In a Different Key

by John Donvan

  Over the next decades, hundreds of children were educated there. Meanwhile, many of them made cameo appearances in some of the most forward-thinking research ever undertaken in the quest to understand what autism really is.

  —

  ONE DAY IN 1967, a woman and a man, both of them scientists, showed up at Florence Road hauling along a heavy wooden box, roughly the size of a window air conditioner. A hole cut in one side was just large enough for a child’s head to poke through. For the next several hours, a long line of boys and girls with autism sat down in a chair in front of the box and did just that—they stuck their heads inside, as the pair watched and took notes.

  Data was everything to the two scientists who, in the compact universe of London experimental psychologists, were already research legends. The Australian-born Neil O’Connor and the German-born Beate Hermelin were seen as superb designers of brilliant experiments that shed light on how the minds of children with autism worked—and how they worked differently from everyone else’s. Before O’Connor and Hermelin, almost no one had bothered to ask these questions.

  The pair worked out of Maudsley Hospital, a psychiatric facility in South London. Redbrick and baronial, “the Maudsley,” as it was always called, was paired operationally with Bethlem Royal Hospital, the direct descendant of the notorious thirteenth-century insane asylum known as “Bedlam.” In the twentieth century, the Maudsley’s wards were used for teaching by the Institute of Psychiatry, the country’s top postgraduate training program. Most of Britain’s leading psychiatrists passed through on the way to becoming fully certified.

  O’Connor and Hermelin, who were psychologists, both spent time at the Maudsley. But starting around 1963, their desks were shifted to a set of wobbly wooden huts that stood in the big building’s shadow. Thrown up after the war, those rough structures became home to a research group known as the Social Psychiatry Unit. An undertaking of Britain’s Medical Research Council—the equivalent of the National Institutes of Health in the United States—the SPU brought together a disparate collection of laboratory psychiatrists, social scientists, statisticians, and graduate students. They all wanted to make a mark.

  The unit became an intellectual hothouse. It was London, after all, and it was the 1960s. Iconoclasm ruled. Everything “establishment”—art, music, fashion, humor—was being frisked, mocked, shaken up, and made to account for itself. Something parallel was happening in the social sciences in Britain, especially at the Maudsley. The researchers in the huts—colleagues, friends, and competitors all—egged one another on in pursuit of the same shared objective: to challenge every known tenet of psychiatric and psychological dogma by putting it to the test of experimentation. That was the ethos and the essence of experimental psychology, which was the specialty of Hermelin and O’Connor: Test everything. Demand data.

  In 1963, the pair turned their attention to autism. They wanted to give the kids tests—actual small tasks—so that they could measure performance. At the time, this was generally considered a futile proposition, since so many of the kids did not communicate in a recognizable way or were otherwise uncooperative. Hermelin and O’Connor, however, believed that there were still ways to elicit observable physical and mental responses that could be measured and quantified, ones that did not require much cooperation or conversational communication. They built that odd wooden box with these goals in mind.

  —

  THE INSIDE OF the box was painted completely black. Besides the large hole on one side for the children to put their heads through, there was a tiny peephole-sized one on the opposite side that let the researchers watch the children’s faces while they were in there. That day, more than two dozen kids were brought into a small room set aside for the researchers. Some stuck their heads into the box because they were curious, others because they wanted the candy that was being used as a bribe.

  For the first few seconds, the kids saw only total blackness. Then suddenly, a spotlight flashed on, revealing a human face floating in the far dark corner—actually a live person around back, pressed up against a third opening. For ten seconds, the person’s eyes closed; then they opened again. After thirty seconds, the light was cut, and the box went black again.

  O’Connor and Hermelin were interested in the kids’ eye movements. They wanted to see where the children looked when the light came on and for how long, then contrast this with what happened when the same children were run through the experiment several more times, albeit with a key difference. In these trials, instead of a real live person’s face, the light revealed pairs of upright white cards displaying various images, abstract geometric shapes, plus one that showed a photograph of a face.

  It took a few hours to run the paces with all twenty-eight children who cooperated that day. Then, on a second day, Hermelin and O’Connor took the box to another school to run the tests on a roughly equivalent number of younger children who were matched with the first group in so-called mental age. The children in this second, control, group did not have autism.

  When the experiment was over, O’Connor and Hermelin analyzed the data and found that the children with autism, as a group, differed significantly from the control group in the attention they gave to the dark and empty areas of the box. The control group paid almost no attention to the darkness, but the children with autism, judging by where their eyes went, were quite curious about the shapeless voids and shadows.

  It was a paper-thin distinction, but it was real. And it was typical of the line of inquiry Hermelin and O’Connor pursued. Over a five-year period, through similarly exacting experiments, the pair continued to discover narrow yet quantifiable ways in which children with autism processed the world differently from other children. They discovered, for example, that many children with autism relied on their sense of touch more than they did hearing or seeing. In all, their experiments produced an array of data making clear that autism had a neurological basis. Here at last was proof that autism had to do with the brain, not with a mother’s love.

  Hermelin and O’Connor exerted a profound influence on their colleagues in the huts outside of Maudsley—both by the kinds of questions they asked and the discipline they brought to seeking answers. It was an ethic of inquiry that would persist as the researchers grew in numbers and took on still more novel questions. These included what seemed one of the most elementary of all: How common was autism?

  It was not, and would never be, an easy question to answer. But London was the first place anyone would even try.

  26

  WHO COUNTS?

  Victor Lotter left South Africa for London in 1963 to find out how much autism there might be in the world. He was late getting into psychology because he was late to enter the University of Cape Town—that lateness a result of a case of ankylosing spondylitis that hit him when he was fourteen. A severely painful autoimmune assault on the bones, it knocked him out of school for years and left him with a wicked curve in his back and a marked stutter in his gait. By the time he applied to the university, he was in his late twenties and mostly self-educated. He graduated with a bachelor’s degree, his school’s top prize for anthropology, and a job offer from Neil O’Connor, to work in the huts outside the Maudsley in London.

  Lotter arrived, clean-cut and buttoned up in a jacket and tie, eager to take on something difficult for his PhD thesis. O’Connor was happy to oblige, giving him the gargantuan task of attempting to determine the prevalence of autism in Middlesex County. British health authorities were beginning to come under parental pressure to deliver support services to children with autism, so they had approached the team at the Social Psychiatry Unit to help them figure out how big the need was. Lotter was assigned to count the number of children with autism in a densely inhabited swath of England that skirted the former County of London.

  Remarkably, no one, not even a veteran researcher like Neil O’Connor, had any idea how much autism existed—not in Middlesex County or anywhere else in the world. No one had ever attacked th
e question in any systematic way. It would have to be an epidemiological study, working with a sample population small enough to permit an actual count of everyone in it, yet big enough to have statistical significance. For that purpose, Victor Lotter decided to include only children born in the years 1953, 1954, and 1955, which gave him a set of 78,000 children, who at that point were between eight and ten years old.

  Lotter would have to slog his way door to door, record office to record office, child to child, in person in order to uncover a good deal of the data he would need. That, in itself, was not a small task. But he faced another challenge of an entirely different kind.

  Lotter was supposed to count kids with autism, but the question of whom to count—the matter of deciding whether a given individual had autism—was a mess of diagnostic confusion. When he turned to the medical literature to put together a simple list of defining symptoms for his survey, he discovered a tangle of competing syndromes, each with its own name, laying claim to the same traits Leo Kanner had described years earlier as being autism. In addition to Kanner’s “infantile autism,” there was also Loretta Bender’s “childhood schizophrenia,” Beata Rank’s “atypical child,” Margaret Mahler’s “symbiotic psychosis,” and a long list of other contenders, including “schizophrenic psychosis of childhood,” “dementia praecocissima,” “dementia infantilis,” “prepubertal schizophrenia,” “pseudo-psychopathic schizophrenia,” “infantile psychosis,” and “latent schizophrenia.” These terms were all being used interchangeably to describe children showing the same sorts of behaviors. As the British child psychiatrist Michael Rutter wrote in this period, “It is by no means clear that all these authors are talking about the same condition.”

  It had been twenty years since Leo Kanner laid down his first description of the condition of autism using Donald Triplett and the other ten children as a model. But during the intervening two decades, the outlines of the condition, the definition of what autism looked like, had blurred and wavered continuously as other expert voices chimed in. As early as 1955, Kanner himself was grousing about the fact that there were too many inaccurate, sloppy autism diagnoses being handed out based on little more than “one or another isolated symptom.” He fretted that his whole concept was being watered down by inconsistent standards. As he wrote later of that time, “Almost overnight the country seemed to be populated by a multitude of autistic children.”

  Such a state of affairs could only be mystifying to people outside of psychiatry. Any lay observer might think that autism was autism—simply, objectively and always.

  But that was not the case, and never would be.

  There was no biological marker for autism (then or now): it could not be determined by a blood test or confirmed by a cheek swab. It could only be diagnosed through the observation and interpretation of a person’s behaviors, which meant it was next to impossible to avoid subjectivity in judgment. This was especially true when some of the key indicators Kanner listed were as vague as “an intelligent and pensive” expression and an “affectionate relation to objects.” Doctors interpreting behaviors by those measures were bound to disagree on whether to use the label of autism.

  Autism was, and would long remain, a diagnosis in the eye of the beholder.

  —

  IN THE END, then, since no authority clearly spelled out for him what autism looked like, it came down to Victor Lotter to define that for himself. The burden was historic. A young psychologist-in-training with only a bachelor’s degree, Lotter, with guidance from his advisors, became the final arbiter of who counted and who did not in history’s first-ever prevalence study of autism.

  Lotter designed a questionnaire—a basic tool of epidemiology—that he mailed out to every age-appropriate school in Middlesex County and every mental hospital, along with a letter to the staff in those places requesting that they use his checklist of behaviors and report back to him the names of any children who might be showing autistic traits. This would give him his first big pass at the population of 78,000 and help him narrow down his study population. His checklist included twenty-two items, such as

  Spends most of the time on his/her own

  Carries or collects curious objects such as stones or tins

  Often uses a ‘special’ or peculiar voice

  Very clumsy or awkward in bodily movements

  Tries to examine things in odd ways…by sniffing or biting them

  Lotter constructed this list by improvising. First, he relied on the two criteria Leo Kanner claimed were the essence of autism: extreme self-isolation and obsessive insistence on the preservation of sameness. But he went beyond Kanner and borrowed a second diagnostic framework that went by the oddly delightful name “Creak’s Nine Points.” Published in the British Medical Journal in May 1961, these nine criteria purported to define something called “schizophrenic syndrome in childhood,” which was another of those terms being used interchangeably for children with autistic behaviors.

  A renowned London psychiatrist named Mildred Creak had chaired a panel of thirteen British experts, who spent nine months arguing over and negotiating a symptom list that added traits like “acute, excessive and seemingly illogical anxiety,” odd movements and postures, and “apparent unawareness of [one’s] own personal identity” to the overall picture. But the Nine Points, despite their creators’ best efforts, also proved vague and confusing in practice. “Apparent unawareness,” for example, was not a behavior that lent itself to precise and objective assessment. The list’s inherent blurriness was emphasized by the critic who disparaged them as “an artificially contrived cluster of symptoms.” Even Creak conceded that “subjective judgment” in their application “obviously led to divergences in interpretation.”

  Nevertheless, using these nine points, along with his own interpretation of Leo Kanner’s thinking, Lotter cobbled together his own definition of what autism looks like, so that he could go out and count it.

  —

  THE RETURN RATE on Lotter’s questionnaire was superb. By the time he finished opening up all the envelopes, 97 percent of the entire population of eight- to ten-year-olds in Middlesex County was accounted for. In this first pass, teachers flagged some 666 children as hitting at least some of the marks on his twenty-two-item autism checklist. Relying on his advisors’ expertise and a closer review of each return, Lotter winnowed this group down to 88 children suspected of having autism, at least as he was defining it.

  Next, he headed out to see each of these 88 children himself. He added to his list another 47 whose names he found by scouring the records at fourteen government medical centers for children who were not at school, most likely because of disability, for a total of 135 children. In the autumn of 1963, Lotter and his wife, Ann, who was also a psychologist and acting as his research assistant, launched the study. Together they went to public schools, mental hospitals, and the so-called training schools where some of these children had been placed. They observed each child, ran intelligence and language tests, and talked to the staff, who presumably knew the children as well as anyone. This process took months, but by the spring of 1964, it allowed the Lotters to eliminate more than half of the 135 children as candidates for the diagnosis of autism.

  Next came the most emotionally trying part of the investigation as the Lotters began visiting the homes of the sixty-one children still remaining. The purpose of the home visits was to collect detailed medical and behavioral histories from the parents. Some of the children still lived at home, but in other cases, the child was absent, residing in an institution somewhere. In both situations, the Lotters could see the strain the families lived under and could feel their despair. Indeed, by virtue of this long journey and their tour of institutions and these households, the Lotters were collectively experiencing a larger dose of the reality of autism in the lives of families than anyone ever had before. Victor Lotter could not have called himself an autism expert when he started his project, but these months of study certainly made him one. />
  All sixty-one children still left on his list showed some autistic traits, but Lotter now had to make a judgment call of his own: which children’s traits added up to autism, for the purposes of his count?

  His solution was yet one more improvisation—a kind of decree on his part. He simply split the children into two groups. He organized their sixty-one names into a list that had the children with the highest levels of impairment at the top and those least impaired at the bottom. Then, under the thirty-fifth name, he drew a line. The children above the line, he would count as having autism; the children below the line, he would exclude. The full rationale of this selection was not clear even to Lotter himself. His top thirty-five, he wrote, were merely those “it was thought should be included.” But he was bluntly honest about the subjective nature of the choice he made. “The point where a line is drawn,” he wrote, was merely “arbitrary.”

  —

  THIRTY-FIVE KIDS out of 78,000. Or 4.5 in 10,000. It was autism’s first so-called prevalence rate. In subsequent years, its importance would grow—not just as a matter of historical interest but as the baseline against which all subsequent measurements of autism prevalence would be compared. Regardless of when or where later prevalence rates were published, they would invariably be cited in contrast to Lotter’s rate, often as if it were a solid, objective, and universal truth.

  Lotter never saw it that way. He made it clear in his 1966 paper summarizing his survey that with only a few minor alterations in the assumptions he made along the way—a few different choices about where to draw the line around symptoms and their varying intensity—his 4.5 statistic could have come out significantly higher or lower.