al. (1989): Schizophrenic men are socially incompetent and therefore more easily detected in the commission of a criminal act, including violence, and more easily detained. As a result, it is difficult to compare the rate of criminal offenses among the mentally ill with that of the general population. The last two studies in this group (Convit et al., 1988; Herrera et al., 1988) point to the fact that both illegal drugs (PCP) and legal drugs (haloperidol—an antipsychotic medication) can increase the frequency of violent acts in schizophrenic men. These findings cast further doubt on the specificity of any association between schizophrenia and the tendency to commit violent acts.

Only two studies (Table 4) from a vast literature are reported here; the relationships among alcohol use, brain damage, psychopathy, and violence are very complex and beyond the scope of this review. One study (Buydens-Branchey et al., 1989) deals with the intervening variable of deficit of serotonin metabolism as a precursor of antisocial behavior. The other study by Coid (1982) reviews the literature and concludes that although violence is an alcohol-related problem, the relationship is not a direct one. Further research into alcohol-related damage to the brain is the most fruitful area of study.

A number of studies on violence in neuropsychiatric populations have focused on epilepsy. Before discussing this research, some definitions are needed. An important distinction concerns the division between acts of violence or aggression that allegedly occur in the course of an actual clinical seizure and are properly to be considered part of the seizure or ictus itself (i.e., ictal manifestations) and violent or aggressive behavior that occurs between seizures (i.e., in the interictal period).

Research on interictal aggression is summarized in Table 5. Before discussing these studies, however, it is necessary to review the ictal data. With respect to ictal manifestations, an earlier study by Ajmone Marsan and Ralston (1957) reviewed several hundred seizures in epileptic persons, induced in the laboratory by injection of metrazol for diagnostic purposes. The results of their review indicated that aggressive behavior or angry feelings were extremely rare. Much more commonly reported were feelings of

fear. Similar observations about the rarity of ictal rage or aggression were made by Gloor (1967) and Rodin (1973) and were reviewed by Mirsky and Harman (1974). Epileptic patients may thrash about in the course of a generalized seizure, or may appear to strike out if attempts are made to restrain them; however, the impaired cognitive state associated with a generalized seizure is scarcely compatible with a directed attack upon another person.

Although it seems unlikely that ictal violence or aggression is a significant human problem, there has nevertheless been a scale devised to rate aggressive behavior during seizures and to certify whether such behavior was truly ictal (Delgado-Escueta et al., 1981). This scale (Table 6) was derived from videotaped seizures suggestive of violent behaviors; it is of interest that it was based on a population of 13 patients out of a pooled sample of more than 5,000 cases. Approximately one-quarter of 1 percent of epileptic persons thus show ictal violence or aggressiveness; this minuscule figure is not in disagreement with the earlier findings with metrazol-induced convulsions by Ajmone Marsan and Ralston (1957) and others.

Concerning interictal violence and aggression, in contrast, the picture is considerably more murky. One of the models for conceptualizing this behavior is to assume that it is state-related. That is to say, the violence represents an occult or hidden seizure equivalent for some persons with epilepsy; if one were fortunate enough to have recording electrodes deep within the appropriate region of the limbic system of the brain of the patient, one would be able to record evidence of seizure (i.e., EEG spike) activity. This is notwithstanding that there may be no other overt indications of convulsive activity. Enormous controversy has raged over whether or not there has ever been a convincing demonstration of this phenomenon, since it provides a model and rationale for surgical treatment (i.e., ablation) of the offending seizure focus (Mark and Ervin, 1970; Valenstein, 1980).

An alternative view of interictal aggression or violence in persons with epilepsy is that it is a trait associated with this type of cerebral disorder rather than an ictal occurrence. According to this view, the emotional mechanisms within the brain (usually referring to the limbic system) of such a person are damaged or modified in some way by the seizure disorder such that aggressive or violent outbursts are more likely to occur in them than in nonepileptic persons. A number of questions are raised by these purported findings. Are there actually patients who conform to this description? To what extent are they representative of the

entire population of persons with epilepsy? To what extent are those behaviors, if they occur, a characteristic of the age, sex, and socioeconomic characteristics of the patients, as opposed to their epilepsy? If the behaviors occur at rates greater than those seen in appropriately selected control subjects, are there other equally plausible interpretations of the behavior rather than what might be called temporolimbic irritability?

Another question pertaining to the design of research in the area concerns whether one starts with patients suffering from seizure disorders and examines the prevalence of violence/aggressiveness in these cases. Such a study might be done using admissions to a state hospital or a clinic (e.g., Tardiff and Sweillam, 1980). The alternative is to start with persons who have been involved with the criminal justice system (or other medical-legal authority) because of dangerous or assaultive behavior and to ascertain the prevalence of seizure disorders. An example of this approach is the study by Bach-y-Rita et al. (1971) (summarized in Table 5). The first approach (i.e., to start with seizures) suffers from the potential problem of sampling bias. To what extent is a state hospital or large clinic dealing with a truly representative sample of seizure cases? What proportion of patients, unknown to the system, seek and receive treatment from private practitioners (or other sources of care and never enter the statistical pool? The second approach (i.e., to start with the aggressive behavior) may also suffer from a sampling bias. Those cases most likely to be referred and to be the objects of clinical attention are those in whom some brain abnormality is suspected or demonstrated. It is also more likely that persons who have engaged in violent/aggressive/assaultive behavior during their lives would have suffered from head injuries leading to seizure disorders. The association between the abuse received during childhood (including possible head injuries) and later abusive behavior toward others appears to be a clinical truism (e.g., Tarter et al., 1984; Dodge et al., 1990). What is sorely lacking in studies of this type are (1) a national register of persons with seizure disorders and/or (2) an epidemiologic sample of a large and varied catchment area. The availability of such sources of information would enable us to draw unambiguous conclusions about associations and, possibly, about the direction of causality. We return to this issue later.

Some of these questions are addressed in the studies reviewed here; definitive answers are not available to most of them. At issue, particularly with respect to interictal violence/aggression, is whether or not surgical treatment (i.e., resections of brain tissue)