hear him rushing down the hall to get to our study on time, “Ver ist ze baby, I need to take hiz blood!” Together we proved that, contrary to what was widely believed, the human newborn, unlike the rat pup, had a very reactive HPA axis. This evidence was sufficient to help me secure a five-year grant from the National Institute of Child Health and Human Development (NICHD).
From Studies in Newborns to Studies in Toddlers
Why the newborn nursery? Well, those assays in saliva that were on the horizon when I was in Levine’s lab were not yet on the market. So, I needed blood. When is blood being taken on otherwise healthy children? During the metabolic screening exam conducted when they are newborns. To obtain the blood for my studies, I hired a nurse as my research assistant. She could take blood for the newborn metabolic exam, and while doing so, she could then take a few extra drops to assay for cortisol. We simply timed blood sampling for the screening exam to coincide with our experiments. Through these experiments, we learned that almost anything distressing to the newborn would elevate cortisol. We also showed that while crying and cortisol were correlated, they were regulated separately. With Howard Stang and Lenny Snellman, who were pediatricians from an HMO (Group Health),Footnote 2 I also conducted the first study using a nerve block to reduce pain for infants undergoing circumcision, a study that led to a noteworthy change in clinical practice.
By the mid 1980s salivary assay methods were available, and I was released from the newborn nursery. Expanding on my behavioral studies of fear and coping, my initial focus was on whether HPA axis reactivity was related to fearful temperament. I used an observational temperament assessment with nine-month-olds and followed them at twelve-months of age with a thirty-minute separation and stranger interaction sequence. The results were a disaster. At nine months the babies showed some small elevations in cortisol, but by twelve months I was getting almost nothing. I called Levine and complained. “The adrenal isn’t doing what it is supposed to.” He responded, “Yes, it is, you just don’t understand it.” Not helpful, Levine! But true. Indeed, my job was to figure out what the adrenal was telling me about the development of stress regulation.
To trace the development of the HPA response to stressors, I left the lab and turned again to pediatricians. This time with the help of Joe Rigatuso, another pediatrician from Group Health, I followed babies over their first two years, observing and assessing them during their well-child checkups and childhood inoculations. Michael Lewis was doing similar work at his Rutgers University lab, and we were both seeing a marked decrease in responding by the second year of life.
Secure Attachment and Social Stress Buffering
Was this reduction in responsiveness of the adrenal like the hyporesponsive period in the rat pup? I didn’t think so because some babies were still responding at eighteen months as much as they did at six months. Perhaps it is relevant that at this time my office was directly above that of Alan Sroufe, a leader in attachment theory and research, and I would joke that attachment theory wafted into my office through the air vents. Whatever the case, I decided to test whether attachment security had anything to do with this lack of HPA response that we saw emerge over infancy. It did – big time. Securely attached infants did not elevate cortisol when frightened, while those who were insecurely attached, did. The HPA axis, then, was being buffered by the presence of the parent in secure relationships. The paper showing this was published in 1996.
In 2013, with my students, Cam Hostinar and Jena Doom, I turned to the question of whether parents are effective social buffers of stress beyond infancy. We found they are, but only until about the midpoint of puberty. With puberty the presence of the parent no longer prevents or reduces cortisol increases, at least not when the stressor is psychological. Notably, the midpoint in puberty is also when there is an increase in major depression, a disorder that is related to problems in HPA axis regulation. The association between HPA dysregulation and depression in adults is one of the most reliable findings in biological psychiatry, and both Emma Adam, a former student, and Ian Gotlib have evidence linking HPA axis dysregulation as a precursor of depression emerging in adolescence.
Temperament
Parental buffering of the HPA axis was a major finding, but it did not help me understand temperament’s relation to stress reactivity. Luckily, about this time my firstborn entered the lab preschool in ICD and I began hanging out in the observation booths. Watching the children, I began to wonder whether, without parents present as stress buffers, we would see associations between temperament and stress reactivity. I used some non-sponsored funds to collect cortisol by having kids play a “tasting game” – mouthing a few grains of drink mix on a cotton dental roll, each day before a snack. Whoa! I found that some children were periodically producing really high cortisol levels, and these were the more temperamentally reactive children. Seeing this, I contacted my program officer at the NICHD and asked if I could divert my grant funds from infants to preschoolers. He agreed and I began studies in the preschool that revealed more confusing findings.
Specifically, when the school year began, it was the outgoing children, not the shy ones, who showed the highest cortisol levels. We found the same results for elementary school children. When we followed preschoolers over the school year, though, we found that by spring the shyer children showed increasing cortisol levels. However, this increased pattern was not the case for the shy children who continued to play more by themselves but, rather, for the ones who were trying to be social. What was going on? The best explanation, I thought, was that it wasn’t the child’s temperament, but whether the child was embracing the challenge of making friends and the hard work of learning how to play with others. In short, embracing challenge and uncertainty rather than shyness seemed to be the active ingredient in relating temperament to cortisol reactivity.
Childcare and Cortisol Increases
By 1997 my graduate students started complaining that we should not focus only on the lab school with its half-day programs because most children were in full-day childcare. They were right, of course. So, for our next study we sampled cortisol at childcare centers midmorning and midafternoon and examined whether temperament was related to cortisol levels or changes over the childcare day. Cortisol typically decreases from midmorning to midafternoon, but to our surprise we found that for every single child cortisol was rising over the day. And the increases were larger at centers that were lower in quality. In our lab preschool work, we had not found that cortisol levels in the afternoon classes were higher than in the morning classes. Therefore, simply put, studying cortisol in childcare we had found that full-day out-of-home care is stressful. This was not a finding that was going to endear me to my feminist friends, as Eleanor pointed out to me.
Do children always increase in cortisol when they are away from home all day? To answer that question, we studied infants and toddlers, preschoolers, and young grade-school children (the latter during the summer) when they were in full-day, out-of-home, childcare. Studied cross-sectionally, we saw no cortisol increase for infants, the largest increase for toddlers, then preschoolers, and among five and six years of age and older children, cortisol was not increasing over the childcare day. Why the greatest increases among the toddlers? We suspected it was because playing with others, which begins in earnest among toddlers, is especially difficult and uncertain when both partners are just beginning to develop social skills. The idea that it was the challenge of interaction that was activating the adrenal was further suggested by evidence that cortisol decreased over nap time (when children were not interacting), even if the child did not sleep.
Recently a group in MontrealFootnote 3 has shown that if you implement a training program that helps children learn how to play well, and randomly assign some childcare centers to training or not, you eliminate cortisol increases in those receiving the training. And it is well established that in higher quality childcare settings, staff focus significantly on supporting children as they learn to play with one another. Thus, it is likely that the quality of childcare findings and the results of the social training study fit together.
Does it matter that cortisol increases at childcare? We suspected that it would depend on the state of the brain when cortisol is elevated. Cortisol can increase for many reasons, beyond feeling threatened. Aerobic exercise, for example, elevates cortisol. At childcare, cortisol might rise because of the effort of playing with others all day, even among children who are not shy and fearful. But the elevations in cortisol might be more likely for shy, fearful children than for outgoing children to correspond to periods when the amygdala is actively processing situations that trigger shyness and anxiety in the child. Under such circumstances, cortisol would be expected to enhance activity in those circuits, and thus might increase the child’s fearfulness over time. Indeed, with my friend and childcare expert, Deborah Phillips, we did find that over time shy children developed more internalizing problems if they were in childcare settings that produced rising cortisol levels daily, while outgoing children showed little effect on either internalizing or externalizing behaviors.
Early Life Adversity
It took ten plus years of working out a basic understanding of the HPA system before I felt we knew enough to tackle the big question of how early life adversity might impact that HPA axis and influence mental and physical health. Luckily for me, it was then that a wonderful student, Jordan Hart, joined the lab. She wanted to study maltreated children. Not knowing the first thing about maltreatment, I picked up the phone and called Dante Cicchetti who was still at the University of Rochester, though he later moved to ICD. I asked him if I could send Jordon to him during the summer and if we could collect cortisol from the preschool children at his Mt. Hope Family Center. Dante was wonderful, and for several summers, with Jordan doing the work, we studied preschool children and then school-aged children in his summer camp. During this work we got hints that cortisol was not elevated but suppressed for children experiencing chronic stress.
I came out of that work realizing two things: (1) in the rodent literature there is a sensitive period early in development when maternal care shapes the set point for the HPA axis. Because that period seems to be prior to weaning, it was very unlikely that it extends across all of childhood. Thus, if I was going to study something analogous, I needed to find children who experienced harsh care isolated prior to the equivalent of weaning; (2) children adopted from orphanages would fit this bill. And remarkably, two career changing events occurred as I was coming to this realization.
First, Dana Johnson, a pediatrician at the University of Minnesota gave an ICD bag lunch colloquium pleading for help. In his International Adoption Clinic, he was beginning to see an influx of children from Romania and the former Soviet republics. He could address their medical issues, but not their confusing behavioral problems. Would ICD faculty please help? Second, Maya Carlson and Tony Earls asked me to accompany them to Romania to collect salivary cortisol in an orphanage. The children were around three-years-of-age, and we compared their cortisol patterns over the day to those produced by similar-aged, family-reared Romanian children. The family-reared children had normal daily rhythms, while not one of the orphanage children did.
I contacted a program officer at the National Institute of Mental Health (NIMH) and she said I needed to determine my sampling frame and obtain evidence that the children I had access to were experiencing problems relevant to NIMH before I had a chance of getting funded. Not knowing anything about adoption, I contacted Harold Grotevant, an expert on domestic adoption then at the University of Minnesota, for advice on how to find families who had adopted internationally. He gave me the name of the head of Adoption and Guardianship in Minnesota’s Department of Social Services, Robert DiNardo. Perhaps because of Hal’s good name, DiNardo was very excited by my question. The state had all the records, but they were on microfiche. He could not let me into those records, but if I supplied the funds, he could hire people to do the abstraction. He could also send out the survey we had in mind if we paid for the mailing. The participant ID-identified survey then would be returned to the university, not the state, to increase trust that the survey answers were anonymous.
Given that cooperation, we proceeded, and NIMH was willing to fund that initial survey. While the names and addresses were being abstracted, I went on a listening tour around Minnesota with Meg Bale, a retired post-adoption social worker. We attended meetings of parent groups and worked with the major adoption agencies to develop a survey that, while containing measures of mental health (via the Child Behavior Checklist) that I needed, the survey also included questions that parents wanted answers to, for example, information on problems the families were facing with the school system. The results of the survey indeed showed that many internationally adopted children were having problems of interest to NIMH. Furthermore, in response to the survey, families of over 4000 children entered a registry of families interested in being contacted for research.
I have always felt that I could not work with children and families who come needing help and study only my own questions. Those questions revolved around early adversity and regulation of stress, while the parents’ questions revolved around why their children had such problems regulating attention, were overly friendly with strangers, and had trouble taking the perspective of others. To address both my interests and their concerns, armed with the survey data and registry, I wrote and received a five-year grant that covered both.
I conducted this work with Seth Pollak who was at the University of Wisconsin, Madison. Seth was an expert in cognitive neuroscience, an expertise I lack, and he had also been approached by parents to help them understand their orphanage-adopted children. Rather than write competing grants, we worked collaboratively. Together we showed that a dearth of responsive, social care in the first year of life is associated with impairments in competencies (attention regulation, executive functions) that depend on the prefrontal cortex, a brain region that presumably develops later in development. Our results also suggested that impacts on the functions subserved by prefrontal brain circuits might reflect ongoing dysregulation of the HPA axis. Specifically, we were finding evidence of a sensitive period when the quality of care shapes the reactivity and regulation of the HPA system. That period seems to close between eighteen and twenty-four months of age. Children adopted after this age show a marked cortisol hyporesponsivity that is predictive of disorders of social engagement, problems with attention regulation, and through these pathways, problems with peers.
Much more recently (2019), my students and I have shown that with puberty the system recalibrates, resulting in cortisol responses that do not differ from those of children who never experienced early institutional care (Gunnar et al., Reference Gunnar, Reid, Gelman and Waxman2019, PNASFootnote 4). Sadly, this recalibration appears to confront youth with increases in anxiety and depression and problems with emotional regulation. It is as if they developed to handle a “Volkswagen” stress system and are now trying to learn to drive a car with a “Ferrari” engine.
Networks and Networking
One of the joys of my career has been participating in collaborative research groups and think tanks. In 1990, I was asked to join a William T. Grant consortium pondering the problems of temperament and stress where I met Tom Boyce, Ron Barr, and Steve Suomi. In 1999, I obtained NIMH funds to form a research network of rat, monkey, and human researchers studying early life stress and the HPA axis. This network went through three iterations over twelve years, finally becoming a full-blown Center on Stress and Neurobehavioral Development that I co-directed with Phil Fisher. While I was starting this network, I was asked to join the Canadian Institute for Advanced Research program on Human Development and Population Health. This group was putting developmental flesh on the bones of Michael Marmot’s studies of the social determinants of health and had produced the book The Developmental Health and the Wealth of Nations (Hertzman & Keating, 1999) that introduced the concept of the biological embedding of early experiences. Along with others (e.g., Michael Meaney), I was brought in to add knowledge of human biology that might be involved in how “experiences get under the skin to influence brain and behavioral development” (i.e., biological embedding).
That program, which has gone through many iterations, is now titled Child and Brain Development. Among the influential researchers I have had the opportunity to learn from there are Michael Rutter, Bryan Kolb, Michael Kobor, and Thomas McDade, along with Michael Meaney. The expertise in the program ranges from fruit fly genetics, to epigenetics, to neurobiology, to artificial intelligence/machine learning, to pediatrics, epidemiology, and biological anthropology. Every meeting stretches me.
Circling back to 1998, I was asked to join the National Academy of Medicine’s panel on Integrating the Science of Early Development that was led by Jack Shonkoff and co-directed by Deborah Philips. We produced the highly influential 2000 report, Neurons to Neighborhoods, and Shonkoff went on to secure funds to keep us going as the National Scientific Council on the Developing Child. The Council’s function is to continue to review emerging findings in developmental science and work with experts in the science of how people understand and use information in order to translate our science more effectively so that it influences policy and practice. As Jack frames it: “closing the gap between what we know and what we do for children and families.”
Concluding Remarks
My scientific journey, as this essay shows, has not been only a journey of ideas. It is also a journey of relationships that have supported and shaped the work. Serendipity. Recognizing and seizing opportunities. Curiosity. And in my case, following my nose. All along the way I have had the wonderful opportunity to work with and learn from amazing students. I hope that this chapter helps the next generation know that if they aren’t sure of where they are going, they should nevertheless take risks, and have faith in their own ability to journey successfully into the scientific unknown.
