When the director of the National Institute of Mental Health (NIMH) requested a review of the behavioral research supported by the institute, he wanted to know how that body of knowledge could be applied to clinical populations. The director, Steven E. Hyman, M.D., is himself a molecular neurobiologist.

Neither that experience nor his training as a psychiatrist had given him a deep understanding of basic behavioral research. In his four years as NIMH director, however, he has come to appreciate the potential contribution of the social sciences to alleviating mental illness. "We have to value all basic disciplines as important potential sources of ideas and approaches," he said in a recent interview.

Despite NIMH’s large investment in basic behavioral research throughout its history, Hyman said, the results of that research have not contributed sufficiently to understanding mental illness. Basic cognitive science and social psychology could be used to comprehend underlying processes, for example. He pointed out that in schizophrenia, cognitive deficits ultimately may prove to be more disabling than delusions or hallucinations, which tend to respond better to antipsychotic medication. Knowledge about cognition derived from basic behavioral science has simply not been brought to bear on understanding schizophrenia as much as it should have been, he said.

Similarly, better objective diagnostic techniques are needed for attention deficit/hyperactivity disorder (ADHD). "Basic behavioral scientists have been studying attentional processes and impulse control for decades," Hyman noted. "It seems critically important that we use this information to develop better developmentally sensitive, more objective diagnostic tests that can be used to follow outcomes of treatment."

Hyman believes that the basic behavioral sciences have yielded less useful information about mental illness than have the biological sciences. He attributes this disparity to differences in the choice of research questions and to methodological limitations.

The crux of the problem may be the difficulty of working with psychiatric patients. "The sorry truth is that the most rigorous work is being done in basic departments where many scientists are unwilling to take the risk of undermining this rigor by using difficult populations—meaning clinical populations," Hyman said.

But he pointed out that the culture in which a science is embedded also profoundly influences its direction. "Many of the outstanding basic behavioral scientists are in colleges of arts and sciences, where they have been performing very rigorous and highly acclaimed, well-funded, and well-published basic investigations. They’re getting professional acclaim. From their point of view, there’s nothing wrong."

Hyman said that NIMH has to entice these scientists to study problems related to mental illness. "We have to say to them, ‘Don’t you think you’ll grow more? Don’t you think you’ll have a more important impact if you begin to explore clinical populations and expand to collaborate with people using other tools?’ "

Ultimately, such research will not only serve the institute’s public health mission, Hyman believes, but also "will enliven and enrich basic behavioral science and the lives of the individuals doing the research." Hyman said that he wants behavioral scientists to have richer two-way connections with neuroscience on the one hand, and public health-related disciplines on the other.

Better interaction with public health professionals may eventually improve epidemiologic studies, for example. "If you look at how we ascertain who has mental illness in our communities, we’re using some very old constructs," he said. "We’re certainly not capitalizing on what bases behavioral scientists have learned about cognition, emotion, and social processes."

One of the old constructs that needs revision is the present diagnostic system, according to Hyman. Among its deficiencies are the present syndromes, which don’t always mesh with new evidence. "Genetically, for example, mood and anxiety disorders don’t segregate so neatly within families," he noted.

Ultimately, new behavioral research should be reflected in the DSM, Hyman said. Evidence suggests that genetics influence the way in which affect, especially negative affect, is processed. "It may be that the separation we have between certain mood and anxiety disorders is just not right. Whether that leads us to a dimensional system or to revised syndromes, we need to change the way we’re thinking."

Another important change is needed in the diagnosis of ADHD, Hyman said. "There’s no developmentally relevant guidance to clinicians. Development has been ignored for a long time. That to my mind is nothing short of disastrous."

He acknowledged that DSM-III was an "enormous leap" from idiosyncratic diagnostic practices in use before its introduction. But the system has become rigidified and has caused the profession to lose sight of the questionable validity of its categories, he said.

"I think cognitive science and other behavioral sciences have a lot to offer," he said. "If we’re going to map the brain’s circuitry, we have to know how to slice this up. Where are the boundaries?"

Hyman pointed out that diagnoses continue to be made at the behavioral level. "We would do very well to have the most sophisticated input on how we understand and evaluate cognition and emotional processing, emotional reactivity, and social cognition. I think this can only make us stronger."

He was particularly critical of the lack of exposure to behavioral sciences in psychiatry residency training. Many developing psychiatrists don’t even know that relevant psychology journals exist, he said. The best work of the behavioral sciences can only enrich psychiatry, in his opinion.

The goal of the new emphasis on clinically relevant behavioral science, Hyman said, is a model of mental illness that is less reductionistic than those that predominate at present. "Our romance with the dexamethasone suppression test or the serotonin metabolite level is over. We found out that these tests provide useful information, but that information doesn’t go very far."