Social Beings, Ch. 10

SocialBeingsCh.10READING.docx

Home >Psychology homework help >Social Beings, Ch. 10

CHAPTER 10 AGGRESSION: ANTISOCIAL BEHAVIOR

Richard Russo's novel, Empire Falls, focuses on a small town in Maine. High school student John Voss—described earlier as comatose, unknowable, dressed in strange thrift-shop clothes, and without a single friend—walks into art class after an unexplained absence of several days.

Without looking at anyone, John sets his folded grocery bag down with a dull thud. … Justin Dibble is the first to speak. “Hey, John,” he says, as if this were a normal day, just another class. “What's in the bag?”

At first he doesn't appear to hear. When he finally reaches into the bag and takes out the revolver, it seem to Tick that he may have done so in response not to Justin but to a voice in his own head. The revolver looks like an antique, or maybe a stage prop, with its wooden grip and a long barrel. He points it and pulls the trigger without hesitation, and then Justin Dibble vanishes in the roar. He simply isn't sitting there anymore. (Russo, 2001, pp. 446–447) 1

John Voss proceeds deliberately to shoot the art teacher, another classmate, and the high school principal before trying to shoot himself. This scene, reminiscent of school shootings around the world, certainly depicts aggression, as most people and most social psychologists understand it.

Consider the behavior of another character in the book, Mrs. Francine Whiting, who has inherited from her late husband's family most of the small town's commercial property, and with it, the single most dominant hand in the town's finances. She has particular control over the finances of the middle-aged manager of the Empire Grill. And she has reason to resent him. As a boy decades before, he had witnessed his mother's brief affair with Mrs. Whiting's late husband. Her long-term strategy for revenge on his family is revealed as she says:

“Payback is how we endure, dear boy. Now, before you say another word in anger, for which I shall have to punish you, you'll have to stop and consider not just your own future, but your daughter's. She may require assistance with her university expenses in a couple of years, much as you did.” She paused to let this sink in. “And of course there are your brother and the others who depend on the Empire Grill for their admittedly slender livelihoods. In the end, though, it's up to you, just as it always has been.”

“Power and control. Right, Francine?” …

“Ah,” she said in mock delight, “you were paying attention to my little lessons, weren't you, dear boy!” (Russo, 2001, p. 435) 2

Many social psychologists would call her behavior just as aggressive as more direct physical violence.

Within aggression, social psychologists study wide varieties, using techniques that run from laboratory experiments on cognitive structure to archival studies of homicide rates. These data matter, as this chapter shows, because aggression raises chronic social issues involving the roles of media violence, guns, and alcohol, which we describe in order to set the societal context. Fortunately, a substantial amount of theory-based research has examined the problem of aggression. Much research starts from the core social motive of understanding, targeting the surprisingly powerful role of antisocial cognitions (think of aggressors being suspicious and quick to take offense). Other core social motives that run through aggression include frustrated efforts at controlling others (think of road rage, when all the other drivers seem intent on blocking your way) and protecting one's self-esteem (think of people becoming angry when their feelings are hurt). Significant theories and research traditions reflect each of these motives: cognitive theories of aggression (understanding), conflict theories (controlling), and image protection theories (self-enhancement).

WHAT IS AGGRESSION?

Conceptual and operational definitions, chronic social problems, and the core social motives all introduce the variety of ways that social researchers tackle aggression. Given the topic, we devote some space to the chronic social problems associated with aggression.

Conceptual Definitions

Aggression entails any behavior whose proximate intent is harm to another person. Injury must be the intent, according to social psychologists (Anderson & Bushman, 2002a; Baron & Richardson, 1994; Berkowitz, 1993; Bushman & Huesmann, 2010; Geen, 1998) and laypeople (Lysak, Rule, & Dobbs, 1989). Otherwise, parents, surgeons, and drill sergeants, who sometimes inflict mental or physical pain in the line of duty, might be viewed as aggressive, when their proximate intent is helping the other. Similarly, accidents are not aggression because they do not intend harm. And many aggression theorists point out that sadomasochism is not aggression because the victim's masochistic pleasure, not harm, is the immediate goal.

The definition, then, focuses on behavior with a particular intent. Aggressive thoughts, without the behavior, are not aggression. Behavior that harms another person, without the intent, is not aggression. Thus, contrary to popular parlance, assertiveness is not aggression. One can assert one's needs, rights, or views without the primary intent being harm to others.

Focus on the proximate (closest and most immediate) intent is distinct from focusing on the primary (ultimate) intent (Bushman & Anderson, 2001a). Aggressors operate with multiple motives, and determining the primary one would be tricky. School shootings may have the primary intent of revenge, suicide, or fame. Terrorist attacks may have various primary intents: revenge, escalating tensions, genocide, political control, moral influence, personal salvation, or publicity. Domestic abuse may have a primary intent of control, self-enhancement, relief from tension, or fulfilling a role. In the fictional example, Mrs. Whiting's apparent charity had a major element of revenge (“payback … dear boy … power and control”). In sponsoring the college education of the son and granddaughter of the woman responsible for her husband's betrayal, one might at first assume her motive was forgiveness and prosocial charity. But her dialogue makes clear that she finds power and control over the other family to be a satisfying form of revenge. Because people's motives are often complex, focusing on the most immediate, closest, or proximate goal seems more fruitful than trying to decide the primary one.

A related distinction has endured for decades but may be passing from usefulness (Bushman & Anderson, 2001a). Researchers traditionally have distinguished between two forms of aggression: instrumental aggression, which seeks harm as a means to another end (control, money, status), and hostile aggression, which seeks harm as the primary goal (e.g., Baron & Richardson, 1994). Instrumental aggression, in this view, is premeditated and controlled, whereas hostile aggression is angry, impulsive, and automatic. However, in apparent contradiction, hostile aggression sometimes can be controlled, as when an angry person plots revenge over time, and instrumental aggression sometimes can be impulsive, as when a child hits another child to get a toy back. Moreover, some researchers consider all aggression to be instrumental, to have the goal of obtaining some kind of reward, so that all aggression is coercive action (Malamuth & Addison, 2001; Tedeschi & Felson, 1994). Either way, the distinction loses its utility.

Researchers also distinguish between proactive and reactive aggression (Dodge & Coie, 1987). Proactive aggression takes the initiative: uses physical force to dominate, gets others to gang up on a peer, or threatens and bullies others. Reactive aggression instead responds to a perceived threat: when teased, strikes back; blames others in fights; or overreacts angrily to accidents. Researchers find that both types of aggression elicit social rejection from others, but they differ in their basic mechanisms, as we shall see.

Another set of distinctions creates a taxonomy of aggression (Buss, 1961; see Table 10.1 ). Although not empirically based and not currently in use, it is worth considering because it expands laypeople's assumptions about what can constitute aggression. The taxonomy points out three dimensions: physical-verbal(involving body versus words), active-passive (by doing versus not doing), and direct-indirect (targeted versus roundabout). Active, direct, physical aggression (face-to-face bodily harm) comes to everyone's mind as the prototype of aggression. More unusual examples involve passive aggression (not speaking to someone, failing to help) or indirect aggression (harm via a third party or after a delay). As we will see, verbal and physical types of aggression are highly correlated (e.g., Carlson, Marcus-Newhall, & Miller, 1989). Researchers rarely study passive aggression (a pity, as it's uncommonly irritating in a colleague or close relationship!). However, some work targets ostracism and the silent treatment, but more from the victim's responses than the perpetrator's intent (e.g., Sommer, Williams, Ciarocco, & Baumeister, 2001; Williams, 1997; Williams, Shore, & Graphe, 1998). Direct versus indirect aggression does receive some research attention, for example, in gender differences that we cover later.

TABLE 10.1 Conceptual Types of Aggression

	Active		Passive
	Direct	Indirect		Direct	Indirect
Physical	Stabbing, punching or shooting another person	Setting a booby trap for another person; hiring an assassin to kill an enemy		Physically preventing another person from obtaining a desired goal or performing a desired act (as in a sit-in demonstration)	Refusing to perform necessary tasks (e.g., refusal to move during a sit-in)
Verbal	Insulting or derogating another person	Spreading malicious rumors or gossip about another individual		Refusing to speak to another person, to answer questions, etc.	Failing to make specific verbal comments (e.g., failing to speak up in another person's defense when he or she is unfairly criticized)

Source: Taxonomy from Buss, 1961.

Operational Definitions

Social psychologists face a challenge in operationalizing the social contexts that give rise to aggression. More than any other topic, harm-doing presents ethical difficulties for researchers (Darley, 1999). Consider the Stanford prison study described in the first chapter (Haney, Banks, & Zimbardo, 1973). The actual degree of aggressive harm inflicted on role-playing prisoners by role-playing guards was severe enough that researchers terminated the study. Much of the fame of that study stems precisely from its extraordinary impact. The more impactful and realistic the setting, the more the ethical issues: One cannot inflict actual harm, merely in order to study it. Hence, most social-psychological studies of aggression use laboratory studies in which people think they are harming another person, but they actually are not.

For example, one of the classic laboratory methods (Anderson & Bushman, 1997) entails physical aggression using an aggression machine (Buss, 1961), a method simultaneously and apparently independently pioneered as a shock generator in Milgram's famous obedience experiments (1963, footnote 3; see Chapter 13 ). The method goes like this: A participant and a confederate appear to be randomly assigned as teacher and learner, respectively, in a study allegedly focused on learning processes. The learner's wrong answers are allegedly punished by electric shocks (though no shocks are actually received by the confederate). The teacher-participant can choose among levels of shock by pushing buttons labeled as barely perceptible to excruciatingly painful. This continuum allows researchers to measure intensity, number, and duration of shocks delivered. Researchers have also used other noxious stimuli, such as blasts of white noise supposedly delivered through earphones.

Critics might wonder whether the Buss aggression machine properly operationalizes aggression because, for example, the participant plays a role that socially sanctions harming the learner, so harm might not be the immediate intent. However, in aggression studies, the learner (or the experimenter) typically provokes the teacher (participant) in some fashion, and the participant's aggression then rises above the minimal level necessary to fulfill the role. More recently, a popular paradigm for studying aggression has included a competitive reaction time task, in which the winner who reacts faster gets to punish the loser (Taylor, 1967; see Giancola & Zeichner, 1995). Both the Taylor and the Buss measures of physical aggression are valid (Bernstein, Richardson, & Hammock, 1987). Another clever method involves the amount of hot sauce given to a person who hates spicy food (e.g., Lieberman, Solomon, Greenberg, & McGregor, 1999). Where tested, these operationalizations of physical aggression correlate with other measures, such as verbal aggression.

Verbal aggression measures typically include oral or written responses to a confederate (Anderson & Bushman, 1997). Researchers record and code the severity of direct verbal attacks. Alternatively, researchers elicit written ratings of a confederate, in settings where the potential aggressor knows that negative comments and evaluations will harm the confederate's future outcomes. Also, verbal measures include both explicit self-reports and implicit measures of aggressive tendencies (Richetin, Richardson, & Mason, 2010).

A critic might argue that verbal and physical measures of aggression tap entirely different forms. For instance, some people might talk tough but not act; others might act tough but not talk much. However, the same people and situations indeed do elicit both kinds of aggression, according to meta-analysis (Carlson et al., 1989). That is, when the same participants aggress in more than one way within a single study, the verbal and physical measures correlate highly. What's more, in studies that manipulate situational antecedents of aggression, the effects on physical and verbal aggression are the same.

Beyond verbal and physical expression of aggression, one might assume that physiological indicators would most objectively assess aggressive tendencies. Unfortunately, just as aggressive behavior is complex, so too are its neural correlates (Baskin, Edersheim, & Price, 2007) and its peripheral physiological correlates (e.g., heart rate, electro-dermal responses; Lorber, 2004). Although physiological indicators lend themselves to both laboratory and field studies of violence, neither setting clearly supports their use, yet. Nonetheless, the possibilities are intriguing (Carnagey, Anderson, & Bartholow, 2007).

Critics might also wonder whether the laboratory studies of aggression generalize to the real world. They do. In the real world, indicators of aggression range from horn-honking to violent crime (murder, assault, robbery, rape), all of which clearly reflect an intent to harm the victim. Researchers often use archival data (e.g., murder rate as a function of daily temperature) and observational methods (e.g., taunting or hitting in classrooms and summer camps, as a function of social status). Based on these data, clear and consistent evidence supports the external validity of conclusions from laboratory studies of aggression (according to meta-analyses; Anderson & Bushman, 1997). That is, the same kinds of individuals and situations elicit aggression both in the laboratory and in the field. Thus, the two types of evidence clearly converge. As in many other areas of social psychology, effects obtained in laboratory studies hold up quite well in the field. In any case, programmatic research that combines laboratory and field data strengthens the conclusions shared by both methods.

Chronic Social Issues: Aggression and Social Artifacts

When people consider aggression as a social problem, various images come to mind, including media violence, guns, and alcohol. Certain social artifacts(human products) facilitate aggression, as researchers from a variety of theoretical traditions have shown. Unlike other topics, the study of aggression is grounded in a series of social problems that revolve around these cultural artifacts, so we will ground our examination of aggression in the societal context provided by these applied problems.

MEDIA VIOLENCE

American society is addicted to television. In average American households with children, the television is on 28 hours per week for preteens and 23 hours per week for teens (Donnerstein, Slaby, & Eron, 1994), accounting for half of children's leisure time. Prime time shows average 5 or 6 acts of violence per hour, and Saturday morning children's programs average 20 to 25 per hour. The most violent programming appears before school (6–9 a.m., 497 violent scenes) and after school (2–5 p.m., 609 violent scenes). If children watch 2 to 4 hours of television daily, they will have witnessed over 8,000 murders and over 100,000 other acts of violence by the time they leave elementary school (Donnerstein et al., 1994). These figures do not include cable, Internet, video, or video games, which have even more graphic violence.

Research unequivocally demonstrates that media violence facilitates aggression. Although the public remains broadly isolated from this knowledge, review after review supports the conclusion: Video game violence, for example, predicts increased aggressive behavior, aggressive cognition, and aggressive affect as well as decreased empathy and prosocial behavior (Anderson et al., 2010). In 586 separate laboratory tests, exposure to media violence had a large effect on aggression; in 556 studies conducted outside the laboratory, with fewer controls, exposure to media violence had medium effects on aggression (Paik & Comstock, 1994). A more recent meta-analysis similarly found larger effects in laboratory than field studies (Anderson & Bushman, 2002b). Differences in the effect size could result from a host of factors, including the recency and concentration of exposure (higher in the laboratory) or precision of operationalization (also higher in the laboratory). For experimental designs, exposure typically consisted of viewing edited excerpts of violent programs, whereas in the field, exposure typically entailed reported frequency of watching television programs identified as violent.

Across both kinds of studies, the most frequent kinds of aggression measures involved self-reported behavioral intentions or actual actions on an aggression machine, but also interpersonal aggression during informal social interactions (e.g., hitting or pushing during play). Experimental studies in general show medium effects of media violence on aggression in the subset of settings that involve unstructured social interaction, again slightly larger in laboratory than field experiments (Wood, Wong, & Chachere, 1991). In all these reviews, only 6% to 10% of the tests show reversed effects—that is, media violence decreasing postexposure aggression—which could be interpreted as catharsis (i.e., letting off steam by watching violence instead of engaging in it; we will come back to this). The vast majority of studies show an effect significantly larger than zero, so media violence clearly facilitates aggression. The link of violent media to aggression is one of the most reliable findings in social psychology.

In one study typical of the laboratory paradigms (Meyer, 1972), undergraduate men participated in a study supposedly about grading each other's work by means of electric shocks. After writing an essay on the importance of a college education, participants received 8 shocks (representing a graded scale of 1 to 8); this presumably aroused hostility toward their partner (a confederate). While waiting for their partner to write an essay on the same topic, they participated in a supposedly unrelated study that entailed watching a 2- to 3-minute film segment. As Table 10.2 indicates, they watched either real violence (an execution with a knife during the Vietnam war), fictional violence (a movie knife fight), exciting nonviolence (a cowboy riding a half-broken wild horse), or nothing. For participants who viewed violent film segments, a voiceover either justified the violence, labeled it as unjustified, or was absent.

TABLE 10.2 Number of Shocks, as a Function of Exposure to Filmed Violence

	Justification
Type of Filmed Violence	Justified	Neutral	Unjustified
Real	7.36	4.96	4.96
Fictional	7.16	7.40	5.20
No violence		5.20
No film		4.76

After the film, participants went back to the alleged main experiment and graded an essay by assigning 1 to 10 shocks to the confederate. As the table shows, a brief prior exposure to justified violence increased the number of shocks, compared with unjustified violence or no violence. Thus, when primed by justified violence, participants were more likely to aggress against the confederate in the service of revenge. (In the neutral condition, which lacked a voiceover, the fictional violence drew more aggression than the real violence did, probably because of baseline differences in how intrinsically justified the incident seemed.)

A noteworthy recent study examined the effects of television viewing on aggression in both childhood and adolescence, tracking 707 randomly sampled families over 17 years (Johnson, Cohen, Smailes, Kasen, & Brook, 2002). A clear relationship emerged between time spent watching television (in adolescence and early adulthood) and aggressive acts. The study is one of the longest and largest longitudinal datasets and takes the measures of violence into severe forms, including assault and robbery. And it still controls for possible confounds such as previous aggressive behavior, childhood neglect, family income, neighborhood violence, parental education, and psychiatric disorders. Although it measured television viewing in general and not exposure to television violence per se, over 60% of television programs contain violence, so this is not a fatal flaw (Anderson & Bushman, 2002a), especially given the staggering number of other studies isolating the effects of violent media.

How do violent media affect aggression? Watching media violence increases angry feelings, aggressive thoughts, physiological arousal, and aggressive behavior, but decreases helpful behavior and physiological sensitivity to violence (Anderson et al., 2004; Carnagey, Anderson, & Bushman, 2007). Identifying with video warriors makes gamers especially aggressive (Konijn, Nije Bijvank, & Bushman, 2007). Essentially, media violence primes short-term accessibility of aggression, creates long-term norms about its prevalence and acceptability, provides role models, desensitizes viewers, and in the case of video games, actively involves consumers in violence (Anderson & Bushman, 2002).

For decades, the weight of evidence from hundreds of studies clearly has shown that media violence facilitates viewer aggression in the short run and the long run. Violent video games have precisely the same effects as television and film violence (Anderson & Bushman, 2001; Anderson & Dill, 2000). Given the clarity of the evidence, a disjunction between scientific findings and media coverage is puzzling indeed (Bushman & Anderson, 2001b). The empirical correlations, in a cumulative meta-analysis, show larger and even more reliable media violence effects over time. At the exact same time, news reports have presented the relationship as ever more tenuous and moderate as the studies accumulate.

Conflict of interest (pleasing their advertisers) would seem a sufficient explanation for this neglect of the scientific evidence. But the media appear to be unaware of one crucial fact. Television violence does not necessarily help their advertisers. Television violence actually decreases memory for advertisements embedded in those segments (Bushman, 1998a, 2005; Bushman & Bonacci, 2002; Bushman & Phillips, 2001). Sex and violence in the ads themselves also impairs memory for the product (Bushman, 2007). Viewers of violence get into an angry mood, and anger impairs their memory—hardly the result desired by advertisers.

Nevertheless, oblivious to this business liability, media executives and advertisers apparently believe that violence attracts viewers (which it does, especially if accompanied by warning labels; Bushman & Stack, 1996). But if the advertisers knew that the viewers were not recalling their ads, they would be less interested in buying time on violent programming. Thus, despite clear and convincing evidence—evidence almost as strong as the smoking-cancer link and even stronger than links between such well-known health effects as condom use and HIV prevention or lead exposure and IQ damage (Bushman & Anderson, 2001b)—media violence continues as standard fare, with predictable results on aggression. These results occur in Europe and Asia as much as in the United States (Anderson et al., 2010; Mummendey, 1996).

Various psychological processes contribute to violent media effects on aggression. Later in the chapter, we see how the mechanisms of violent media alter people's social motives, including how they understand aggression, how they control others, and how they enhance themselves.

WEAPONS

Another cultural artifact is the easy presence of aggressive weapons. Countries differ, for example, in the ease of gun ownership, with the United States having both the highest rate of gun ownership and the highest rate of firearm homicide among industrialized Western countries (Berkowitz, 1994). Across 16 countries in Europe, each country's availability of firearms correlates with its firearm homicide rate.

A more direct comparison is Seattle and Vancouver, two cities similar in location, size, income, education, unemployment, and, most important, rates of burglary and robbery. However, one has gun control and one does not. Over a six-year period, Seattle residents ran only a slightly higher risk of aggravated assault (1.16 to 1), but the rate of assault with firearms was fully seven times higher (Sloan et al., 1988). The risk of dying from homicide in Seattle was significantly higher than in Vancouver (1.6 times), a difference completely explained by a nearly fivefold increase in the risk of being murdered by a handgun. Rates of homicide by other means did not differ, evidence suggesting to the authors that “restricting access to handguns may reduce the rate of homicide in a community.”

Advocates of unrestricted gun ownership argue that people need guns to protect themselves, but other studies suggest that gun owners are more likely to be hurt by their own guns than to hurt an intruder. For example, keeping a gun at home is associated with a nearly threefold increased risk of homicide, in virtually all cases involving a family member or an intimate (Kellermann et al., 1993). Households with recent purchasers of handguns are at greater risk for both homicide and suicide, a risk that endures for more than five years after purchase (Cummings, Koepsell, Grossman, Savarino, & Thompson, 1997; Wintemute, Parham, Beaumout, Wright, & Drake, 1999). Guns make aggression more lethal when it does occur, not a surprising effect. But (more surprising) guns also provoke aggression in their own right, simply by being there.

Social psychologists have shown one mechanism by which guns provoke aggression. Cultural learning imbues certain objects with aggressive associations (Berkowitz & LaPage, 1967): “The presence of the aggressive objects should generally lead to more intense attacks upon an available target than would occur in the presence of a neutral object” (p. 203).

In a study allegedly involving shocks as evaluations for problem-solving performance, undergraduate men encountered a partner who delivered seven shocks, provoking the participant, or only one shock (control). When it was the participants' turn to deliver evaluative shocks, the table containing the shock key also contained either (a) a 12-gauge shotgun and a .38 caliber revolver, said to be associated with the partner's research project, (b) the same guns, said to be associated with someone else's research project, or (c) nothing. As an additional control, some of the provoked participants also saw badminton racquets said to be associated with the partner (see Table 10.3 ). The unprovoked (one-shock) participants gave the least shock in return, and the presence of weapons did not matter. For provoked participants, the mere presence of weapons, either associated or not, led to the most shocks, whereas the absence of weapons led to more moderate levels of shock. These classic results, soon known as the weapons effect, fit the idea that weapons cue more aggressive behavior than would otherwise occur (Berkowitz, 1993, 1994).

Table 10.3 Number of Shocks Delivered, as a Function of Aggressive Cues

	Shocks Received
Cue Condition	1	7
Associated weapon	2.60	6.07
Unassociated weapon	2.30	5.67
No object	3.07	4.67
Badminton racquet	—	4.60

Meta-analysis (Carlson, Marcus-Newhall, & Miller, 1990) indicates that the weapons effect is most reliable for two kinds of participants who are least likely to monitor their own behavior in the experiment. First, being naïve (according to a postexperimental interview, unaware of the hypothesis) means not being suspicious and thus not monitoring their behavior as much. Second, being low on evaluation apprehension (not worried about making a favorable impression) again means not monitoring their behavior as much.

Aggressive cues need not be limited to weapons. The generality of the aggressive cue phenomenon (i.e., beyond the weapons effect) appears most reliably in work that essentially primes people. Some work uses the same name for an aggressor and a subsequent potential target of aggression, and other work exposes people to hostile verbal cues, such as bumper stickers. All in all, uncovering this phenomenon anticipated more recent interest in the unconscious activation of motives: “Situational stimuli can exert relatively ‘automatic’ control over socially relevant human actions” (Berkowitz & La Page, 1967, p. 203).

Overall, the data clearly link the presence of guns with number of homicides, but this result is neutral about policy implications, and reasonable people can disagree. The solutions to this problem are many and beyond the scope of current scientific social psychology to determine. People favor solutions ranging from gun control to safety locks to training to increased punishment for crimes committed with firearms. The most important message, regardless of one's view of the appropriate solution, is the demonstrable link between guns and lethal aggression. What one does about it is another matter. Later, the bulk of the chapter examines the psychological mechanisms that predispose people to be violent by whatever means.

ALCOHOL

About 50% of violent crimes are committed while the assailant is intoxicated (Anderson & Bushman, 1997; Steele & Southwick, 1985). In controlled laboratory studies, likewise, intoxication facilitates aggression. Meta-analyses show a small to medium effect (Bushman & Cooper, 1990; Ito, Miller, & Pollock, 1996; Steele & Josephs, 1990; Steele & Southwick, 1985) that depends on dosage. Moderate to high doses appear to facilitate aggression, whereas low doses do not.

Moreover, the study's control condition matters to the effect size. Studies showing the largest effects compare alcohol to a placebo (an inactive substance indistinguishable from the active drug). For example, people might drink ginger ale and peppermint oil as a placebo or the same drink with alcohol added. This isolates the physiological effect of alcohol on aggression, and it is large. Comparing the placebo to a simple control, in which people correctly know they are not drinking alcohol, isolates the effects of mere expectation. That is, in both the placebo and control conditions, no one receives any alcohol, but the placebo participants think they are receiving it. Surprisingly, the sheer expectancy of having consumed alcohol causes no effect on aggression (Bushman & Cooper, 1990; Hull & Bond, 1986).

Alcohol affects aggression in part by alcohol myopia (Steele & Josephs, 1990), that is, by impairing cognitive processes (Hull & Bond, 1986; Steele & Southwick, 1985). According to Steele's inhibitory conflict model, many social behaviors occur under contradictory impulses, simultaneous desires to act and not to act. Aggressing constitutes a central example, but so can sexual adventuring, self-disclosing, gambling, and excessive drinking (see Table 10.4 ). When people drink, they undermine their ability to process conflicting responses, so they respond to only one of two conflicting impulses. This results in more extreme, less moderated behavior. Often, the inhibiting response is weaker and more complex, whereas the instigating response is powerful and simple, so it is easier to follow the original impulse when one is not thinking clearly.

TABLE 10.4 Level of Conflict and Effects of Alcohol

	Conflict Level
Behavior	High	Low
Aggression	1.32	0.17
Drinking	0.88	−0.38
Gambling	0.38	0.05
Self-disclosure	1.34	0.07
Sexual interest	0.42	−0.06

Consistent with this idea are findings that self-awareness (getting people to think about themselves; see Chapter 9 ) eliminates the differences between drunk and sober people's aggression (Ito, Miller, & Pollock, 1996). Thus, when people are drinking, they cannot think as easily about their own internalized standards, which might otherwise inhibit their aggressive or other destructive impulses. If they are prompted to think about their own standards, they can indeed control their behavior, even when drunk. Alcohol myopia accounts for the rewarding features of drinking, in that it impairs stress-inducing thoughts, particularly when one is distracted (Steele, Southwick, & Pagano, 1986). Alcohol enables people to escape from self (Baumeister, 1991a), which can be relaxing but also can facilitate aggression.

Further consistent with the alcohol myopia idea are findings from meta-analyses that the effects of alcohol on aggression are, oddly, even stronger when the partner may be able to retaliate (Bushman & Cooper, 1990). Thus, when people are sober, they notice when aggression will elicit retaliation—they see both sides of the response conflict—so they inhibit their aggressive impulses. But when they have been drinking, they ignore the possible ill effects of their own aggression. This explains how some of the worst fights occur when both parties are drunk and unable to consider the consequences.

Finally, alcohol myopia helps explain the effects of alcohol on marital violence (Murphy & O'Farrell, 1996). People with higher alcohol consumption are more likely to be physically violent toward their partners. But if they stop drinking, their violence drops to levels typical of the general population, a finding consistent with the short-term effects of alcohol on disinhibiting aggressive impulses.

Alcohol does not, by itself, create aggression. Alcohol's effect on aggression occurs when people are provoked, not distracted from it, and lack a nonaggressive alternative (Bushman & Cooper, 1990). Certain conditions exaggerate the effects of alcohol. Studies that involve fasting for at least four hours, consuming vodka, or consuming other distilled spirits show stronger effects than studies involving shorter fasts, beer, or pure alcohol. Unfortunately, experimenter expectancy effects (see Chapter 2 ) also seem to operate, making effects much bigger when experimenters are not blind to condition. This suggests using not only placebos but also doubleblind trials (in which neither experimenter nor participant knows whether alcohol or placebo is involved). Carefully designed experiments are crucial to document the role of alcohol in aggression (Pedersen, Aviles, Ito, Miller, & Pollock, 2002). Nevertheless, the facilitating effects of alcohol on aggression are already clear-cut.

CONCLUSION

The chronic social issues of media violence, weapons, and alcohol all concern social artifacts that facilitate aggression. Research shows that these cultural factors indeed contribute to real, persistent dangers to people's safety. However, people are still the aggressors, and the most frequently studied antecedents of aggression are interpersonal conflict (Geen, 1998, p. 318). Beyond social artifacts that may facilitate aggression, people themselves motivate aggression, and various social psychological theories account for it.

Core Social Motives

Social-psychological research on aggression primarily focuses on three motives: understanding, controlling, and self-enhancing. Surprisingly, understanding is the single motive most studied in the aggression field, a focus perhaps contrary to laypeople's perceptions of what should motivate aggression.

UNDERSTANDING

Even as people decry its effects, aggression often follows from shared social understandings. People learn the cultural cues for threat and retaliation. Thus, for example, people learn norms for acceptable and unacceptable aggression, in order to maintain group membership, as a later section shows. Aggression on behalf of the group cements belonging to the ingroup. Conversely, aggression in defiance of the group provokes rejection, incapacitation, or punishment. Therefore, as we will see, some aggressors are just following orderly norms regarding aggression: shared understandings of acceptable provocation, methods, sequences, and consequences. The chapter's next section takes up this kind of social learning and the resulting cognitive structures. If cognition seems like a dry approach to aggression, consider the horrifying role of education, indoctrination, and shared group understanding in teenage gangs and terrorist training camps.

Does shared understanding always shape aggression? What about people who are not socialized, who are aggressive toward others in general, even ingroup others? What about an isolate who turns on others? In most cases, such aggressors see their actions as justified retaliation for harm previously caused to them. As we will see, they are playing by the rules as they understand them. This may be viewed as a case of shared understanding gone awry. Perpetrators apply what they think are appropriate rules for aggression, a form perhaps of antisocial understanding that later sections will elaborate.

CONTROLLING

Remember Mrs. Francine Whiting's principles of “power and control” as she threatens revenge on the small-town grill owner whose family she hates? Although laypeople rarely consider aggression as control, think of a mugger holding a gun to someone's head. Sometimes aggression is clearly coercive, when people use aggression to control their outcomes or to obtain resources. As noted earlier, some researchers define aggression solely as a coercive influence tactic (Malamuth & Addison, 2001; Tedeschi & Felson, 1994). From this perspective, people have evolved instrumental aggression modules (or mods), defined as specific, inherited psychological mechanisms that potentiate aggression as a coercive response to certain environmental conditions. This is the strongest evolutionary stance for human aggression, which views aggression as a broadly applied type of social influence.

A more socially functional stance views aggression primarily as a response to perceived threat. In this view, aggression defends self, intimates, or ingroup as part of an adaptive impulse to protect. It thus entails controlling others, in this case, coercing the potential perpetrator to prevent or cease inflicting harm. This important view of aggression broadens its scope to include female aggression, more than otherwise, for example in defense of others. Given gender differences in who traditionally takes care of whom, the nature of aggression's sought-for control differs dramatically by gender. The operation of control in aggression becomes more evident upon a closer look at gender differences in patterns of aggression.

More than in any other topic within social psychology, men's and women's aggression differs, particularly their manner of aggression. According to meta-analyses, men are more aggressive than women, especially when inflicting physical, rather than psychological, harm (Eagly & Steffen, 1986) and especially when unprovoked (Bettencourt & Miller, 1996); these results occur across both laboratory and field studies (Anderson & Bushman, 1997). Women and girls are more likely to use verbal aggression, relative to men's and boy's emphasis on physical aggression (Geen, 1998). Again, Francine Whiting's tactics contrast with those of the high school shooter John Voss.

Research suggests that both boys and girls engage in socially sanctioned forms of aggression, fitting their respective gender roles in elementary school. Girls are more involved in relational aggression, that is, behaviors intended to damage another child's friendships or sense of belonging to a peer ingroup (Crick, 1995, 1997; Crick, Casas, & Mosher, 1997; Crick & Grotpeter, 1995; Crick & Werner, 1998). Examples include exclusion, gossip, or manipulation, and it starts as early as preschool. Girls not only are more likely to engage in relational aggression but also are more provoked by it when it does occur. Boys more typically engage in physical aggression.

Note that, as always, gender differences do not specify psychological processes; they could be due to social roles or biology or some combination. A social role analysis is supported, for example, by meta-analysis of gender differences in perceived consequences of aggression: Women perceive more harm to the victim, danger to self, and guilt or anxiety, so they aggress less (Eagly & Steffen, 1986). Men whose social roles include belonging to all-male groups (e.g., fraternities, sports teams) are more likely to endorse and report sexual aggression against women, according to meta-analysis (Murnen & Kohlman, 2007). A more explicitly biological analysis is supported, for example, by meta-analyses finding that higher levels of the masculine hormone testosterone are associated with male aggressive behavior (Archer, 1991; 1994; Archer, Birring, & Wu, 1998; Archer, Graham-Kevan, & Davies, 2005). And testosterone varies with the social situation; for example, male winners of male-male competitions show increased testosterone levels (Gladue, Boechler, & McCaul, 1989). Other underlying gender differences could explain differential aggression by a combination of biological and social factors: Women develop social cognitive skills earlier than men, decreasing their need for blunt, physical forms of control (Bennett, Farrington, & Huesmann, 2005).

Regardless of its specifically biological or social origins, aggression may still be viewed as an effort to control others, but who and how differs by gender. An evolutionary account posits that men have to compete for mates more than women do. Recall the parental investment theories ( Chapter 7 ; Trivers, 1971): Because of pregnancy and lactation, women have had to invest more resources in each child they produce than men do. An account based on sexual selection (the selection for male versus female reproductive success) would argue that men have a higher theoretical maximum number of progeny and therefore higher variance in number of progeny. Thus, the argument runs, men should compete for mates more vigorously (Daly & Wilson, 1988). Men can further their access to mates by aggression against other men. In this argument, men can also acquire status by a reputation for aggression, and women are supposed to seek high-status mates, to enhance the survival of their offspring. Finally, men are supposed to be more sexually jealous because paternity is more ambiguous than maternity, so they would have to guard against sexual infidelity of their mates.

These predictions have been supported in a number of evolutionary psychology studies. For example, consistent with male competition for mates, male-male aggression is more frequent than male-female aggression. Male-male aggression is directed to strangers, who might compete for mates, but male-female aggression, when it occurs, is directed toward sexual partners, presumably as a form of control (Hilton, Harris, & Rice, 2000; Kenrick & Sheets, 1993). Men report that their strategies to retain mates include threats of violence to other men and debasement of their own partners, especially the younger and more attractive their partner (Buss & Shackelford, 1997). Men who are violent against their spouses are also highly controlling, jealous, and debasing (Wilson & Daly, 1996). Also chillingly consistent with this line of argument, stepfathers are 500 times more likely than genetic fathers to kill their infants and toddlers (Daly & Wilson, 1996). Children of all ages are 70 times more at risk from stepfathers than from fathers (Daly & Wilson, 1988).

However, not all the datasets agree on this point (Temrin, Buchmayer, & Enquist, 2000). Moreover, these accounts of male violence can be explained in more social-cultural terms, as demands of male gender roles that define sole possession of women as a necessary form of honor. Evolutionary explanations can work through culture, but not all cultural practices necessarily result directly from evolutionary pressures. (We will come back to cultural differences in aggression.)

Female-male homicide occurs rarely, but when it does, it typically occurs in defense against male violence (Wilson, 1989). Women physically aggress against men more than against other women, but in both cases of female violence, the victims are more likely to be intimates than strangers (Hilton, Harris, & Rice, 2000; Kenrick & Sheets, 1993). The motives here may differ; such violence may be less an expression of control over possessions and more an effort to control (maintain) the family unit.

Whereas male responses to stress tend to focus on fight-or-flight (aggression or escape), female responses to stress focus more on nurturing and attachment activities newly termed tend-and-befriend (Taylor et al., 2000; see relationships chapter). Thus, viewing aggression as the core social motive of controlling, as in the coercive features of male aggression, finds a possible exception in female responses, which more often emphasize belonging motives. Note, however, that women protecting themselves or their offspring may be aggressive, but the underlying motive seems to differ. Of course, the evidence for relational aggression suggests that women aren't exactly heroines.

SELF-ENHANCING

Using aggression as a coercive influence tactic could follow from a motive to self-enhance almost as much as to control. Acquiring resources, status, and reputation partly occurs in the service of enhancing one's image. Rather than focusing on control over one's external outcomes, a self-enhancement explanation focuses on internal feelings of self-worth. For aggression as a coercive tactic, the explanation in terms of control is more direct, whereas the explanation in terms of self-enhancement constitutes an inference.

The related view of aggression as an effort to counteract perceived threat similarly could follow from a motivation to protect the self from damage, almost as much as from an effort to control. Here, the protection and promotion of self is more evident. What is less evident is the direction of the effect. One might think that people with low self-esteem would be the most aggressive, in an effort to repair their self-esteem, but the research says otherwise, as we see later.

Summary of Definitions, Social Issues, and Motives

Aggression is behavior whose proximate intent is harm to another person. It may be more or less hostile, instrumental, or coercive. Aggression varies on how active-passive, direct-indirect, and physical-verbal it is. Operational definitions of aggression most often entail active, direct aggression. Physical aggression studies often provoke participants and then measure their reactions on the (fake) shock generator usually called the aggression machine. Verbal measures of aggression include negative ratings that can harm the other person. Verbal and physical measures of aggression do correlate with each other. More generally, laboratory measures correspond well to field measures (observations, archival records).

Aggression interacts with chronic social issues, such as media violence, guns, and alcohol. Social psychologists have shown that each facilitates aggression in people who have been provoked. Various theoretical approaches account for these effects, as we will see. The main motives identified in aggression theories are shared understanding of social norms, controlling others, and self-enhancing, which can make aggression seem adaptive for the perpetrator if not the victim.

COGNITIVE THEORIES OF AGGRESSION: ANTISOCIAL UNDERSTANDING

A surprising amount of aggression research focuses on cognitive approaches, which may seem a cool approach to a hot topic, if ever there was one. On the other hand, consider the background that novelist Richard Russo gives to John Voss, the high school boy who shoots classmates and teachers. When he was noisy as an infant and toddler, his parents hung him in a laundry sack on a closet door to get him out of the way. And their abuse did not stop there. His warped understanding of the social world had deep roots. Similarly, Francine Whiting's idea that revenge should be visited on the descendants of the woman who wronged her is unusual in modern American culture but not so unusual an understanding in other times and places. Aggression clearly results from people's construction of social situations, and understanding (cognition) is how they get there. Social learning theories address understanding how and when to aggress, giving both actions and consequences. Attributional theories address people's understanding of why to aggress. And cognitive structure theories address what habits of aggression to acquire and why.

Social Learning: Understanding How and When

Attitudes, as described in an earlier chapter, can develop via modeling (imitating the behavior of another person) and via vicarious conditioning or observational learning (watching someone else get rewarded or punished). Initial work in the social learning of aggression emerged from an effort to show “no-trial” learning in social settings (Bandura, 1965). Normally, learning operates through a series of trials, direct experiences with actions and consequences. For aggressive behavior, one's own, direct, trial-and-error learning could prove maladaptive to the point of fatality. Thus, a more indirect approach to understanding aggression would be safer. Such an approach requires cognitive representation. Counter to many of the original learning theorists, who did not believe in cognition, Bandura argued for the function of representational processes. Learning processes are mediated by “imaginal and verbal” symbols (p. 47), an early way of talking about cognitive structures. In this view, cognitive matching processes occur when people learn by first watching others, represent that scene in their minds, and then perform the same behavior. A cognitive representation needs to bridge the gap between one's observation and one's subsequent action.

Aggression clearly can develop by imitating aggressive models, whether real people, filmed people, or cartoon characters (Bandura, Ross, & Ross, 1963a). For example, nursery school children watched a model aggress against a Bobo doll (a large inflated tippy clown with a weighted bottom). The aggression included novel and unusual forms, such as kicking the doll in the air, saying “Sock him in the nose!” and “Pow!” Observers then frustrated the children (by not allowing them to play with some desirable toys) and then watched the children in a room with other toys, including a smaller Bobo doll. Aggression was coded as imitative, partially imitative, or nonimitative. Imitative aggression included the unusual forms of aggression clearly derived from the model's behavior. As Table 10.5 indicates in the last line, children were almost twice as aggressive overall when they had observed an aggressive model than when they had observed no model (about 91 versus 54 actions). Most relevant to Bandura's theory, the pattern was clearest for imitative aggression (16 versus 3 acts) and less so for nonimitative aggression (39 versus 29). Although the imitative aggression was less frequent overall, its appearance after only one exposure to the model speaks volumes. Children demonstrably learn novel forms of aggressive behavior through modeling, thereby arguably learning how to aggress.

TABLE 10.5 Imitation of Novel Aggression from Models

	Type of Model
Type of Aggression	Live Model	Filmed Human	Cartoon	Control
Imitative	21.3	16.4	12.0	2.8
Nonimitative	34.1	34.2	49.6	29.1
Total (including partially imitative and gun play)	83.0	91.5	99.0	54.3

Source: Calculated from Bandura et al., 1963a.

In addition to modeling how, social learning teaches when to aggress: what effects aggression has. Nursery school children (in a paradigm similar to the previous example) observed an aggressor win control of toys and snacks, marching off with a sack of loot, singing “Hi ho, hi ho, it's off to play I go.” Other children saw the aggressor lose control of the toys and get spanked by the child who had originally forfeited the toys. When children see aggression rewarded, they are more likely to imitate it (Bandura, 1965; Bandura, Ross, & Ross, 1963b).

Clearly, such modeling of observed aggression flies in the face of people's hunches that violent media might provide a catharsis or harmless outlet for aggressive impulses. (We will come back to catharsis shortly.) Social learning theory has illuminated the ill effects of violent media documented earlier (Donnerstein, Slaby, & Eron, 1994; Mummendey, 1996). Violent media have the most impact under the very circumstances predicted by social learning theory: when the media violence occurs with (a) efficacy as an instrument to reach one's goals; (b) omission of observable consequences to the victim; (c) similarity between viewer and perpetrator; and (d) entailing the viewer's frustration, emotional arousal, or aggressive predisposition (Comstock & Paik, 1991).

Social learning theory has also guided work on the long-term development of antisocial behavior patterns (Patterson, DeBaryshe, & Ramsey, 1989). In particular, it addresses self-regulatory mechanisms, treating them as self-sanctions (self-rewards and punishments). For example, if people normally inhibit their own aggression toward human targets by self-sanctions, then dehumanization of the targets disinhibits aggression because the self-sanctions no longer apply (Bandura, Underwood, & Fromson, 1975). Similarly, removing self-sanctions can disinhibit aggression by other routes, such as blaming victims, ignoring their harm, linking one's aggression to a worthy cause, or diffusing personal responsibility.

Such moral disengagement (believing that ethical standards do not apply to oneself in a particular context) reduces self-censure and eases aggression (Bandura, Barbaranelli, Caprara, & Pastorelli, 1996). In some of the most evil acts of aggression, such as terrorism, these mechanisms (worthy cause, vilification of victims) clearly operate (Bandura, 1999). Moral disengagement, moral exemption (Staub, 1990), moral hypocrisy (covered in Chapter 9 ; Batson et al., 1997; 1999), socially induced evil (Darley, 1992), and socially sanctioned violence (Jackman, 2002) all name forms of violence that seem prosocial to the perpetrators but aggressive from the perspective of victims or observers. Going back to Chapter 9 's principlism, the moral principles that can motivate prosocial behavior, note that moral conviction has a darker side (Skitka & Mullen, 2002). Perceived moral mandates can lead people to excuse any means—ethical or not—in the service of their valued goals. People sometimes believe so utterly in their own righteousness that they ignore other human values and procedural safeguards. People learn cognitive representations, sanction themselves, and manage their moral self-regulation in the service of desired ends.

Because it addresses how aggression is acquired, instigated, and regulated, social learning theory addresses developmental issues more directly than much of social psychology (Bandura, 1973). Social learning theory focuses on well-learned patterns of voluntary response. It assumes that aggression has a substantially intentional component, as an instrumental activity in the service of acquiring rewards (adding a control motive to the more obvious understanding motive). However, other cognitive theories subscribe to a more automatic, less controlled view of aggression.

Cognitive Structural Approaches: Understanding What Habits of Aggression to Acquire and Why

As noted, social learning theory early on advocated a role for mental representations. This stand flew in the face of the narrowest forms of learning theory available at the time, which viewed anything except stimulus and response as an unscientific fiction. Nevertheless, it remained for other theories to spell out the nature and function of the posited cognitive structures.

AGGRESSION SCRIPTS

Children learn complex sequences of behavior by observation and by doing. Cognitive representations called scripts—habitual programs of behavior ( Chapter 4 )—encode attended material, which can strengthen by rehearsal (mental repetition), and retrieved if accessible (Huesmann, 1988; Huesmann & Eron, 1989). Violent media contribute to this process (Huesmann, 1986), but so does the child's own behavior. According to a 22-year longitudinal study (Huesmann & Eron, 1984), children with intellectual deficits are more likely to behave aggressively then and later, perhaps because they are easily frustrated by ordinary problem solving (see Figure 10.1 ). As patterns of aggression repeat, they are encoded and rehearsed, becoming more accessible in the future, a self-perpetuating process. Cognitive rehearsal of aggression—fantasizing—both predicts and is predicted by overt aggression. What's more, both rehearsal and aggression correlate with television viewing habits. A reciprocal model emerges, in which aggression, television violence, and scripts reinforce each other. Intellectual incompetence could affect several of the variables, abetting the cycle, as just noted.

Figure 10.1 Model of Mass Media Effects on Violence

Individual predispositions to aggress combine with specific learning experiences to produce stability in aggression. That is, intellectual incompetence combines with scripts learned from television and personal observation, creating persistent patterns of aggression. Aggression scripts suggest plausible behaviors, but these filter through normative beliefs about appropriate and acceptable behavior (Huesmann & Guerra, 1997). Normative beliefs about aggression and individual differences in aggression each predict the other. That is, people who aggress often believe their responses are appropriate, according to the norms.

Knowing this role of normative beliefs, the substantial gender differences in physical aggression make sense. In the Bandura social learning studies cited earlier, for example, boys were typically twice as aggressive as girls. A meta-analysis of gender differences (Bettencourt & Miller, 1996) as noted demonstrates that unprovoked men are generally more aggressive than unprovoked women. When provoked, men are more likely to perceive the provocation as intense and to view retaliation as not dangerous. Maybe some of the same processes are operating in unprovoked situations. That is, maybe men are more likely than women to interpret an ambiguous situation as a provocation and to minimize the danger of retaliation, hence their higher baseline levels of aggression.

However, at high levels of provocation, the gender differences attenuate, perhaps because the norm against female aggression is trumped by the norm allowing for retaliatory aggression. In short, normative beliefs—one component of aggression scripts—can mediate effects of gender roles on aggression.

Given all these roles for script processes in aggression, the stability of aggression over the life cycle makes unfortunate sense. Aggression at age 8 years tends to predict criminal behavior, spouse abuse, traffic violations, and self-reported physical aggression in adults (Huesmann, Eron, Lefkowitz, & Walder, 1984). Aggression across generations within one family is even more stable than aggression within one individual. This stability within individuals and families can result from any number of factors—such as social class, parenting patterns, modeling, or inherited temperament—but the script analysis provides one viable account. That is, one learns aggressive scripts from one's family and continues acting on them throughout one's life.

COGNITIVE NEO-ASSOCIATION THEORY

If social learning theory explains the ways people develop an understanding of how and when to aggress, and if aggressive scripts explain what aggressive habits people acquire over time, the cognitive neo-association theory describes why, the immediate conditions that instigate aggression (Berkowitz, 1990, 1993). People's transitory situations can activate feelings and understandings that provoke impulsive aggression.

The model hinges on associations in memory. Start with ordinary, unpleasant situations that range from frustration and provocation to noise, heat, crowding, and smells. These all lead to unpleasant feelings, including rudimentary fear or anger (see Figure 10.2 ). The negative rudimentary affects in turn activate cognitions and behavior related to fight-or-flight responses: respective impulses toward aggression or escape. Either impulse can prevail, depending on person and situation. At the same time, situational cues can influence a chain of more complex emotional reactions to aversive situations: Anger, envy, anxiety, hurt, sadness, and depression can emerge as more complicated cognitive processing unfolds. Thus, these more complex emotions result from complex cognitions, such as interpretations, attributions, and explanations. (Recall from Chapter 3 , Ordinary Personology, and Chapter 9 , Helping, that complex cognitions such as attributions of control over an aversive event, for example, lead to complex emotions such as anger and pity.) One main point here is that this model posits both rudimentary associations (unpleasant feelings, fight-or-flight impulses) and more complex associations (cognitive interpretation and complex emotions).

Figure 10.2 Model of Aggressive Cue Effects on Aggression

Source: As described by Berkowitz, 1993.

Discovery of the weapons effect, mentioned earlier, spurred this model. In an aversive situation (such as the provocation in those studies), situational cues (such as guns) instigate aggression. As noted, numerous studies have replicated this effect, at least when participants are not suspicious or self-conscious.

Recent research supports the outlines of the more general cognitive neo-association model. Automatic priming does result from exposure to a weapon (Anderson, Benjamin, & Bartholow, 1998). For example, people can identify aggression-related words faster after they have just seen a weapon name or picture. This suggests that weapons make aggressive thoughts more accessible. Violent media also make aggressive thoughts more accessible (Bushman, 1998b): After seeing a violent video, people list more aggressive associations to both ambiguously aggressive and nonaggressive words, and they identify aggressive words faster. Aggressive cues do seem to activate aggressive associations, as the model would predict.

Individuals differ in their typical levels of aggressiveness, and the cognitive associations of chronically aggressive individuals contain, indeed, more aggression. Their associations among aggressive words are stronger, as are their associations between ambiguously and clearly aggressive words (Bushman, 1996). These richer associative links help explain why aggressive individuals respond more strongly to violent media: choosing to watch it more often, becoming more angry when they do watch it, and aggressing more afterward (as illustrated in Chapter 2 , Bushman, 1995; Josephson, 1987).

Gender differences also provide indirect support for the cognitive association model of impulsive aggression. When people ruminate, mood-relevant thoughts remain accessible, and angry moods are no exception: Angry rumination primes angry thoughts and actions (Bushman, Bonacci, Pedersen, Vasquez, & Miller, 2005). Women are more likely to distract themselves from ruminating when they feel angry than when they are feeling neutral, but men do not (Rusting & Nolen-Hoeksema, 1998). This could contribute to gender differences in aggression.

Gender differences and individual differences combine. For some men, a constellation of associations links power motivation, sex, and aggression. For men identified as likely to sexually harass or attracted to sexual aggression, power is linked to sex (Bargh, Raymond, Pryor, & Stack, 1995). In these men, high levels of power motivation and strong power-sex associations predict typical forms of sexual aggression, which are direct (seeking younger or drunk women who could not resist sexual advances as easily, punishing sexual resistance with silence or changed feelings, or giving expensive gifts with the clear expectation of sexual favors in return). In women, power combines instead with strong intimacy-affiliation motives to predict sexual aggression, which tends to be more indirect (dressing seductively or playing hard to get, as deliberate seductions) (Zurbriggen, 2000). Although both men and women can link aggression and sex, priming with sex-related words facilitates aggression only for men (Mussweiler & Foerster, 2000). For women, priming sex facilitates perceptions of aggression in others. Overall, given different social learning, the genders differ in their associations among power, sex, and aggression. Demonstrating consistent clusters in the cognitive networks of associations among aggressive thoughts and feelings lends support to the cognitive neo-association theory.

GENERAL AGGRESSION MODEL

The similarities among the social learning and cognitive structural (script and neo-association) approaches led to the general aggression model. It relies on the idea of knowledge structures (expectations in Chapter 4 ), linked to affect, arousal, and behavior (Anderson & Bushman, 2002a). The model focuses on a single event or episode, involving a person-in-a-situation unit. Within an episode, people make automatic appraisals, rapid evaluations of threat in the environment (after the appraisal idea in emotion theories; see Zajonc, 1998). Such automatic priming clearly cues aggression (Todorov & Bargh, 2002). People also make controlled reappraisals, more complex cognitive analyses. Aggressive inputs (personality and situation) are mediated via these processes to determine aggressive (or nonaggressive) responses. The model is entirely consistent with the other cognitive theories of aggression, so data that support them also support it, and vice versa (e.g., Anderson, 1997; Lindsay & Anderson, 2000). But it has a more up-to-date understanding of social cognitive processes.

Some of the model's original impetus came from observing the link between high temperatures and aggression, which is well established in field settings (see Figure 10.3 ; Anderson, 1989; Anderson, Bushman, & Groom, 1997). In the laboratory, high temperatures sometimes cause aggression (Anderson, Anderson, Dorr, DeNeve, & Flanagan, 2000; Baron & Richardson, 1994), and low temperatures also increase hostility (Anderson, Anderson, & Deuser, 1996) and sometimes aggression (Anderson et al., 2000). Curiously, the low-temperature effect does not occur in the field; perhaps societies are better at counteracting the negative effects of cold by warming people than at counteracting the negative effects of heat by cooling them. In any case, when people are hot, they experience discomfort and hostility, so they are more easily provoked or cued to aggress. This provides indirect support for the general aggression model's predictions about automatic appraisals.

Figure 10.3 Monthly Occurrence of Assaults

In one series of studies testing many parts of the model, participants were selected as falling into the top or bottom third of the trait hostility scale (from a mass pretest, including items such as “I easily fly off the handle with those who do not listen or understand.” And “I often feel like a powder keg ready to explode.”). Thus, trait hostility was one independent variable. As another independent variable, half the participants were randomly assigned to the pain condition, which consisted of holding their nondominant arm horizontally in the air for 2.5- to 3-minute intervals, with 30-second rests, throughout the experiment (try it); control participants did nothing. Pain participants rated their discomfort at 3.31 on a 5-point scale, not major pain, but twice as uncomfortable as the controls (1.65). Finally, as a third independent variable, some participants viewed pictures of weapons (aggressive cues) and some viewed pictures of nature.

The individual difference variable, trait hostility, had the most consistent results across measures, sometimes interacting with the situational variables (pain and aggressive cues). People who scored high on trait (chronic) hostility not surprisingly (a) also rated themselves as higher on state (temporary) hostility (e.g., outraged, burned up, discontented, not cooperative, not friendly), and (b) they wanted to escape. More interesting from the model's interactive predictions, trait hostility predicted (c) more accessible aggressive thoughts, mainly when provoked by pain and the aggressive cue, and (d) more aggressive punishment of an opponent during a reaction-time game, especially when already primed by the aggressive cues. These results are consistent with the general aggression model, but also its forebears, the cognitive neo-association model, the script approach, and the social learning approach.

Attributional Approaches: Understanding Why

As noted at the outset, both laypeople and scientists define aggression as requiring intent. Interpreting one's own and another's intent thus commences a potentially aggressive encounter. Attribution theories ( Chapter 3 ) describe how people understand intent, and two major cognitive theories have addressed the role of attributions in aggression. The first focuses on people's attributions about others' intent. The second focuses on people's attributions about their own aggressive feelings. Both fit the cognitive structural assumption that people mentally represent the social conditions for aggression, in this case aggressive intent.

HOSTILE ATTRIBUTION BIAS

All the cognitive theories discussed so far bear on the core social motive of understanding, but all produce an anti social understanding, a warped version of shared social understanding. The aggressive perceiver may view this understanding as shared, or at least as socially justified, but it is not. In fact, aggressive people have particular biases in their understandings, which encourage aggression, bad reputations, and a spiral of rejection.

Compared with other children, chronically aggressive children tend toward a hostile attribution bias (Nasby, Hayden, & DePaulo, 1979), interpreting ambiguous behavior as aggressive. In an early study (Dodge, 1980), peers and teachers nominated various boys as generally more or less aggressive. The boys then experienced a negative, frustrating outcome, portrayed as resulting from an anonymous peer's hostile, benign, or ambiguous intent. All boys accurately interpreted the clearly hostile or clearly benign intents, but aggressive boys saw ambiguous intent as more hostile than did the nonaggressive boys. These biases result in more aggressive responses (Dodge & Frame, 1982).

The cycle partly reflects the lived experience of aggressive boys, who are indeed frequent targets of peer aggression. However, aggressive boys act as perpetrators even more often than as targets of aggression. The hostile attribution bias thus seems to exaggerate a selective aspect of their experience. How does it do so? In one study (Dodge & Tomlin, 1987), children heard stories in which (for example) a peer spilled milk on them, and various eyewitnesses gave mixed testimony suggesting intent both hostile (“I saw him laughing at you”) and benign (“He wasn't looking where he was going”). Participants then interpreted the peer's intent as hostile or benign and explained why. Their explanations mostly included the cues provided as well as other people's typical behavior toward themselves personally. As Table 10.6 shows, aggressive children were more likely to attribute hostile intent. In doing so, compared with nonaggressive children, they used less of the relevant evidence (second column of data). And they specifically weighted benign cues relatively less (third column) and self-schemas relatively more (last column).

TABLE 10.6 Hostile Attribution Biases and Underuse of Social Cues

Participant Group	Attributing Hostile Intent (Percent of Participants)	Use of Relevant Cues (Percent of Participants)	Breakdown of Cues Actually Used (of All Cues)
			Benign	Hostile	Self-Schema
Aggressive	44%	67%	.34	.33	.33
Nonaggressive	30%	79%	.54	.26	.20

Oddly, they did not weigh aggressive intent relatively more than other children do. The problem seems to be a failure to notice and credit benign explanations. In judging self-relevant social situations, aggressive boys respond faster, which causes them to attend and to recall less (Dodge & Frame, 1982; Dodge & Newman, 1981). These deficits and biases increase under social anxiety and personal threat (Dodge & Somberg, 1987).

The dynamics suggest that certain boys react defensively to others, partly based on bad experiences that become self-fulfilling prophecies. Aggressive children develop normative beliefs, akin to aggression scripts, about the permissibility of retaliation. Belief in retaliation predicts biased understanding of aggressive situations a year later, which also predicts aggressive behavior a year later (Zelli, Dodge, Lochman, Laird, & Conduct Problems Prevention Research Group, 1999). The problematic situations revolve around social interpretations of provocation. That is, the hostile attribution bias occurs in reactive (responsive) aggression, not proactive (initiated) aggression (Dodge & Coie, 1987), which are distinct patterns of antisocial behavior (Dodge, Lochman, Harnish, Bates, & Pettit, 1997). Effective interventions can target hostile-attribution biases, generate more socially competent responses, and devalue aggression (Dodge, Godwin, & the Conduct Problems Prevention Research Group, 2013).

The hostile attribution bias spotlights the social construction of aggression (Mummendey, 1996). Certain pairs of boys react aggressively to each other, beyond their own individual aggressiveness toward others (Hubbard, Dodge, Cillessen, Coie, & Schwartz, 2001). For example, initiators and responders both may view their own behavior as appropriate and the other's as inappropriate. Attributions of aggressive intent carry evaluative overtones, as forms of social understanding.

EXCITATION TRANSFER, NOT CATHARSIS

Arousal has surfaced in several contexts already, some specific to aggression and some related to other affect-laden behavior in this book. The rapid and incomplete attributions of aggressive children under perceived threat suggest a state of chronic arousal and vigilance that other people experience only sporadically. We saw a role for short-term arousal in the general aggression model as well. In discussing attribution previously ( Chapter 3 ), we saw arousal combine with people's attributions to produce their feelings in general: Schachter's two-component theory defines experienced emotion as resulting from cognition (attribution) plus arousal. In that context and in discussing passion in close relationships ( Chapter 8 ), we described studies in which people misattributed fear-based arousal (from a suspension bridge or a horror movie) to heterosexual attraction (Dutton & Aron, 1974; Zillmann et al., 1986).

The phenomenon of excitation transfer documents how arousal from an irrelevant prior source (e.g., the bridge or the movie) then persists and spills over to the next setting (e.g., attraction or provocation), to which it is then misattributed (Zillmann, 1988). Some surprising misattributional mix-ups occur. For example, exposing undergraduate men to explicit erotica (pornography) increased their aggression in response to provocation (Ramirez, Bryant, & Zillmann, 1982). That is, the students saw a short, sexually explicit film (or sexually suggestive photographs, or nonerotic snapshots of faces), with an experimenter who repeatedly accused them of not following instructions. When later given a chance to evaluate the experimenter to a university committee that would allocate research funding, they expressed more annoyance and delivered more hostile ratings after viewing the explicitly sexual film than in the two control conditions. The excitatory potential of the erotica continues afterward and then is misattributed to the provocation, so the combination particularly increases retaliation (Zillmann, Bryant, & Carveth, 1981). Women show the same aggressive response to explicitly erotic films as men do (Cantor, Zillmann, & Einsiedel, 1978). However, milder sexually oriented materials (which are not arousing) have, if anything, a dampening effect on people's retaliatory aggression (Zillmann & Sapolsky, 1977) because aggression and romantic affection are incompatible responses.

Other sources of arousal support the misattribution of arousal resulting in aggression. For example, the effects of extreme heat on aggression are consistent with the excitation-transfer idea (Anderson, 1989). For misattribution to occur, the actual source of arousal (excessive heat) must recede from awareness, so that arousal can be attributed instead to provocation. Noise also arouses people and makes them aggressive, but only in the context of provocation (Donnerstein & Wilson, 1976).

Violent media make people more aggressive, as we saw earlier, and arousal is one conduit for its effects. As noted earlier, the competing hypothesis has always been catharsis: the idea that aggressive feelings could be expressed vicariously and so diminished in actual encounters. Scientific evidence does not support the catharsis hypothesis (Bushman, Baumeister, & Stack, 1999; Geen & Quanty, 1977). For example, aggressive video games, as noted, do not help the player let off steam but instead increase aggression in both the short and long term (Anderson & Dill, 2000). Standard attempts at catharsis instead maintain the aggressive arousal and elaborate the cognitions that justify the aggression. Nevertheless, people continue to believe in catharsis, apparently holding the belief that venting anger will make them feel better, although it in fact makes them more aggressive and does not reduce negative affect (Berkowitz, 1962; Bushman, Baumeister, & Phillips, 1999; Bushman, Baumeister, & Stack, 1999). What's more, even if level of excitation holds constant, people who vent become more angry and aggressive (Bushman, 2002). Angry people may experience transitory pleasant feelings when they aggress and may sometimes deter the person who provoked them. But these positive experiences only reinforce future aggression (Baron & Richardson, 1994).

Summary of Cognitive Theories

Antisocial understanding develops through the same routes as shared social understanding: learning, mental representation, and subsequent interpretation. The understandings promoted by some people and some situations encourage aggressive reactions. Social learning theory teaches people how and when to aggress, through modeling of imitative aggression and vicarious learning of sanctions. Cognitive structures such as aggression scripts develop through rehearsal of aggressive episodes, teaching normative beliefs that stabilize patterns of aggression. Cognitive neo-association theory links affect to the cognitive structures of aggression. The general aggression model, consistent with all the preceding ones, posits knowledge structures that aid automatic appraisals, which may be modified by controlled appraisals. All these kinds of prior knowledge about aggression affect people's attribution of aggressive intent to others and to self. Some people are biased toward hostile attributions about other people, underusing benign cues, especially when threatened. In attributing aggressive feelings to oneself, prior irrelevant arousal may be misattributed to provocations, encouraging aggression. All these types of understanding—learning, cognitive structures, and attributions—build on theories covered in earlier chapters but given new significance in this context.

CONFLICT: CONTROLLING OTHERS

Whatever their basis of understanding, aggression serves people's efforts to control their outcomes by controlling other people's behavior. Some accounts of instrumental aggression, as noted, aim to express this kind of motive. Accounts of aggression as coercive influence, also described earlier, more explicitly express this viewpoint. Aggression as control especially emerges in the classic frustration-aggression hypothesis and in newer work on bullying and intimate abuse.

Frustration-Aggression Hypothesis

According to the original frustration-aggression hypothesis, frustration typically leads to aggression in some form, and all aggression results from frustration of some kind (Dollard, Doob, Miller, Mowrer, & Sears, 1939). In this work, frustration is the blocking of any goal-directed sequence of behavior. Early evidence was often flawed by confounding frustration with other factors such as anger (Baron & Richardson, 1994). Although another interpretation would be that anger mediates the effects of frustration on aggression, that was not the original hypothesis.

People certainly do sometimes aggress as a response to other people thwarting their goals. In one study (Geen, 1968), some male undergraduates underwent initial instigation to aggression: rewards for shocking a peer in a teacher-learner paradigm; others shocked but received no rewards. They also either (a) worked on an unsolvable puzzle, (b) worked on a solvable puzzle but with constant interruptions, (c) worked on the solvable puzzle but with subsequent insults by a confederate, or (d) experienced neither frustration nor insults. Their aggression against the confederate (who acted as the learner) was low in the control condition (d), moderate in the two frustration conditions (a and b), and highest in the insult condition (c). Thus, both task frustration (unsolvable problems) and social frustration (interruptions) certainly can lead to aggression. (Rewards also increased aggression.)

Despite a few promising early studies, the frustration-aggression pattern has proved elusive and unreliable in the laboratory. At a minimum, some moderators(facilitating and inhibiting conditions) do predict when frustration does indeed provoke aggression (Baron & Richardson, 1994). In agreement with the cognitive approaches, all the moderators relate to the types of attributions frustrated people might make. For example, the magnitude of the frustration matters. Larger frustrations facilitate aggression perhaps because they are more likely to trigger a (perhaps biased) search for an explanation, which would favor dispositional attributions about the agent of frustration and so facilitate aggression. Similarly, arbitrary frustrations, lacking justification, can prime both hostile attributions and aggression. The presence of aggressive cues, as in the Geen study, also facilitates aggression. Here, too, aggressive cues may prime hostile attributions. So one mediator (mechanism) shared by all the moderators might be the attributions triggered.

The moderators—magnitude, arbitrariness/justification, and aggressive cues—all fit another mediator. Frustrations may provoke aggression only when they arouse negative affect (Berkowitz, 1962, 1989, 1993). The greater the frustration's magnitude, perhaps the greater the negative affect, all else being equal. As another example, if someone constantly interrupts for a justified reason (hearing difficulty), rather than for an arbitrary reason (whim), that could elicit pity rather than anger. People also examine justifications for their own negative situation and feel accordingly. The attributional models of helping ( Chapter 9 ) address the role of justification regarding people's negative situations and the resulting feelings of anger or pity. Finally, in keeping with the cognitive neo-association model, frustration is an aversive event, and if interpreted as an aggressive cue, it can prime primitive anger and aggression.

Direct aggression constitutes only one prong of the frustration-aggression hypothesis, which specifies aggression in some form (not necessarily directed toward the agent of frustration). Displaced aggression—directed toward someone other than the instigator of frustration—also fits the frustration-aggression hypothesis. The results here are clearer than those for direct aggression. A meta-analysis of laboratory studies (Marcus-Newhall, Pedersen, Carlson, & Miller, 2000) finds a sizeable effect of provocation on aggression toward an innocent third party when retaliation against the provoking person is not possible. Frustrations on a task, for example, produce aggression, just as much as do various kinds of provocation (such as negative evaluations, verbal attacks plus shocks) and irritants (such as odors and extreme temperatures). In accord with the affective mediators proposed by Berkowitz's cognitive neo-association model, feelings of displeasure likely mediate the effects of provocation and perhaps frustration on displaced aggression (Pedersen, Gonzales, & Miller, 2000).

In the field, evidence for displaced aggression has provided a real mystery. An early study found awful correlations between drops in the price of cotton in the U.S. South and white lynching of blacks (Hovland & Sears, 1940); this suggests displacement of aggression from economic frustrations onto an outgroup scapegoat. Some subsequent studies have corroborated these results (for reviews, see Berkowitz, 1993; Geen, 1998), showing community-level effects of job loss on violence, for example (Catalano, Novaco, & McConnnell, 1997). However, the group-level frustration-aggression results remain controversial and difficult to replicate (Green, Glaser, & Rich, 1998). Moreover, at these societal levels of analysis (i.e., macro), the mediating psychological processes (i.e., micro) remain unidentified.

Work on displaced aggression in the laboratory, however, remains convincing, so we know that at least sometimes it can occur. In any case, the frustration-aggression research fits other models of aversive situations or retaliation scripts, at least when mediated by hostile attributions or anger.

All the work inspired by the frustration-aggression hypothesis relates to the core social motive of control. When someone else thwarts one's goal-directed activity, the conflict is precisely about control. Aggressive responses that attempt to remove frustrating blockages obviously stem from direct efforts to control someone who is (in that respect at least) in a position of strength. Less obviously, displaced aggression also may result from indirect efforts to restore a sense of control. In other domains—for example, interpersonal perception—diminished control in one setting prompts efforts to create control in other, irrelevant settings (Pittman & Pittman, 1980).

Controlling the Weak

Displaced aggression often focuses on a weaker, safer target than the frustrating agent. Related work on bullying and partner abuse shows aggression against others as a pure exercise of control over someone in a weaker position.

BULLYING: PEER ABUSE

Some people, bullies, consistently initiate aggression against weaker people, victims. Such proactive aggression is not unique to particular personal relationships but seems to be a function of particular actors likely to bully everyone they can and particular targets likely to be victims of several bullies (Hubbard et al., 2001; Juvonen & Graham, 2004).

On the bully side, some studies suggest that bullies are disagreeable, controlling people in general. Some children endorse beliefs that aggression (hitting, yelling, pushing, insulting) is okay, especially when they are angry. These children are disproportionately aggressors (Huesmann & Guerra, 1997), consistent with the script model emphasizing normative beliefs. Bullies tend to be impulsive, dominating, strong, and aggressive (Olweus, 1995). Among boys, they may seem to have high self-esteem, as viewed by self and peers, but it turns out to be the defensive sort: The bully always wants to be the center of attention, thinks too much of himself, and cannot take criticism (Salmivalli, Kaukiainen, Kaistaniemi, & Lagerspetz, 1999). (Adolescents with genuinely high self-esteem defend victims against bullies or avoid the situation.)

Chronic victims differ by gender. Victimized boys often have low self-esteem, across all measures, and victimized girls may show humble pride, having high self-esteem, though their peers rate them as having low self-esteem (Salmivalli et al., 1999). (The opposite combination, self-belittlers, who do have low self-esteem, though their peers think otherwise, tend to stay out of the situation.) In general, victims seem to be anxious, insecure, cautious, sensitive, quiet, and weaker (Olweus, 1995).

Teasing can be one form of verbal bullying. Although teasing contains ambiguous combinations of aggression and humor, and it may fail the aggression test by the actor's stated intent not to harm, it is often experienced by victims as hurtful. In general, teasers see their motives as benign and friendly (Keltner, Young, Heerey, Oemig, & Monarch, 1998; Shapiro, Baumeister, & Kessler, 1991), although victims may not, especially among children. And the victims may be onto something, given the personalities of teasers. For example, disagreeable, unreliable, extraverted people tease more (Georgesen, Harris, Milich, & Young, 1999). And disagreeable men tease in more hostile ways than do other men and all women (Keltner et al., 1998). They tend to threaten the other person's public image (face) more and not to redress (not to provide sufficient cues that the tease is playful, off-record, and not serious).

Although personality may differentiate who becomes bully (or teaser) and who becomes victim (or target), situations matter too. Sanctions from the environment determine whether or not bullies act on their aggressive impulses. Collective norms communicated by salient peers influence bullying (Paluck & Shepherd, 2012). Schools can also cut down on bullying when adults are warm, interested, involved, and clear that bullying is unacceptable (Olweus, 1995, 1997). Schools are wise to prevent bullying, which correlates with depression and loneliness, predicting absenteeism, low GPA, social maladjustment (Juvonen, Nishina, & Graham, 2000), and even suicide (Sugimori, 1998). The control the bully exerts over the victim, who in turn feels helpless, can be vicious.

WORKPLACE VIOLENCE

We've all seen the tragic news story of someone “going postal,” shooting coworkers and self, if the police do not arrive in time. This evokes images of a random, unpredictable, mentally ill shooter—previously known to be quiet and isolated—who tracks down a supervisor after being laid off, and meanwhile has been secretly abusing family members. In this view, workplace violence cannot be prevented because it is all too random.

On the contrary, according to accumulated research (Barling, Dupré, & Kelloway, 2009), workplace violence involves an angry, hostile shooter with a history of violence who kills because of perceived interactional justice (that is, interpersonal treatment). But distributive justice (actual outcomes such as layoffs) or procedural justice (unfair practices) do not in fact predict workplace aggression. Thus, it is control that aggressors seek, but interpersonal more than procedural or tangible control. Nor, by the way, is the Post Office the site of disproportionate workplace violence; on the contrary, it is actually much safer and more secure than other worksites.

Who are the targets of workplace violence? Targets of directed aggression are often specific, but displaced aggression may endanger others caught in the vicinity. More routine workplace violence—that is, peer bullying—targets those perceived to be weak but threatening, suggesting a role for control motives. For example, workers with low self-esteem but difficult personalities often are victims (Aquino & Thau, 2009). However, because the results are correlational, workplace bullying could lower a person's self-esteem and make the person seem demanding and difficult.

Situations that encourage workplace aggression also undermine people's sense of control. Role conflict, role ambiguity, and lack of control correlate with workplace victimization (Aquino & Thau, 2009). Being victimized is associated with a host of negative psychological effects, including depression, anxiety, stress, shame, and physical illness (Bowling & Beehr, 2006).

BATTERING: PARTNER ABUSE

The dynamics of domestic abuse, like those of peer abuse, partly reflect the dynamics of control. Aggression in close relationships, as we saw in Chapter 8 , recruits some familiar mechanisms: hostile attribution, social insensitivity, and domineering personality. In another twist of misattribution, abusive men blame their partners for their own negative feelings and ruminate on this, resulting in anger (Dutton, 1995a). Blaming attributions, along with feeling emotionally out of control, can result in instrumental, coercive violence that seeks to eliminate those negative feelings.

Attachment styles of abusive men tend to be insecure, mostly anxious and fearful (Dutton, 1995b, 2000; Dutton et al., 1994). Both styles involve a negative and unstable sense of self, which may result from many abusive men's recall of frequent public shaming by their parents (Dutton, 1995b, 2000; Dutton, van Ginkel, & Starzomski, 1995).

Finally, modeling enters into intimate violence; as is well known, exposure to parental violence, vicarious or personal, leads to own violent behavior later (Dutton, 2000), as social learning would predict. Outside the current scope lie some issues, both clinical (e.g., borderline personality disorder) and developmental (parent-child relations). Nevertheless, other aspects of intimate violence carry familiar themes of antisocial understanding, as well as attempts to restore felt lack of control.

Controlling the Strong: Terrorism as Frustration-Aggression

As I drafted the first edition of this chapter, the events of September 11, 2001, had just transformed our nation and the world. No one would doubt that suicide hijacking and anthrax dissemination constitute aggression of the worst kind, targeted at large numbers of innocent people.

At first, social psychologists (and others) understandably focused on empirical analysis of the reactions of victims, not the dynamics of perpetrators, partly because of feasibility (Byron & Peterson, 2002; Fredrickson, Tugade, Waugh, & Larkin, 2003; Galea et al., 2002; Mehl & Pennebaker, 2003; Silver, Holman, McIntosh, Poulin, & Gil-Rivas, 2002). But this chapter, like most social psychology of aggression, focuses on the aggressor, not the victim, so what light can we shed on terrorism here? As social psychology, this chapter speaks mainly to interpersonal aggression and less to either intergroup hostility (but see the next chapter) or the interpersonal processes that allow individuals to participate in large-scale aggression against civilians.

Nevertheless, some social-psychological analyses are beginning to explain how people become terrorist aggressors against noncombatant populations. First, terrorism is a political and combat tool, like any other tactic (e.g., aerial bombing of military bases). As a tool, it serves certain goals (Kruglanaski & Fishman, 2006). Terrorists, by this logic, view the techniques as furthering their objectives, better than comparable means, and not interfering with other important goals (e.g., ingroup long-term welfare). To discourage the use of terrorism requires undermining its perceived effectiveness, feasibility, and hindrance of other important goals. Easier to say than do, of course, but this analysis takes terrorism out of the realm of “senseless violence” and into a domain where one can begin to understand how to attempt averting it.

Another, compatible perspective describes how people can become terrorists, as a multistep process, like climbing a staircase past multiple floors (Moghaddam, 2005). In the first place is fraternal or group-level relative deprivation, which is the perception that one's ingroup is suffering relative to other groups. Lacking individual mobility or out of strong ingroup identification, many people seek to improve their group's situation. If they perceive the deprivation to be unfair (i.e., attributing it to deliberate, illegitimate injury by others), then they are more likely to be angry and frustrated, the second floor on the staircase. Anger and frustration, as we have seen, can trigger displaced aggression, onto an enemy designated by leaders. At the third level, a minority of frustrated group members may begin to endorse gradually the morality of terrorist organizations as a solution (as in the means-ends analysis of terrorism as a tool). The more justified they see the terrorist strategy, the more receptive they are to terrorist recruiters on the fourth level. Within a terrorist organization, the enemy is viewed as categorically “them,” not at all related to “us.” Finally, an even smaller number of these deprived, frustrated, angry converts get to the last level, where they view civilians as part of the blameworthy enemy and distance themselves from the civilians, so they are willing to hurt others and themselves in what they see as a necessary, even heroic sacrifice for their cause. Prevention efforts would target each of these stages, the earlier the better, because more people inhabit the lower than the upper floors.

Other processes that generate terrorism and mass killing stem partly from the social psychology of intergroup biases, group dynamics, and social influence (so later chapters will touch on topics relevant to mass violence), but large-scale aggression also results from economic, political, and religious dynamics beyond the scope of this book (see Jackman, 2002; Newman & Erber, 2002; Staub, 1989). Terrorists always believe in the virtue of their cause, whether religious beliefs (Bushman, Ridge, Das, Key, & Busath, 2007) or secular ones prime and allegedly justify their aggression.

Summary of Control Theories

Felt lack of control begets violence. Frustration, paired with other aggressive cues, often leads to aggression, either direct or displaced. Mediators include blaming attributions and angry affect. A similar effort to assert control may underlie bullying. Bullying of peers relates to defensive self-esteem and settings that permit bullying. Again implicating control, abuse of intimates relates to unstable, insecure attachment and sense of self. Intimate abuse also relates to antisocial understanding, given the roles of blaming attributions and parental modeling of abuse. Familiar social psychological processes explain even terrorists' psychology of aggression.

PROTECTING ONE'S IMAGE: SELF-ENHANCEMENT

Conventional wisdom holds that people are aggressive because their low self-esteem limits their ability to express their needs in more constructive ways. However, social psychology suggests just the opposite. Two lines of social-psychological work on aggression illustrate how high self-esteem—but of a defensive, inflated, fragile sort—can lead to aggression. This suggests a role for the core social motive of self-enhancement in aggression.

Narcissistic Rage

Threatened egoism underlies a variety of aggression, according to a review of work on aggression, crime, and violence (Baumeister, Smart, & Boden, 1996). Some of the most aggressive people do not show low self-esteem. They are, instead: not depressed, but rather maniacally grandiose; psychopaths, who are narcissistic; men, who on average have higher self-esteem than women; or intoxicated, which inflates self-esteem. Insecure arrogance describes the narcissism of people most likely to aggress (Baumeister, 1997). As we saw among bullies, aggressors often have unstable self-esteem, and domestic abusers often have an insecure sense of self. High but unstable self-esteem remains vulnerable to threat and predicts both anger and hostility (Kernis, Grannermann, & Barclay, 1989).

For example, in one study (Bushman & Baumeister, 1998), laboratory aggression was irrelevant to conventional self-esteem (measured by items such as “I feel that I have a number of good qualities”; “I take a positive attitude toward myself”; and “I am able to do things as well as most people”; see Chapter 5 ). However, researchers also measured narcissism (that is, excessive self-love, measured as agreeing with items such as “If I ruled the world, it would be a much better place”; “I am going to be a great person”; and “I am more capable than other people”). The combination of high narcissism and insult predicted high aggression toward the source of the insult (see Figure 10.4 ). People who view themselves more favorably than other people view them, having unrealistically high self-esteem, are likely to receive frequently disappointing feedback as a matter of course.

Figure 10.4 Aggression as a Function of Narcissism and Threat

Narcissism becomes narcissistic rage only when insult and shame are added. Shame involves a public threat to the entire self. Individual differences in shame proneness clearly predict maladaptive responses to anger (Tangney, Wagner, Hill-Barlow, Marschall, & Gramzow, 1996), as do situations that invoke shame (Tangney, 1992). Bullies and batterers also emerge from experiences of shame. Shame is negatively related to empathy; that is, the more focused one is on one's own shameful exposure, the less receptive one is to other people's experiences.

In a similar vein, sensitivity to rejection—a disposition to defensively expect, readily perceive, and intensely react to social rejection—often predicts aggression (Ayduk et al., 2000; Downey, Feldman, & Ayduk, 2000; Downey, Lebolt, Rincon, & Freitas, 1998). Consistent with this idea, sexual aggressors discount and are suspicious of the veridicality of women's communications (Malamuth & Brown, 1994), as if they are vigilant to being shamed and humiliated by women's rejection.

Sexual aggression seems to be a particular case in point. Men who are self-centered, rather than sensitive to the needs of others, are more likely to act on their aggressive sexual fantasies (Dean & Malamuth, 1997). Hostile masculinity entails being insecure, defensive, hypersensitive, hostile, and distrustful, as well as gratified from dominating women. It predicts sexual aggression and conflictual relationships with women, even as much as ten years later (Malamuth, 1986; Malamuth, Linz, Heavey, Barnes, & Acker, 1995). Hostile masculinity comes from attitudes supporting violence against women, but more relevant here, also from masculine role stress (feeling stressed about failing at the traditional male role by being unemployed, subordinated to a woman, intellectually inferior, shorter than a woman, sexually inadequate, and so on). Similarly, hostile sexism (which includes the beliefs that some women are trying to control men; Glick & Fiske, 1996; see Chapter 11 ) predicts likelihood of sexual harassment (Begany & Milburn, 2002). Thus, aggression can result from particular insults to some men's masculine self-esteem.

Men with unrealistically inflated self-esteem may try to dominate others in an effort to confirm their fragile sense of superiority. Sexual aggressors link dominance to sexuality, as demonstrated by a variety of evidence, some noted earlier. For example, some men's needs for power and dominance, along with aggressiveness and anger, predict attitudes accepting of rape (Anderson, Cooper, & Okamura, 1997). At the automatic, unconscious level, power and dominance prime sexuality for men who are likely to sexually harass or aggress (Bargh et al., 1995; Pryor & Stoller, 1994). Certain people and certain situations facilitate sexual harassment; again, a person-situation combination predicts better than either variable alone (Pryor, Giedd, & Williams, 1995). Men with a need to dominate women, coupled with harassment-accepting norms, may bolster their fragile self-esteem by sexual harassment and aggression. These patterns suggest general proclivities consistent with narcissism.

The pattern of narcissistic rage, shamed narcissism erupting in aggression, does not have to be always a matter of individual differences. Situational threats to the continuity of the self have similar effects on most people.

Interpersonal rejection can make almost anybody angry and aggressive. Several individual differences moderate the relationship between perceived rejection and aggression (Leary, Twenge, & Quinlivan, 2006). And several mechanisms may mediate this relationship; rejection may be experienced as pain, frustration, or self-esteem threat, and aggression may improve one's mood, influence others, reestablish control, or inflict revenge. Or people who feel rejected may simply feel relieved of the necessity to attend to social norms and control themselves, allowing expression of aggressive impulses. The rejection-aggression relationship would seem overdetermined.

Some degree of rejection is inevitable, so is aggression also inevitable, beyond individual differences that dampen potential aggression? Even dispositional narcissists can reduce their normally aggressive responses to criticism if they perceive a prior unit relationship (i.e., a bond) between themselves and the other person (Konrath, Bushman, & Campbell, 2006). So perhaps anyone who is feeling ego-threat and therefore prone to aggression, can be soothed by inclusion and social bonds.

Consider also, as a situational threat, mortality salience—reminders that, like all living things, one ultimately will die—can provoke aggression (McGregor et al., 1998). Mortality salience (e.g., manipulated by writing about one's own death or a control topic) makes people cling to values (e.g., ethical, religious, political) that will endure beyond their own life span. Given the salience of their own mortality, then confronted with a person who disparages their worldviews (e.g., political beliefs), participants responded aggressively (assigned the person to consume more hot sauce). Derogation (verbal aggression) operates in the same way. One could view mortality salience as a threat to ordinary people's narcissism, in the sense of our unspoken belief that we will live forever. Wounded narcissism again lashes out.

Culture of Honor

Moving up a level of analysis from individuals and situations to cultures, one might imagine an entire culture that is shame avoidant. The culture of honorhypothesis holds that, because of local norms, southern males more often value their reputation for toughness and retaliation to insult. They more frequently endorse violence for the protection of self and possessions and respond with greater anger to insults (Nisbett, 1993). The United States South, and portions of the West originally settled by southerners, have higher homicide rates than the North and East. These values and the correlated aggression might be linked historically to herding cultures, in which one's wealth can be easily stolen, and a reputation for aggressiveness may shield one's investments. In field experiments, southern and western employers were more forgiving of job applicants with a history of killing someone in an honor-related conflict, and southern and western newspapers crafted more sympathetic stories about a stabbing in response to a family insult (Cohen & Nisbett, 1997). Note that these results are specific to honor-related violence, for example, arguments, not to all forms of violence. In laboratory experiments, southern men insulted by a confederate more often thought their masculine reputation threatened, felt upset, were primed for aggression cognitively and physiologically, and behaved aggressively and dominantly (Cohen, Nisbett, Bowdle, & Schwarz, 1996).

Consistent with the cultural transmission idea, greater social organization is associated in the South and West with greater violence on several levels: argument-related homicides; violence in choices of media, recreation, and vocation; and votes on gun control and national defense (Cohen, 1998). Similarly, southern and western states' laws reflect greater acceptance of violence as a form of self-protection: This appears in laws related to gun control, defense of self and family, and foreign policy (Cohen, 1996). In southern former slave states, some laws also reflect greater acceptance of violence for coercion and punishment: This appears in laws related to spouse abuse, corporal punishment, capital punishment, and foreign policy.

In interpersonal interactions, the stuff of social psychology, southerners are more polite and less sensitive to minor cues of hostility. Northerners were more likely to give and receive small doses of hostility (Cohen, Vandello, Puente, & Rantilla, 1999). Southerners were more likely to ignore annoyance until it accumulated but then to react more severely. Argument-related violence in northern and southern cities reflects similar patterns.

Summary of Self-Enhancement Theories

Narcissistic rage appears in individuals with inflated but fragile self-esteem when they are threatened. Research on narcissism, shame proneness, rejection sensitivity, hostile masculinity, male dominance, and mortality salience all illustrate aggression in response to ego threats. On a cultural level, male northerners may exchange hostilities at lower levels, being less polite, but male southerners defend their honor when insults reach a serious level, endorsing aggression as a form of self and family protection.

CHAPTER SUMMARY

We opened the chapter with one example of overt interpersonal physical aggression (a school shooting) and one example of subtle, relational, verbal aggression (revenge by power and control over an entire family for generations). Aggression is any behavior whose proximate intent is harm to another person. Proximate (immediate) intent may differ from the ultimate (primary) intent, as in instrumental aggression, which may be a means to an end. Although all aggression may contain elements of coercive influence, that is not a necessary component, as when someone reacts mainly out of anger. Proactive aggression takes the initiative, and reactive aggression responds to perceived threat. Aggression differs along several dimensions: physical-verbal, active-passive, and direct-indirect.

The most common operational definitions in the laboratory include physical responses (aggression machine, reaction-time competition with noise blasts or shocks, doses of hot sauce) and verbal responses, which show parallel results. Field studies include archival crime data, direct observation in schools, and public behavior such as horn-honking. Results from laboratory and field studies clearly converge.

Chronic social issues around cultural artifacts—media violence, guns, and alcohol—continue to raise controversy. In all three cases, according to decades of social-psychological research, unrestricted access enhances aggression, when people are provoked. Media violence operates via various psychological processes, especially social learning. The presence of weapons operates by cuing aggressive associations. And alcohol myopia operates by making people less able to consider conflicting impulses not to aggress.

Core social motives prominent in aggression research include antisocial understanding, controlling resources via inherited mods, and self-enhancing for those with high but unstable self-esteem. In service of understanding, social learning teaches how and when to aggress, via modeling and vicarious conditioning. Its operation is particularly clear in imitative aggression (how), but people also learn self-sanctions (when). Observational learning suggests that catharsis does not result from observing aggression. Social learning posits cognitive representations that link observation of behavior and its imitation.

More recent accounts of cognitive representations include scripts, associations, and episodes. The script theory addresses how rehearsal of aggression facilitates stable patterns of aggression in individuals and family members. Scripts are accompanied by normative beliefs about appropriate behavior. Cognitive neo-association theory describes people's immediate associations to aggressive cues, including rudimentary emotions related to fight or flight, as well as more complex cognitive and emotional responses. The general aggression model relies on the broad idea of knowledge structures to integrate several previous theories. To these, it adds automatic appraisal and more controlled reappraisals. Further, cognitive theories focus on the what and why of aggression.

Attributional approaches include the hostile attribution bias, in which aggressors ignore benign cues in ambiguous settings and infer malignant intent. Excitation transfer focuses on attributions for the causes of arousal, including the misattribution from prior irrelevant sources to aggressive situations. This again undermines the notion of catharsis. Attribution tells people why to aggress.

Interpersonal conflicts focus on controlling others. Frustration often results in anger and displaced aggression and sometimes also aggression in the service of removing an obstacle. In a sense, it is an effort to control someone stronger. Efforts to control people in a weaker position include bullying and partner abuse. Both of these hinge on a combination of people prone to aggression and situations that permit it.

Aggression also results from efforts to shore up a weak or unstable sense of high self-esteem, resulting in narcissistic rage. The culture of honor's link to aggression may be seen as a cultural analog to narcissism. In the next chapter, we take up the cultural phenomenon of intergroup relations, which can result in aggression when people interact as representatives of conflicting groups.

SUGGESTIONS FOR FURTHER READING

1. Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27–51.

2. Aquino, K., & Thau, S. (2009). Workplace victimization: Aggression from the target's perspective. Annual Review of Psychology, 60, 717–741.

3. Barling, J., Dupré, K. E., & Kelloway, E. K. (2009). Predicting workplace aggression and violence. Annual Review of Psychology, 60, 671–692.

4. Bushman, B. J., & Huesmann, L. R. (2010). Aggression. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 833–863). New York: Wiley.

5. De Dreu, C. K. W. (2010). Social conflict: The emergence and consequences of struggle and negotiation. In S. T. Fiske, D. T. Gilbert, & G Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 983–1023). New York: Wiley.

6. Fiske, S. T. (2010). Interpersonal stratification: Status, power, and subordination. In S. T. Fiske, D. T. Gilbert, & G Lindzey (Eds.), Handbook of social psychology (5th ed., pp. 941–982). New York: Wiley.

7. Mummendey, A. (1996). Aggressive behavior. In M. Hewstone, W. Stroebe & G. M. Stephenson (Eds.), Introduction to social psychology (2nd ed., pp. 403–435). Malden, MA: Blackwell.

8. Pyszczynski, T., Greenberg, J., Koole, S., & Solomon, S. (2010). Experimental existential psychology: Coping with the facts of life. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., pp. 724–757). New York: Wiley.

Notes