How a fraudulent experiment set psychiatry back decades

In January 1973, Science (along with Nature, the most influential general science journal in the world) published an article that immediately captured major media attention. David Rosenhan, a Stanford social psychologist, reported that eight pseudo-patients had presented themselves at a variety of mental hospitals, 12 in all, complaining that they were hearing voices saying ‘hollow, empty and thud’, but otherwise behaving completely normally.

All of them, he reported, were promptly admitted, and all but one diagnosed as suffering from schizophrenia (the other receiving the somewhat more hopeful diagnosis of manic depressive psychosis). It took weeks for them to be released, though they were instructed to show no symptoms once admitted. When they were finally discharged, they received the damning diagnosis of ‘schizophrenia in remission’, a label that promised to compromise their futures ever afterwards.

At the time the paper appeared, psychiatrists were already under siege. One of their number, Thomas Szasz, had recently pronounced that mental illness was a myth, and that his fellow professionals were little better than gaolers. And the sociologist Erving Goffman had published his best-selling book Asylums, which suggested that mental hospitals created rather than cured mental illness, and could be usefully compared to concentration camps. Now shrinks found themselves ridiculed as ‘experts’ who were incapable of the most basic of tasks, deciding who was mad and who was sane. Rosenhan’s paper caused a sensation, and his exposé resonated powerfully outside the cloistered world of academia.

The leadership of the American Psychiatric Association was thrown into a panic, and within weeks desperately sought ways to render psychiatric diagnosis more reliable. But then, as now, no scans, no blood tests, no laboratory findings allowed them to distinguish the mad from the sane. Perforce, like 18th-century doctors, psychiatrists depended upon symptoms and patient self-reports, and the new system they created in the years that followed thus relied on a tick-the-boxes approach to diagnosis: perm any six symptoms from ten and voila! one was given a diagnosis of schizophrenia; exhibit another set of symptoms, and one was manic depressive, and so forth and so on, for the new Diagnostic and Statistical Manual ran to some 494 pages and 256 different ‘illnesses’.