Violence and Serotonin: Influence of Impulse Control, Affect Regulation, and Social Functioning

There has been a great deal of interest concerning the role of central nervous system (CNS) serotonin in aggression. Findings in various areas, including human and animal studies, pharmacological, imaging, and receptor subtyping research indicate the importance of serotonin in certain types of aggressive behaviors. The literature reviewed in this article is pertinent in understanding the relationship between serotonin and aggressive behavior in the context of three important domains related to aggression, namely impulse control, affect regulation, and social functioning.

Extensive literature dating back to the 1960s provides evidence of an association between animal aggression and reduced serotonergic function. Beginning in the late 1970s, studies focused on human aggression, examining violence toward others as well as suicide, which was viewed as a form of violence, i.e., self-directed violence. Serotonergic function was a marker and predictor of both violence and suicide in various populations. However, not all aggressive behaviors were associated with decreased serotonin function and this association needed further characterization. These aggressive behaviors were often maladaptive in some ways, whether impulsive, emotionally driven, or socially dysfunctional. Impulse control, emotional regulation, and social functioning appear to be important qualifiers of the violent behavior associated with serotonin dysfunction. They also serve to characterize further the overall functioning of the violent individuals whose difficulties are rarely limited to aggression alone.

Of these three areas, impulsivity has received the most attention. Serotonin is seen as playing a role in the inhibition of impulses other than aggressive ones. Researchers have investigated the relationship of serotonin to various impulsive behaviors and to “impulsivity” as a trait. The regulation of emotions is relevant to both violence and impulse control. Strong emotional states often accompany violent acts that are considered impulsive. Irritability, temper outbursts, or a greater emotional responsiveness to others are often present in violent or impulsive individuals. Many of these individuals have general problems with emotional self-regulation.

Social functioning is linked to violence in many ways. An aggressive individual is more likely to show various social problems, including impaired social cognition and social affiliation. Dysfunctional aggressive behavior can be viewed as representing one extreme along a continuum of social interactions; the opposite pole consisting of skilled affiliative behaviors and social cooperativeness. However, it is important not to limit the analysis to the person's social characteristics. Aggressive acts occur in a social context. The same characteristics can lead to dissimilar behavioral outcomes, depending on the attributes of other group members. The channeling of impulses and emotions are expressed in a social context and are the products of social interactions. Viewed in this fashion, aggression is a complex phenomenon that must be analyzed at the biological, psychological, and social levels of organization. Basic structures and processes at one level help define the other levels.

METHODS

Serotonin Function in Aggression

Impairment in serotonin function has been documented through a variety of methods, including studies of serotonin metabolites, especially 5-hydroxyindoleacetic acid (5-HIAA), in the cerebrospinal fluid (CSF). Low levels of 5-HIAA in the CSF indicate a low level of serotonergic activity in the brain.¹ Serotonin activity has also been assessed by peripheral measures, including plasma tryptophan and platelet serotonin uptake. Central serotonergic function has been studied through prolactin responses to agents that enhance serotonergic activity.

Recent research has focused on serotonin receptor subtypes that contribute to aggressive behavior and on genetic factors related to these subtypes. Mutant mice lacking specific receptors have been studied. Mice lacking the Type 1B serotonin receptor are more aggressive. They attack any intruder faster and more intensely than do the wild-type mice² and show greater motor impulsivity.³ In humans, various receptor subtypes have been explored that account for variability in aggression, impulsivity, and central nervous system serotonergic responsivity.⁴ There is a high density of serotonin type 2 receptors in the prefrontal cortex,⁵ an area that has been implicated in aggressive and violent behavior. Higher density of this receptor has been noted in the postmortem brain of suicide victims, especially in the prefrontal area.⁶

Animal Literature:

The serotonergic system was seen as mediating different types of aggressive responses in a variety of species, but most studies were initially conducted in rodents, where aggression was noted to increase after depletion of brain serotonin. Decreasing serotonin in rats, for example, results in increased killing of mice.^7,8 The same increase in aggression occurs with destruction of serotonergic neurons, and this effect can be reversed through administration of serotonin-mimetics.⁹

Even in rodents, no simple one-to-one relationship exists between aggression and biological variables. Aggressive behavior, once understood in terms of its function, is no longer a unitary concept. Conversely, the same function can be achieved through behaviors other than aggression; the animal, therefore, will become aggressive only when these other behaviors are unsuitable or when the animal has insufficient opportunity to develop other responses.¹⁰

As we move up the phylogenetic scales, more factors interact with serotonergic function, and serotonin itself affects a broader range of behaviors and characteristics. While low serotonin is often associated with adaptive aggressive acts in rodents, it is often related to maladaptive aggressive behaviors in nonhuman primates and humans.¹¹ Studies of nonhuman primates have helped define the type of aggression and other impairments associated with serotonergic dysfunction and the role of social factors in aggression.

The aggressive behavior associated with low serotonin in nonhuman primates is severe, unrestrained, and dysfunctional. It often results in the aggressor being wounded or killed by other monkeys, and it is associated with high levels of stress.¹² In a study of adolescent rhesus macaques,¹³ subjects with low CSF 5-HIAA exhibited more serious forms of aggressive behavior. While these more severe forms of aggression correlated negatively with CSF 5-HIAA, the total rate of aggression did not. Furthermore, the CSF 5-HIAA concentrations were significantly lower in those subjects who showed evidence of physical wounding. In other studies of both male¹⁴ and female¹⁵ monkeys, individuals with low CSF 5-HIAA exhibited severe, unrestrained aggression. In contrast, competitive, restrained aggression that is used to maintain status was not correlated with CSF 5-HIAA. Low CSF 5-HIAA concentrations measured early in life predict excessive aggression and premature death as a result of aggression later in the lives of male rhesus monkeys.¹⁶ The dead or missing animals initiated unrestrained aggression, which has a high probability of injury.

It is important to note that the impairment in these monkeys is not limited to violence but also includes social dysfunction (see below, “aggressive behavior and social functioning”) and impulsive or risk-taking behaviors. In the above study,¹³ monkeys with low CSF 5-HIAA were more likely to take dangerous, long leaps high up in the forest canopy. This was interpreted as evidence of loss of impulse control. In studies mentioned above,^14,15 CSF 5-HIAA concentrations were negatively correlated with impulsive behavior and predicted excessive risk taking later in life.¹⁶ This impulsive behavior and risk taking are positively correlated with severe, unrestrained aggression, but not with overall rates of aggression.

Direct manipulation of serotonergic function in nonhuman primates has an impact on aggression that is consistent with the correlations observed in natural settings. Pharmacologically decreasing serotonin activity increases aggression,¹⁷ while serotonergic augmentation, through dietary increases in serotonin precursors or pharmacological blockade of serotonin reuptake, reduces aggression.¹⁸

Studies of Violence and Aggression in Human Populations

The relationship between serotonin and violence in man has been established in multiple studies. The aim of this article is to understand the complex interplay between serotonin and aggression through the study of impulse control, affect regulation, and social functioning. We have therefore limited the review to studies that allow us to delineate further the behavioral disturbances that exist in the violent populations or to define more precisely the aggression associated with serotonin dysfunction or the lowering of serotonin levels. We located articles through a MEDLINE and PsychLit search from 1980 to 2002, combining the keyword “serotonin” with the terms “violence” or “assault.” We limited the review to adult populations. Suicide was considered only if it was investigated together with violence or assault directed at others. Articles were included in the final review if the measures used or the description of the aggressive acts or subject populations provided information about impulsivity, emotional regulation, or social functioning. Aggression was defined broadly and included not only actual physical assaults, but also hostile or aggressive responses on experimental tasks.

The studies are divided into two major areas:

(1) Studies of associations between aggression and measures of serotonergic function, as assessed through serotonin metabolites or through endocrine studies, diminished activity of the serotonergic system being evidenced through a blunted response to agents that normally enhance serotonin.

(2) Studies where serotonin levels are experimentally modified. These studies provide a more direct investigation of serotonin effects than correlational studies. Serotonin levels have been reduced through precursor depletion and enhanced through precursor augmentation. Data suggest that acute reduction of the precursor tryptophan may decrease brain serotonin synthesis by approximately 90%.¹⁹ Enhancement has also been achieved through administration of certain drugs, especially selective serotonin-uptake inhibitors (SSRI), though these drugs have some impact on other monoamine systems.²⁰

Methodological Issues:

In order to interpret properly these studies, one must consider methodological issues that are related to the main theme of this review. Violence, impulsivity, affect, and social functioning have been defined and measured in a wide variety of ways, including observed behavior, experimental tasks, and self-report questionnaires. These measures are far from analogous. Homicide, for example, represents a very different form of aggression than delivering electrical shocks to an imaginary opponent. Aggression in normal populations may be a distint phenomenon from aggression in violent offenders.

Scales often assess other characteristics besides the ones they purport to measure. This is particularly important with regard to the concepts of interest in this review, since measures of aggression often include items that are not aggression proper but assessments of social functioning, emotions, or impulsivity. For example, the Brown-Goodwin Assessment for History of Lifetime Aggression,^21,22 which has been used in several of the studies reviewed below, includes various social dysfunctions. The Life History of Aggression Scale,²³ which is based on the Brown-Goodwin scale, includes also difficulties in interpersonal relationships, temper tantrums, and irritability. The “aggression score” provided by these scales includes these areas of functioning.

Self-reports are often used in many of these studies. These are strongly affected by the subjects' capacity for self-reflection and their willingness to disclose symptoms. Such capacity and willingness may also vary as a function of gender or diagnosis. In populations characterized by violence and impulsivity, the results may be especially biased by inaccurate self-perceptions.

Since the range of behaviors or characteristics studied is restricted in some populations, negative results may simply reflect a range that is too narrow to assess the relationship. Aggressive acts per se do not occur frequently, and the ability to detect relationships is more limited as a result.

Studies of the Associations Between Serotonergic Function and Aggression:

An early influential study²¹ found that “lifetime aggression” and suicide were associated with lower levels of CSF 5-HIAA in a group of 26 military men. Presence of personality disorder was an inclusion criterion for the study. The subjects had lifelong psychopathology and current difficulties adjusting to military life. “Lifetime aggression” was assessed through nine categories of behaviors, only two of which consisted of actual assaults or fighting. The authors state that the other seven categories were included because they are “related to history of aggressive behavior.” They included a history of temper tantrums, school discipline problems in childhood, antisocial behaviors, problems with getting along with supervisors, and military disciplinary problems in adulthood. The “aggression score” that was negatively related to serotonin levels included these problems. In another study by the same group,²² similar results were obtained. “Life history of aggression,” as defined above, was significantly associated with lower CSF 5-HIAA levels. The 12 subjects in this study were all diagnosed with borderline personality disorder and presented with lifelong problems in various areas of functioning. A history of suicide attempts was also associated with lower CSF 5-HIAA.

One study compared 16 murderers, 22 suicide attempters, and 39 controls.²⁴ The suicidal patients, but not the murderers, had significantly lower levels of CSF 5-HIAA than the controls. Though 15 of the 16 murderers were considered “impulsive,” they did not have lower serotonin levels. There was, however, a small subgroup of five subjects who differed from the other murderers in that they had killed their sexual partners. These subjects had very low levels of CSF 5-HIAA. The authors assume that these crimes were committed in a state of intense negative affect: jealousy, frustration, or fear. Thus, they conclude that serotonin dysfunction is associated with violence that occurs “in states of emotional turmoil.” This assumption, however, remains very questionable, given the very small sample size of the homicide group and the posthoc character of these analyses.

An important group of studies by Finnish and American collaborators emphasized that low serotonin function is associated specifically with “impulsive violence.” They defined the crime as impulsive when it was committed “without provocation or premeditation.” The first such study²⁵ investigated serotonergic function in 36 Finnish violent alcoholic offenders who had committed acts of “particular cruelty.” Low CSF 5-HIAA was found in the “impulsive” group (i.e., those who did not premeditate their criminal act), but not in the “nonimpulsive” group. The authors concluded that low CSF 5-HIAA concentration may be “indicative of impulsivity rather than violence.” Of the groups studied, impulsive violent offenders who had attempted suicide had the lowest 5-HIAA. In another such study,²⁶ impulsive and nonimpulsive violent alcoholic offenders and healthy volunteers were assessed. The impulsive offenders had lower CSF 5-HIAA than normal controls. The nonimpulsive violent patients had higher concentrations than both the impulsive offenders and the normal controls. No explanation was offered for the association between nonimpulsive violence and increased serotonergic function.

The violent offenders and arsonists, whose CSF 5-HIAA findings are reported above,²⁵ were followed up for an average of three years after their release from prison.²⁷ Recidivists had lower CSF 5-HIAA than did nonrecidivists on the intial assessment.

A number of problems exist with the definition of impulsivity as delineated in these studies. The violent crime was classified as impulsive when it occurred without provocation or premeditation, i.e., if the subject did not know the victim and investigators could not detect a rationale for the aggression during the psychiatric examination. The perpetrators themselves were then termed “impulsive” on the basis of the initial impulsive crime. A combination of two behaviors, such as externally directed violence and suicide, was interpreted as evidence of stronger impulsivity. Yet, violence occurring without identifiable provocation or premeditation may be indicative of greater psychopathology and not necessarily impulsivity. Differences other than impulsivity exist between the “impulsive” and “nonimpulsive” subjects that may account for the dissimilarity in the violent behavior. The “impulsive” subjects had more pervasive problems and were diagnosed with more severe disorders. In one of the above studies,²⁶ for example, all the impulsive subjects were diagnosed with either antisocial personality disorder or intermittent explosive disorder, and almost all (93%) received the additional diagnosis of borderline personality disorder. None of the nonimpulsive patients were diagnosed with antisocial personality disorder or intermittent explosive disorder, and only 27% were diagnosed with borderline personality disorder. Clearly, these subjects differ in areas other than the “impulsive” crime they committed; and that crime itself may be reinterpreted in the context of the overall pathology. One must also consider the impact of alcohol, as most subjects in these studies, impulsive, and nonimpulsive were alcoholics. Alcohol might have a different effect on the two groups.

These investigators also studied arson, though it does not entail interpersonal violence, as they considered it an “impulsive crime.” Arsonists had reduced CSF 5-HIAA.²⁸ The definition of arson as impulsive broadens further the definition of impulsivity and renders it even more imprecise. This is discussed below.

In some studies of violent populations, impulsivity was explored as a dimension separate from violence. Other characteristics such as anger, and irritability were also studied. In one study, serotonin blood platelet uptake was lower in male outpatients with episodic aggression, as compared to nonviolent controls.²⁹ This measure was negatively correlated with ratings of impulsivity, as measured by the Barratt Impulsivity Scale, but did not show any relationship to measures of anger.

In one study,³⁰ the authors examined the association between aggression and CSF 5-HIAA concentrations in a group of 64 patients with various psychiatric disorders and no past suicidal behavior. Aggressive and nonaggressive groups were defined by a median split on a six-item history of “adulthood aggressive behavior” based on the Brown-Goodwin scale. Three of these items consisted of actual aggression, but the other three included difficulties in getting along with supervisors at work and antisocial behaviors. The aggressive group had lower CSF 5-HIAA levels than the nonaggressive group. The aggressive group had higher scores on self-reported measures of hostility, as measured by the Buss-Durkee Inventory (BDI), and impulsivity, as measured by the Schedule for Interviewing Borderlines impulsivity subscale. The CSF 5-HIAA, however, did not correlate with either the hostility or the impulsivity.

Another study reports a negative relationship between CSF 5-HIAA levels and “aggression” in 17 normal subjects.³¹ The aggression was measured by the “urge to act out hostility” subscale of a self-report questionnaire: the Hostility and Direction of Hostility Questionnaire. The investigators interpret these results as indicating that decreased serotonin turnover is associated with “aggression and dysregulation of impulse control.” Yet, there were no significant relationships between CSF 5-HIAA levels and the total hostility score or the scores on the other subscales of the questionnaire. Furthermore, the “urge to act out hostility” subscale includes other dysfunctions, such as irritability, anger, and various antisocial attitudes or behavior problems (e.g., “I don't blame anyone for trying to grab everything he can in this world,” “In school I was sometimes sent to the principal for misbehaving”).

Central serotonergic function has been assessed in endocrine studies. Coccaro and co-authors³² investigated the relationship between aggression and serotonergic function in patients with personality and mood disorders. In both groups, reduced prolactin response was related to a history of suicide attempts and alcohol abuse. In the personality disorder patients, an association between a blunted prolactin response and aggression was found, as measured by the Buss-Durkee “assault” score. A negative correlation with irritability and impulsivity was also observed. These relationships, however, were not found in patients with mood disorders. These findings underscore the need to define carefully the study population.

In another study,²³ the score on the Life History of Aggression questionnaire was negatively correlated with prolactin responses to fenfluramine in 24 male and female subjects with personality disorders. In a study of 119 nonpatient men and women³³ with no Axis I diagnoses (Axis II diagnoses were not obtained), there were negative relationships between prolactin response to fenfluramine and the life history of aggression score, as well as with impulsivity (as measured by the Barratt Impulsivity Scale) and “Angry Hostility”. These relationships were found in men but not in women. However, no relationships with any of the hostility indices, as measured by the Buss-Durkee Inventory, were discovered.

Experimental Manipulation of Serotonin Levels in Human Populations:

Experimental modification of serotonin levels provides a more direct investigation of serotonin effects than correlational studies. Both clinical and nonclinical populations have been studied in this fashion. Modification of serotonin levels in normal populations can provide important information since it demonstrates the impact of serotonin on aggression and on other relevant characteristics in the absence of preexisting psychopathology.

In one study,³⁴ the administration of the selective serotonin-uptake inhibitor (SSRI) fluoxetine to 40 personality-disordered individuals with impulsive aggressive behavior and irritability resulted in a decrease in scores on the Irritability and Aggression subscales of the Overt Aggression Scale.

In one study,³⁵ the impact of increased serotonin on hostility was examined as part of a broader investigation of its effect on affect and social behaviors. A selective serotonin reuptake inhibitor (SSRI), paroxetine, was administered to 26 normal volunteers, while 25 other subjects received a placebo. Patients with Axis I disorders, dysthymia or with a history of psychotropic medication or substance abuse were excluded. Presence of personality disorder was not assessed. Treatment reduced hostility, as measured by the Buss-Durkee Hostility Inventory, and negative affects (e.g., hostility, fear, anxiety). It enhanced affiliative behavior in a cooperative experimental task, in which the subjects had to collaborate with others to solve puzzles. SSRI-treated partners scored higher on affiliative behavioral responses consisting of increased suggestions, decreased commands, and decreased unilateral solution attempts. Significant differences between the SSRI and placebo group on all of these measures were revealed. Furthermore, the magnitude of changes in assaultiveness, irritability, negative affect and affiliation was correlated with plasma levels of SSRI.

Tryptophan depletion or enhancement has been studied in nonpatient populations. In two studies, one of college males³⁶ and the other of community males,³⁷ tryptophan depletion had no effect on aggressive behavior, measured on the basis of the intensity of electric shocks delivered to a fictitious partner. In the second study, the subjects were further subdivided on the basis of family history of alcoholism. The tryptophan-depleted individuals with a family history of alcoholism responded in a more impulsive fashion on an experimental task than did tryptophan-depleted individuals with no family history of alcoholism or than did subjects who received the balanced mixture of an amino acid.

In two studies, consisting of ten³⁸ and eight³⁹ nonpatient males, tryptophan depletion increased aggressive responding on an experimental task, which consisted of withdrawing points that were exchangeable for money from a fictitious opponent. Subjects were provoked to retaliate when their points were subtracted presumably by the fictitious opponent. In both studies, the subjects were recruited from the community; only Axis I but not Axis II diagnoses were excluded. The authors of the second study note that the subjects consisted mostly of unemployed men with possible personality disorders. In a posthoc analysis of the second study,⁴⁰ the subjects were split into high- and low-hostility groups. The increases in aggression with tryptophan depletion were limited to the four subjects with hostility scores above the median. The authors suggest that hostile men “may be more prone to behavior change induced by the perturbation of the serotonin neurotransmitter system.” In another study,⁴¹ subjects were divided on the basis of preexisting aggressive traits into high- and low-aggression groups. Both tryptophan enhancement and depletion were used. In the subjects with high aggression, acute decrease in tryptophan levels resulted in increased subjective and objective aggression, while tryptophan enhancement had the opposite effect. In the low-aggression group, on the other hand, there was no consistent effect on subjective or behavioral measures of aggression.

Interactions with other factors, such as alcohol intake, have also been considered. Pihl and co-workers⁴² investigated the effects of alcohol in combination with tryptophan depletion or enhancement. Alcohol in combination with depletion increased the shock intensity delivered to a fictitious partner. In the absence of alcohol, however, the differences between depletion and enhancement were small and inconsistent. The authors suggest that subjects with low brain serotonin levels may be particularly susceptible to alcohol-induced violence.

In summary, many of the studies report a relationship between serotonergic dysfunction and aggression. The literature, however, may suggest greater coherence and consistency than exists in reality. Reasons for this include:

Many researchers conclude that their studies have demonstrated such a relationship when, in fact, it is present only for a specific type of aggression or hostility that is characterized a postiori. The relationship sometimes exists only for one of several scales used in the study.

Measures of “aggression” often include behaviors or characteristics that are not aggressive, such as anger and interpersonal difficulties. Even when the assessment is limited to aggression, the concept still differs markedly from study to study.

The meaning and impact of serotonin dysfunction may differ considerably across studies. Serotonin dysfunction may be a marker of early trauma or alcohol abuse, which may result in aggression independently of their effect on serotonergic function. Furthermore, decreasing serotonin acutely, as is done in experimental studies, is quite different from long-term deficits that may have been present at critical developmental stages.

In many of the studies, subjects were diagnosed with personality disorder. Even nonpatient populations, often termed “normal,” may include such subjects (as only patients with Axis I diagnoses were excluded). Thus, considerable psychopathology in the subjects is present, which may contribute to the violence independently of or in interaction with the serotonergic dysfunction.

The presence of alcohol abuse and its role in eliciting the violence have not always been clarified in these studies. Alcohol is known to play a role in the emergence of violent behavior and in lowering serotonin levels.

Aggressive behavior is more likely to occur when there is preexisting psychopathology. This often includes disturbances in impulse control, emotional regulation, and social functioning. Personality disturbances and preexisting hostility are often present. In addition, substance abuse, particularly alcohol, appears to play an important role in the emergence of violence.

Impulsivity and Emotional Regulation:

In many of the above studies, the violence associated with serotonin impairment was accompanied by either impulsivity or some form of emotional dysregulation. These two domains have been studied separately for the most part, yet they often coexist in various people who are violent or suicidal. Many patients in the above studies present with personality disorder that entails disturbances in these areas. In borderline personality disorder, for example, both poor impulse control and emotional dysregulation are prominent features of the disorder and characterize many of the behaviors and interactions.

There are important areas of overlap between poor impulse control and emotional dysregulation at various levels of analysis. Most obvious, we find that strong emotional states or faulty emotional regulation often accompany the violence that is described as impulsive. It would seem that impulses are more likely to result in aggression when they are emotionally triggered. Affective states, such as depression or anger, can magnify the intensity of the underlying impulse that moves the person to action. They may also diminish reflection in the decision making process that would inhibit action. Thus in one study,⁴³ impulsive subjects who were depressed were more likely to be aggressive, while impulsive subjects who were not depressed were less aggressive than nonimpulsive subjects. Barratt⁴⁴ has emphasized the role of strong emotional states or temper outbursts in impulsive aggression. He views impulsive aggression as a “hair-trigger response” in some people who do not process information in an adaptive fashion. Action, including violence, may also relieve the tension that accompanies intense negative emotions. This may be particularly true of patients with borderline personality disorder.

In understanding the relationship between emotion and impulsivity, it is important to note that, to some extent, the boundaries between these two areas are not as clear-cut as our terminology would lead us to believe. While the emotional accompaniment of action can best be observed and characterized when there is a full emotional reaction, emotional processes can still impact on action in the absence of full emotional reactions. This may be particularly true of impulsive action. Such a view of emotion can best be conceptualized from a neurophysiological perspective. The full emotional response can be viewed as being comprised of diverse components that are separate functions of different brain areas or circuits. When only some of these are activated, the emotional response is incomplete. For example, LeDoux has postulated that emotional experience, which is part of the full emotional response, will be present only when stimulus representations, affect representations, and self-representations coincide in working memory.⁴⁵

As mentioned above, in this area of research, the concept of impulsivity emerged to characterize the violence associated with serotonergic abnormality.^46,47 While aggression is far from being a unitary concept, it can be easily defined operationally, whereas impulsivity has to be deduced from various behaviors. It has therefore remained vague and variable, often encompassing a broad range of behaviors and characteristics that overlap with abnormalities in emotional regulation and social functioning. In some disorders, the presence of certain behaviors, without further classification, is considered evidence of impulsivity. In borderline personality disorder, for example, drug abuse or suicide attempts are considered indications of impulsivity.

An important distinction must be made between two different meanings of impulsivity that have been occasionally confused in the literature. One meaning is based on the term “impulse,” which is defined as an underlying instinctive urge that incites action. Impulsivity, in this sense, denotes a failure to resist an impulse or urge that exerts undue influence on behavior, despite its harmful consequences. The “impulsive” action, however, may entail considerable planning and premeditation. In contrast, impulsivity has been frequently used to characterize quick and immediate responsivity, a tendency to respond without reflection.⁴⁸ Emphasis is placed on the mechanism of action, which bypasses or shortcuts the usual “pathways” rather than on the loss of control over urges. Serotonin dysfunction and frontal lobe impairment have been linked to both these deficits.

The first meaning is best illustrated by impulse control disorders, in which the main criterion is “failure to resist an impulse, drive or temptation to perform an act that is harmful to the person or to others.”⁴⁹ These disorders involve intense urges and fantasies that can also be accompanied by reflections and planning. In paraphilias or other disorders involving dysregulation of sexual arousal, there are also intense fantasies and urges that result in a loss of control over sexual impulses.⁵⁰

This ambiguity in definition may lead, at times, to some confusion, as is the case with arson, which has been considered the prototype of an “impulsive” crime by the Finnish and American investigators;^26–28,51 they view it as a disorder of poor impulse control rather than a form of violence. The finding of low serotonin levels in this population has been interpreted as further corroboration of the association between serotonergic dysfunction and impulsivity. While patients with pyromania have an irresistible impulse to set fires, this activity is undertaken most often with considerable advance planning and preparation, not in an unreflective fashion. Some authors, as a matter of fact, have categorized it among the least impulsive crimes.⁵²

Soubrie⁵³ has developed a rodent model of impulsivity that is consistent with quick and immediate responsivity. Impulsivity, in that model, is defined broadly as lack of behavioral inhibition or capacity to wait before action is undertaken. Under normal circumstances, external factors, such as unfamiliar situations, aversive stimuli, or lack of reward, inhibit various behaviors, including aggression. With decreased serotonin, there is decreased inhibition, and this results in “facilitation of responding.” Behavior will occur even in the presence of external factors that are usually inhibitory. Serotonin bears no relationship to any predisposition to aggression; but if aggressive impulses are already present, the individual is more likely to act upon these with decreased serotonin.

Affective states and emotions serve as an intermediary construct between serotonin and aggression, as they play a role in aggressive behavior and are influenced by serotonergic mechanisms. In many of the studies on serotonin function in aggression, various emotions, especially anger or irritability and difficulties with affect modulation were associated with serotonergic dysfunction. Irritability, anger, and rage appear to play a particularly important role in violent behavior.⁵⁴ In addition, individuals with faulty regulation of negative emotion are at greater risk for aggression and suicide.

Serotonin plays a role in various affective states, including depression, anxiety, irritability, and anger. Serotonin reduction through tryptophan depletion can induce a depressive affect.^55,56,57,58 In patients with personality disorder, prolactin responses to buspirone challenge correlated inversely with self-assessed irritability.⁵⁹ In open-label studies, the SSRI fluoxetine improved anger in various populations,^60,61 including patients with explosive outbursts of rage.⁶² In a double-blind study of patients with borderline personality disorder,⁶³ a significant decrease in anger was observed in the group treated with fluoxetine as compared to placebo. In Knutson's study,³⁵ enhanced serotonin function reduced hostility in normal subjects through a more general decrease in negative affect.

Aggressive Behavior and Social Functioning:

The relationship between violence and social functioning has been investigated at multiple levels of analysis. While information on the role of serotonin in human social behavior is limited, data on violence and social functioning in children is extensive. Caution must be exerted in generalizing between children and adults. Nonetheless, this information, when combined with knowledge of serotonin influence over aggressive behaviors and social functioning in nonhuman primates, provides a more comprehensive picture of the interplay between biological and social factors in man.

Studies of nonhuman primates have shown that while serotonergic activity is negatively related to dysfunctional violence, it is positively related to good social functioning. Monkeys with higher concentrations of CSF 5-HIAA were found to be more sociable as assessed by various measures of affiliative sociality, including time spent grooming other monkeys, time spent in close proximity to others, and number of neighbors living nearby.⁶⁴

Direct manipulation of serotonergic function impacts on primate social behavior in ways that are consistent with correlations that were observed to occur naturally. Enhancing serotonin functioning increases positive social behaviors, such as approaching and grooming other monkeys.^65,66 In addition to this increase in positive social contacts, there is a decrease in negative interactions, such as avoidance, vigilance, and solitude.⁶⁴ Reducing serotonin functioning, on the other hand, decreased positive social behaviors. The monkeys withdrew from social interactions and avoided social proximity.⁶⁷

The role of social cognition in violence is an important area that, obviously, can be investigated only in man. Social cognitive impairments often accompany violent behavior. The presence of such impairments makes it difficult for an individual to find nonagressive solutions to interpersonal problems other than aggression. Aggressive children often cannot generate flexible solutions to social problems.⁶⁸ They also misunderstand other people's intentions, misreading them as being aggressive. For example, they often misinterpret prosocial overtures as aggressive.⁶⁹ In another study,⁷⁰ aggressive boys evidenced a deficit in accurately interpreting others' intentions and attributed hostile intentions to peers. These biases and deficits were exaggerated under conditions of threat. Such social cognitive impairments may be due in part to faulty emotional processing. The misinterpretation of other people's intentions and actions may reflect some inability to simulate another person's emotional state.

We can get a better understanding of aggression and aggressive individuals by expanding the field of inquiry and taking into account the individual's overall social functioning as well as group interactions and dynamics. Given that aggression does occur, it is important to consider what other behaviors are available to the individual and the group in order to maintain social cohesiveness and promote good functioning. Humans and nonhuman primates have developed important strategies to minimize the impact of conflict and restore social cohesiveness after aggression. A few studies have considered conflict-related behaviors that prevent actual aggression from occurring or promote reconciliation after fights. An ethological study⁷¹ of preschoolers indicated that social tensions are often resolved by conciliatory behaviors, which are then followed by peaceful interactions. In nonhuman primates, various affiliative behaviors, including grooming and various forms of body contact, become more frequent after aggressive interactions. These behaviors lead to a decrease in aggression and greater social tolerance. Similarly, in a study of 5-year-old boys in Sweden,⁷² aggressive behaviors decreased, and play was promoted when one of the participants in a conflict showed affiliative behavior toward the opponent. Individuals with low serotonin may be impaired in their ability to negotiate these complex social interactions. Their excessive and out-of-control aggression may reflect their social ineptitude and poor integration within the group. Monkeys with low serotonin emigrated from their natal social groups at a younger age than did those with higher serotonin levels.⁶⁴ Reducing serotonin, in the above studies, decreased social interactions.

Not all aggressive behaviors are indicative of social dysfunction. Some children, for example, though aggressive, are not rejected by their peers. Despite their aggression, they possess social skills and are able to share and cooperate with others.⁷³ Aggressive children who are rejected, as compared with those who are not, show a variety of behavior problems⁷⁴ and prominent social cognitive deficiencies.⁷⁵ When used in an appropriate fashion, aggression can be compatible with well-integrated social behavior. Thus, social ranking, which is very important in nonhuman primate societies, requires an appropriate integration of aggression with proper social behavior. A few individuals can form alliances and cooperate with each other in order to fight and defeat another animal.⁷⁶ Therefore, a high rank within a primate society may not depend on strength but rather on social skill in obtaining support from other group members.⁷⁷ The aggression that is used to maintain social dominance ranking is not associated with low serotonin levels. In one study, female monkeys with higher levels of serotonin were more likely to attain a high social dominance ranking within their social group than females with below average CSF 5-HIAA.¹⁵

The impact of serotonin on aggression and social status was systematically analyzed in a set of experiments on vervet monkeys⁷⁸ that involved the dominant male being removed from each of several groups. Serotonin levels were then experimentally manipulated in the remaining males. They were treated with either serotonin agonists or antagonists. When serotonergic function was enhanced, the subjects became dominant in the group through increased affiliative behavior and social skills. In contrast, animals with lowered serotonergic function showed higher levels of aggression and lower social rank. The sequence of the behavioral changes shown by the treated males as they acquired dominant status paralleled those seen in naturalistic conditions. These results corroborate the differences between dominance and aggression described above. Serotonergic mechanisms seem to influence the complex social interactions that permit animals to attain a high status of dominance.

CONCLUSION

There is no one-to-one relationship between serotonin and aggression, but rather a complex interplay among various factors. Aggressive behavior cannot be easily dissociated from impulse control, affect regulation, and social functioning. Furthermore, this behavior must be analyzed at biological, psychological, and social levels of organization. Serotonin affects psychological characteristics and social interactions that have an impact on violent behavior, while psychological and social factors exert an influence on serotonergic function.

In monkeys, the dysfunctional aggression and impulsive behaviors associated with lowered serotonergic function are accompanied by social dysfunction, poor social integration within the group, and lower social status. Oftentimes, these dysfunctional individuals are excluded from the group. They are either killed or forced to leave at an early age. Better serotonergic function, on the other hand, is associated with appropriate aggression, increased affiliative behavior, good social skills, and higher social status.

Serotonin influences psychological and social factors, and psychological and social factors influence serotonin. Therefore, bidirectionality exists in this system. Psychological and social factors determine how individuals respond to serotonergic levels. In one of the above studies,⁷⁸ social structure limited the influence of serotonin on aggression and social status. Serotonin exerted its influence on hierarchical relationships among group members only when the dominant males were removed from their groups. When the original untreated dominant males were returned, they always regained dominant status. Social status itself determines the extent to which serotonin is an influencing factor. Serotonin enhanced social affiliation more strongly in monkeys that were higher in dominant social status.⁶⁶

Mutual interactions between biological and social factors are well illustrated by the behavior of a simple organism: the crayfish. In the crayfish, changes in social status are accompanied by changes in the impact of serotonin.⁷⁹ This animal has an important neuron that controls a reflex involved in the fight-flight response, and the neuron's response to serotonin differs dramatically, depending on the animal's social status and experience. Serotonin enhances the response to sensory stimuli, making the neuron more likely to activate in dominant crayfish, but it inhibits activation in subordinate animals. Furthermore, the effect of serotonin on the neuron is altered if the social status of the crayfish changes. Aggression is not determined solely by an individual's biological characteristics, but it is generated within a social context. Serotonergic function reflects and modulates social interactions. The extent to which serotonergic dysfunction is associated with aggression depends on multiple, interacting factors (e.g., other predispositions, alcohol abuse, the type of provocations, and social milieu). These factors exert also an influence on the inhibition of impulses, regulation of emotions, and social functioning.

ACKNOWLEDGMENTS

This article was supported in part by grant #58341 from the NIMH.