* We became concerned with behavioral subtypes of attention deficit disorder (ADD) several years before the recognition of two subtypes in the third volume of the Diagnostic and Statistical Manual of Mental Disorders (DSM) (American Psychiatric Association [APA], 1980). In this article, we provide an historical perspective, through our own work, for the current debate regarding ADD subtypes. We also present a brief overview of studies from other research groups, focusing first on contrasts of children with ADD, with and without hyperactivity, and then on children with ADD, with defiant or aggressive symptoms. Finally, we present arguments for and against subtyping and suggest ways to achieve validation of behavioral subtypes.
A REVIEW OF OUR STUDIES
From the outset of our research with children with attention problems, we recognized the heterogeneity of this group. In our first large-scale study, we classified boys with ADD into three behavioral groups: hyperactive, normally active, and hypoactive (Dykman & Ackerman, 1976; Dykman, Ackerman, Clements, & Peters, 1971; Dykman, Walls, Suzuki, Ackerman, & Peters, 1970) . The rationale for this subgrouping came from the work of Luria (1961), who studied two groups of children who were seemingly without disabilities, yet performed poorly in the classroom. Drawing on Pavlovian experimentation in central nervous system subtyping, Luria described an excitatory type and an inhibitory type. The excitatory type's characteristics corresponded with those of the hyperactive child, then beginning to claim the interest of American child psychologists and psychiatrists (Wender, 1971; Werry & Quay, 1972). Luria's inhibitory type, characterized by slow execution of purposive movements, had not then engaged research attention; but clinicians knew well such children who, though not disruptive, dawdled and day-dreamed through much of the school day (Clements & Peters, 1962). Their teachers complained that these children didn't carry out instructions well, failed to complete assignments, or inadequately recalled material they read or heard presented. Similar complaints were made of hyperactive children, but the underlying reasons appeared to be different.
When we began our initial study, no appropriate standardized rating scales existed for placing students into behavioral categories. Thus we used the classroom teacher's global rating of activity level (5-point scale, ranging from highly disruptive to well behaved) to classify the boys. Those not rated as disruptive were considered normally active unless the teacher and parents described the child as extremely slow in purposive movements, in which case he was classified as hypoactive. As hypothesized, the hypoactive group was significantly slower on laboratory reaction-time measures than the hyperactive group (Dykman et al., 1970).
Moreover, when we studied these elementary school boys again at age 14 (Ackerman, Dykman, & Peters, 1977), the hyperactive group had accrued numerous reports of episodes of conflict with authority figures (home, school, community), whereas the behavioral reports of the nonhyperactive groups were similar to those for the normal control group. On a self-rating scale (The Minnesota Counseling Inventory), the hyperactive and hypoactive groups had significantly different profiles. The hyperactives had elevated scores on a cluster of traits getting at conflict and control (family relationships, emotional stability, and conformity), whereas the hypoactive teenagers had adverse scores on social relationships, mood, and leadership. The normally active boys had profiles more similar to controls. In the laboratory, the hyperactives showed an inability to conform on the reaction-time key only in response to light signals. Most fiddled with the key during intertrial intervals, which significantly discriminated them from the nonhyperactive subjects. In evaluating family data on this sample, we found one major difference, which was corroborated in a subsequent sample (Ackerman, Elardo, & Dykman, 1979). …