During the past decade, theories relating sugar intake to violent behavior have received increasing attention. Once relegated to articles and books directed at food faddists, such theories are now discussed at meetings of criminologists, and are found in books and articles aimed at personnel in the correction and criminal justice systems. Moreover, on the basis of these theories, correctional facilities in several states have revised their dietary policies in an effort to reduce sugar intake and control violent behavior (Gray, 1986).

Interest in the relationship between sugar and violent behavior was sparked by studies by Virkkunen and colleagues suggesting that hypoglycemia was common in criminals and delinquents displaying habitually violent behavior (Virkkunen and Huttunen, 1982; Virkkunen, 1982, 1983a, 1986a,b; Roy et al., 1986; Linnoila et al., 1990). These studies compared glucose and insulin levels during an OGTT between violent male offenders and male controls matched for age and relative body weight. In comparison to controls, men diagnosed as having either antisocial personality disorder or intermittent explosive disorder (American Psychiatric Association, 1987) initially displayed greater increases in blood glucose concentrations during the OGTT, followed by rapid declines in glucose values to levels indicative of reactive hypoglycemia. Comparisons between the two groups of offenders revealed that individuals with intermittent explosive disorder displayed a more rapid decline of glucose levels following the initial hyperglycemia, as well as a more rapid return from hypoglycemic levels to the original baseline values, than individuals with antisocial personality disorder. Men with antisocial personality disorder also demonstrated enhanced insulin secretion compared to controls. This increase in insulin secretion could act to augment glucose uptake by the cells and thus contribute to hypoglycemia. In contrast to men with antisocial personality disorder, men with intermittent explosive disorder did not have significantly elevated insulin values compared to controls. Subsequent correlational analyses suggested that a positive relationship existed between the duration of hypoglycemia during the OGTT and behavioral or sleeping problems, truancy, stealing, number of crimes against property, and multiple prison sentences (Virkkunen, 1982; Virkkunen and Huttunen, 1982; Virkkunen, 1986a,b; Roy et al., 1986).

One explanation hypothesized for the relationship between hypoglycemia and aggression is that a functional deficit in serotonergic

neurons in the central nervous system may lead to abnormalities in glucose metabolism that can be conducive to violent behavior (Roy et al., 1988; Linnoila et al., 1990). Unfortunately, the details of this hypothesis remain to be elucidated.

Although individuals with a history of violent behavior had a greater tendency toward hypoglycemia than controls, this finding cannot be viewed as unequivocal evidence of an association between hypoglycemia and aggression. First, no determination of nutritional status was made in any of the studies examining this association. However, in several papers, the authors noted that habitually violent men generally had poor appetites and may not have consumed food for many hours prior to an act of violence. It is possible that the nutritional status of these men was not adequate. This seems particularly likely because all of the violent offenders in the studies by Virkkunen and colleagues had a history of alcohol abuse. Chronic alcoholics frequently substitute alcohol for much of their normal food intake, and therefore often consume insufficient amounts of protein and essential vitamins and minerals (Shaw and Lieber, 1988). Inadequate nutrition can lead to abnormal glucose responses. Thus, it may not be that hypoglycemia results in violent behavior, but rather that a lifestyle that encompasses alcohol abuse and other behaviors that contribute to inadequate nutrition results in hypoglycemia.

A second problem is that in some studies, violent offenders were given their normal diet for three days preceding the OGTT (e.g. Virkkunen, 1983a), and in others, a hospital diet containing a minimum of 48–55 percent calories as carbohydrate (e.g. Virkkunen, 1986a). The diet of control subjects was not manipulated in any of the studies. It is thus conceivable that differences in dietary intake immediately preceding the OGTT contributed to the differences in blood glucose and insulin responses observed between violent offenders and normal controls. Future research exploring hypoglycemia and aggressive behavior should include assessments of nutrient intake for all subjects.

Another difficulty with this research is that recent work has indicated that the OGTT may not be a good indicator of the changes in blood glucose levels that occur after a normal meal (Crapo, 1985). Thus, the finding that individuals with a history of violent behavior have lower glucose levels during an OGTT does not imply that this occurs under normal feeding conditions. Measurements of glucose and insulin levels in subjects following standard meals would be useful for determining the relationship between hypoglycemia and violent behavior.

An additional problem is that although positive associations were reported between the duration of hypoglycemic responses during the OGTT and a number of measures indicative of behavioral problems, there is no evidence that violent behavior actually occurred when insulin secretion was enhanced or low blood sugar levels were experienced. Examination of mood changes and other experimental behavioral indices of aggressive impulses during the OGTT could help to resolve this problem.

Finally, as previously mentioned, all of the violent offenders studied by Virkkunen and colleagues had a history of alcohol abuse. Although, as pointed out by these investigators, alcohol may enhance insulin secretion and thus lead to a reduction in blood glucose levels, it has a variety of other effects on the central nervous system. These other actions certainly play a role in alcohol's effects on aggressive behavior (see Miczek, Haney, et al., in this volume).

In a series of studies employing a dietary replacement strategy, Schoenthaler (1982, 1983a-c, 1985) investigated the effects of reducing sugar consumption on the behavior of inmates in juvenile detention facilities. A similar experimental approach was used in all studies. At a specific point in time, the institution modified its food policy in an effort to reduce sugar intake. Typical changes in the diet included substituting honey for table sugar; molasses for white sugar in cooking; fruit juice for Kool-aid; unsweetened cereal for presweetened cereal; and fresh fruit, peanuts, coconut, popcorn, or cheese for high-sugar desserts. The dependent variable in all of these studies was the number of disciplinary actions recorded by staff members before and during the change in food policy. On the basis of these studies, Schoenthaler (1982, 1983a-c, 1985) claimed that antisocial behavior in juvenile offenders could be decreased by 21 to 54 percent if sugar intake was reduced. Because this claim has important policy implications, it warrants careful scrutiny.

The first problem posed by Schoenthaler's work is identification of the independent variable. Sugar intake is reported to be the independent variable. However, one does not have to be a nutritionist to appreciate that the dietary manipulations used were of dubious value in limiting sugar intake. Many of the dietary changes merely replaced one sugar for another (e.g. honey for sucrose). Moreover, no measurements of actual sugar intake were made in any of these studies. Thus, it is impossible to determine if the dietary alterations actually led to a reduction in sugar consumption. Intake data are essential to establish if dietary manipulations