Some Things Psychologists Think They Know About Aggression and Violence

Clark McCauley

Professor of Psychology, Bryn Mawr College; Co-Director, Solomon Asch Center for the Study of Ethnopolitical Conflict, University of Pennsylvania; HFG grantee

This article appeared in Teaching About Violence, the Spring 2000 edition of The HFG Review, a Harry Frank Guggenheim Foundation publication that examined topics of violence in depth.

There are two distinctions that are crucial in the psychology of aggression and violence. The first is impulsive vs. instrumental aggression, and the second is individual vs. group aggression. Conveying these fundamental concepts to students is essential for teaching the psychology of intergroup conflict and genocide, where the costs of human aggression are highest.

Impulsive Aggression Is Different from Instrumental Aggression

Psychologists understand aggression to be behavior aimed at harming another member of the same species, and most psychologists distinguish between impulsive and instrumental aggression. Impulsive aggression (also known as irritable, angry, or expressive aggression) is marked by strong emotion, especially anger, and is aimed at hurting another. Instrumental aggression is cooler and the hurt delivered to another is not an end in itself but only the means to some other end. Aggression in a mugging, for instance, is aimed at getting the victim’s money; aggression against Saddam Hussein’s Iraq is aimed at repelling an aggressor.

Of course a particular instance of aggression may involve both impulsive and instrumental aggression, as when a parent spanks a child for going into the street. The spanking may be aimed at keeping the child out of the street in the future (instrumental aggression), but it may at the same time express the parent’s fear and anger in reaction to the child’s danger and disobedience (impulsive aggression). Although pure cases of impulsive or instrumental aggression may be rare, it is often useful to ask about a particular case of aggression whether it is predominantly impulsive or instrumental. Studies of aggression in children, for instance, have found it useful to distinguish reactive (impulsive) from proactive (instrumental) aggression.

Sociologists and criminologists studying violent crime are also beginning to use this distinction. Psychopaths are individuals with defective emotions; they do not experience normal levels of shame, guilt, or fear. Possibly as a result of this defect, they also have weak and transitory social attachments; they treat other people as objects. Psychopaths show predominantly instrumental aggression, as they use aggression coldly, as a means of controlling others. They are likely to continue aggressive acts despite conviction and punishment, and they are over-represented in prison populations.

Compared with instrumental aggression, impulsive aggression may be more easily deterred, or at least recidivism is less for crimes involving anger. A man who kills in anger in a bar brawl is less likely to kill again than a man who kills in the conduct of a robbery. Similarly, it appears that men who assault their female partners or their children in anger are more likely to quit than those who use aggression coldly, as a means of controlling those around them. There is also evidence that impulsive aggression, but not instrumental aggression, is related to low levels of serotonin in the brain.

If impulsive aggression looks like the lesser of two evils, it may be because anger, like shame and guilt, is a moral emotion. Aristotle said that anger is a reaction to insult—a specific form of moral violation. Modern frustration-aggression theory says that anger is a reaction to any noxious experience, including frustration. However, frustration-aggression theory recognizes the importance of moral violation as justifying the expression of anger and aggression. Further indication of the importance of moral violation in anger and aggression is provided by survey data indicating that the most common occasions of anger are perceived infringements of authority or independence, or other threats to positive self-image. Aristotle and modern psychology are agreed in seeing moral violation at the center of anger and impulsive aggression.

Group Violence Is Different from Individual Violence

The psychology of individual aggression is importantly different from the psychology of intergroup conflict. In general terms, it is the least socialized who are disproportionately involved in individual violence, whereas it is often the best socialized who are involved in intergroup violence.

Individual violence is represented in the statistics of violent crime: murder, robbery, assault, and rape. Although violence associated with organized crime groups and youth gangs makes some contribution to these statistics, the overwhelming majority of violent crime offences are individual offences. Violent crimes are committed disproportionately by individuals of lower socioeconomic status—individuals poorer and less educated than average. These are often individuals who do not accept the larger societal norms that give to the state a monopoly of the use of violence. Their weak socialization may be more cultural than personal, that is, the individual may be part of a deviant subculture in which violence is more acceptable than it is in the dominant culture. Still, from the point of view of the state and its norms, violent crime is predominantly a problem of undersocialized individuals.

In contrast, intergroup violence depends upon the leadership and example of well-socialized individuals. The prototype of intergroup violence is the violence of interstate war, and studies of U.S. and Israeli soldiers show consistently that the best soldiers are above average in civilian qualifications. Despite the many films based on a “dirty dozen” of career criminals forged into an effective combat team, the reality is that success in modern war depends on soldiers with middle-class levels of intelligence and education, multiplied by entrepreneurial virtues of initiative and cooperation. As war becomes more complex, success depends more on having good people up front.

The crucial role of well-socialized individuals is evident particularly in the extent to which good people do bad things in the name of the state. It was not the worst but some of the best of American boys who dropped firebombs on Dresden, Hamburg, and Tokyo and atom bombs on Hiroshima and Nagasaki. It was no less obvious then than now that, in time of war, city bombing is killing mostly noncombatants—old people, women, and children. Similarly, studies of police and military torturers reveal these individuals to be unremarkable except perhaps for a stronger-than-ordinary sense of duty and respect for authority.

Thus it is the least socialized who are disproportionately involved in individual violence, whereas the best socialized are the foundation of intergroup violence. The distinction is important because the origins of individual violence are mostly to be found in individual differences that tell us little about group conflict. Rather the origins of group conflict are in the power of group dynamics, and the origins of genocide are in the power of the state.

Government Is the Big Killer

The twentieth century has seen unprecedented accomplishments in killing, and government is the big killer. In a century of two world wars and many smaller ones, including civil wars, about 40 million men have died in uniform. But governments also kill non-combatants, and the total of these deaths for the twentieth century is approximately 150 million. Mao killed class enemies of the people; Stalin killed class enemies and nationalities; Hitler killed Jews and gypsies; Pol Pot killed class enemies and “non-Cambodian” minorities. To these official target groups must be added those who did not accede to state power—liberal elements, former allies, personal enemies of state leaders. The striking aspect of this killing is that it is within rather than across state boundaries: Chinese killing Chinese, Soviets killing within the U.S.S.R., Hitler killing within Germany and German-occupied territories, Pol Pot killing Cambodians.

Some cases of state killing are often hidden under the label of “ethnic conflict”: Turks killing Armenians, Hutus killing Tutsi, Serbs killing Albanians. In these cases we do not have a name to hitch to the killing: no Mao, no Stalin, no Hitler, no Pol Pot. Without a name to blame, we may find it easy to ascribe the killing to “ancient enmities” of tribe or ethnic group. “Ancient enmities” implies mass hatred—most people of one group hating another group, most of one group rising up to kill the most they can of the hated group. In fact, however, mass killing typically involves relatively small groups of killers.

In Rwanda, it is estimated that only about one percent of Hutus actually participated in killing Tutsi. The Hutu elite prepared the population to accept genocide with months of radio broadcasts about Tutsi plans for domination; the killers were prepared with prioritized lists of those, Hutu as well as Tutsi, who were to be killed. The fact that the killing employed mostly low-tech weapons like machetes and clubs does not mean it was not organized. Similarly it seems likely that no more than one percent of Turks participated in killing and transporting Armenians, and no more than one percent of Serbs participated in ethnic cleansing in Kosovo.

The fact that killers are a small minority in ethnic cleansing is inconsistent with ancient-enmities explanations of ethnic conflict and genocide. Mass killing involves not a mass of individuals boiling with hate or fear, but massive organization and planning. Individual hatreds make exceptions for friends and neighbors, and individual killers get tired or run out of victims. Mass killing requires prioritizing victims, transporting and supplying killers, transporting and disposal of those to be killed. Mass killing requires the industrialization of killing with the power of the state—not impulsive aggression by the many but instrumental aggression by a few.

In contrast to government killing, terrorist killing is relatively small. A terrorist group aims to bring down the state by threatening and killing those who support the state. Since World War II, only a handful of terrorist groups have succeeded in replacing the state they were fighting: in Palestine, Algeria, and South Africa; perhaps in South Vietnam and Northern Ireland. Although states have for several decades been concerned about nuclear, biological, and chemical attacks in which terrorists would kill millions, thus far terror from below has not come close to the killing accomplished by state power—terror from above.

Intergroup Violence Is Normal Group Psychology, Not Individual Pathology

The psychology of state violence is twofold. There is the psychology of leaders who plan and order the violence, and there is the psychology of those who perpetrate the violence.

It is tempting to say that anyone who can order millions of innocent people killed must be crazy. The usual specification of the craziness is psychopathy, the diagnostic category described earlier as associated with use of instrumental aggression. Unfortunately for this comfortable view of evil as distant from normality, there is no evidence that leaders such as Mao, Stalin, Hitler, or Pol Pot were psychopaths. Indeed, it is unlikely that a psychopath, with typically impoverished social relations, could develop the relationships and loyalties required for effective leadership. The mega-killers are normal men who devoted their lives to gaining power for themselves and for an idea that justifies killing—often an ideology of class or race advancement.

Similarly, it is tempting to say that anyone who can perpetrate violence against civilians—anyone who can kill old people, women, and children—must be a psychopath. But, as noted above, some of America’s best have been capable of this kind of violence. Individual motives and individual pathology cannot explain intergroup violence; state killers and those who direct them can only be understood at the level of group psychology.

People care about groups, even groups they are not part of, like sports teams and famine victims. Research indicates that group attachments are a better predictor of political opinions than individual self-interest. White support for school desegregation, for instance, is not related to having school-age children but is related to sympathy for blacks. Voters are not asking “What’s in it for me?” but “What’s in it for the groups I care about?” This is what is meant by group identification: caring about what happens to a group. In positive identification, we want good things to happen to a group; in negative identification, we want bad things to happen. Theory and research from a number of perspectives can illuminate the ways in which group identification affects our behavior.

According to terror-management theory, our attachment to important cultural groups is our buffer against mortality. Humans are the only animals that know they are going to die. The human answer to mortality is participation in a group that will not die when the individual dies. Larger groups with longer histories and more glorious futures offer better reassurance against mortality, so family, ethnic, religious, and national groups offer more reassurance than neighborhood, recreational, and occupational groups. Terror-management theory answers the question of why individuals should be willing to die for their national or ethnic group: dying for a cultural group gives meaning to life and an answer to death.

TMT is an extension of group-dynamics theory in relation to the social-reality function of groups. As science is grounded in replication, so individual perceptions are validated by the agreement of others. If someone says she can see a star where I see none, I am ready to admit the star and my weak eyes when a third person says she can see the star.

The central concept of group-dynamics theory is cohesion, the attractiveness of the group to its members. As emphasized by TMT, cohesion is increased by the degree to which group consensus offers an antidote to uncertainty about issues of meaning and value. Cohesion is also increased by more mundane membership rewards, such as congeniality, status, and accomplishment of group goals.

High cohesion leads to strong pressures for uniformity on issues relevant to the group. This is basically a homeostatic prediction, in which differences of opinion that would get in the way of a social-reality consensus or pursuit of group goals must be controlled if members are to experience the rewards that attract them to the group. For group members, higher cohesion thus leads to greater acceptance of group norms, more rejection of deviates, and greater respect for and obedience to group leaders.

Although developed in studies of small face-to-face groups (primary groups), the predictions of group-dynamics theory can be scaled up to apply as well to larger collectivities, such as ethnic and national groups, that are so large that most members cannot be known personally (secondary groups). Identification with secondary groups is equivalent to cohesion in primary groups, and the consequences of primary-group cohesion can be predicted as well from identification with secondary groups. More identification should lead to greater acceptance of ethnic or national group norms, more rejection of those who deviate from these norms, and more respect for ethnic or national leaders.

The power of the group and the consequences of cohesion are explicated in some of the classic experiments in social psychology. In Solomon Asch’s experiments on conformity, 75% of college students denied the evidence of their own eyes to give an incorrect judgment of line length when seven other students (confederates of the experimenter) unanimously gave the incorrect judgment. This power was increased by greater uncertainty (more difficult line discriminations) and decreased by any break in the unanimity of the majority (if even one confederate gave the objectively correct judgment). If the subjects could make their judgments privately rather than aloud, conformity was cut by two thirds.

One clear implication of these results is that group influence depends partly on group power to reward conformers and punish deviates; when the group could not hear the subject’s judgment, conformity was decreased. The conformity remaining with private judgment shows a different kind of group influence, influence based on subjects accepting the judgments of others as valid information about reality. Group power to define reality is even stronger on issues of value, where there is no source of certainty except group consensus.

Both kinds of group power contribute to killing in intergroup conflict. Killers are controlled by the rewards and punishments of their group, and by their acceptance of group norms and the social reality defined by group consensus. In time of war, every state depends on coercion to make men fight: prison or death for refusing conscription, prison or death for desertion, prison or death for disobeying an order. Terrorist killers are probably less controlled by coercion than by the social-reality power of their group, inasmuch as terrorist groups only wish they had the reward and punishment powers of the state.

Even for soldiers, however, the controlling group norms tend to be more the values of the combat group than the values of the state. Research with American soldiers in particular has shown that, in the stress of combat, most soldiers fight less for cause or country or hatred of the enemy than for their buddies. The half dozen or dozen men who share the loneliness of the battlefield are closer than brothers; they fight because to do less is to endanger the group on which they are totally dependent. One of the great lessons of social psychology is that group power is maximized when group members have no other group to turn to. In combat groups as in terrorist groups, the individual’s social world has contracted to just the few around him.

A similar account has been offered for genocidal killing. Christopher Browning describes how middle-aged German men in uniform came to participate in killing Jews. The Reserve Police were sent to Poland, separated from family, friends, and previous occupations in a foreign land where their only home was their unit. Their officers had orders to kill Jews, but, at least initially, did not threaten much coercion against men who could not or would not. In this situation, it is not individual motives of anti-semitism that explain the killing. Browning points instead to the cohesion of “ordinary men,” for whom the best reason to kill Jews was that those who did less put more burden on their fellows.

Browning puts considerable emphasis also on desensitization and routinization of killing in explaining how men slowly escalated their participation in killing. This is the psychology of commitment, much studied in research on dissonance theory. In its modern version, dissonance is a theory of rationalization in which individuals sucked into stupid or sleazy behavior will change their opinions to justify and make sense of their behavior. Probably the paradigm case of this psychology is another classic study, Milgrams’s research on obedience.

Milgram showed that the majority of normal individuals will give a supposed “learner” increasing levels of shock, up to a maximum 450-volt shock labeled “XXX DANGER STRONG SHOCK.” Part of the power of this paradigm is the slow, graded nature of the shocks, which begin at only 15 volts and increase 15 volts with every mistake the “learner” makes. So close is the grading of shock levels that, at each level, to recognize something wrong with giving the next level must imply something wrong with the level already administered. Slow escalation of hurting others is a slippery slope in which each act of aggression becomes a reason for more aggression.

Taken together, group dynamics and the psychology of escalating commitment go a long way toward explaining how normal people can do awful things. Throw in the reward and punishment power of the state, a power that needs move only a small number of people to do the dirty work against a target class or race, and even genocide begins to be comprehensible.

Clark McCauley is Professor of Psychology at Bryn Mawr College and Co-Director of the Solomon Asch Center for the Study of Ethnopolitical Conflict at the University of Pennsylvania. He is an HFG grantee.

Suggested Reading

  1. Averill, J. R. 1982. Anger and Aggression: An Essay on Emotion. New York: Springer.
  2. Berkowitz, L. 1989. Frustration-aggression hypothesis: Examination and reformulation. Psychological Bulletin 106: 59-73.
  3. Browning, C. R. 1992. Ordinary Men: Reserve Police Battalion 101 and the Final Solution in Poland. New York: HarperCollins.
  4. Duckitt, J. 1989. Authoritarianism and group identification: A new view of an old construct. Political Psychology 10: 63-84.
  5. Dodge, K. A. 1991. The structure and function of reactive and proactive aggression. In D. J. Pepler and K. H. Rubin (eds.) The Development and Treatment of Childhood Aggression. Hillsdale: Erlbaum.
  6. Holmes, R. 1985. Acts of War: The Behavior of Men in Battle. New York: Free Press.
  7. Kinder, D. R. 1998. Opinion and action in the realm of politics. In D. T. Gilbert, S. T. Fiske, and G. Lindzey (eds.) Handbook of Social Psychology, vol. 2. New York: McGraw-Hill.
  8. McCauley, C. (ed.) 1991. Terrorism Research and Public Policy. London: Cass.
  9. Pyszczynski, T., J. Greenberg, and S. Solomon. 1997. Why do we need what we need? A terror management perspective on the roots of human social motivation. Psychological Inquiry 8: 1-20.
  10. Rummel, R. J. 1996. Death by Government. New Brunswick, N.J.: Transaction Publishers.
  11. Smith, D. N. 1998. The psychocultural roots of genocide: Legitimacy and crisis in Rwanda. American Psychologist 53: 743-753.
  12. Williamson, S., R. D. Hare, and S. Wong. 1987. Violence: Criminal psychopaths and their victims. Canadian Journal of Behavioral Science 19: 454-462.

Welcome to the website of The Harry Frank Guggenheim Foundation

Sign up here for Foundation news and updates on our programs and research.