Understanding Clinical Anger and Violence The Anger Avoidance Model Frank L. Gardner La Salle University Zella E. Moore Manhattan College

Although anger is a primary emotion and holds clear functional necessities, the presence of anger and its behavioral manifestations of aggression/vio- lence can have serious emotional, health, and social consequences. Despite such consequences, the construct of clinical anger has to date suffered from few theoretical and treatment advancements and has received insufficient research attention. Thus, the purpose of this article is to introduce the Anger Avoidance Model, which is a new conceptualization of clinical anger and its behavioral manifestations. The Anger Avoidance Model suggests that among anger patients, a chronic early aversive history leads to information process- ing biases and emotion regulation deficits, which in turn result in intense efforts to avoid the experience of anger. This avoidance takes the form of hos- tile rumination (cognitive avoidance) and aggressive and violent behavior (behavioral avoidance). This model holds clear implications for research and treatment of this challenging clinical phenomenon.

Keywords: anger; violence; avoidance; aggression; emotion regulation

The emotion of anger is a natural, biologically necessary primary emo- tion innate to all human beings. The subjective emotional experience

of anger may vary from mild (irritation) to severe (rage) and is character- ized by both cognitive biases reflecting exaggerated prediction, interpreta- tion, and oversensitivity to violation by others and physiological arousal, which is reflected by sympathetic nervous system arousal and increased muscular tension (Deffenbacher, Demm, & Brandon, 1986). Anger may also function as a secondary emotion by serving as an affective response to other emotional states. By way of example, fear, which may be experienced

Behavior Modification Volume 32 Number 6

November 2008 897-912 © 2008 Sage Publications

10.1177/0145445508319282 http://bmo.sagepub.com

hosted at http://online.sagepub.com

897

Authors’ Note: Correspondence concerning this article should be addressed to Frank L. Gardner, 1900 West Olney Avenue, Box 842, Philadelphia, PA 19141; e-mail: [email protected]

as intolerable, is often associated with a strong feeling of uncontrollabil- ity/vulnerability and can in turn become a discriminative stimulus for the secondary emotion of anger.

Although, to many, anger may seem like a maladaptive response to inter- personal distress, the emotion is actually intended to serve an adaptive function, as it has as its most basic purpose the preparation of human beings to respond to real threats in the environment (Kemper, 1987). However, when generalized to contexts beyond those in which it is likely to be useful and adaptive, this otherwise normal emotion can lead to chronically height- ened arousal and is associated with dysfunctional and problematic behav- ior. For many individuals, heightened intensity, frequency, and duration of anger, which we have defined as “clinical anger,” are precursors to a vari- ety of interpersonal, health, occupational, and legal difficulties (Del Vecchio & O’Leary, 2004; Kassinove & Sukhodolsky, 1995).

Yet although clinical anger and its behavioral manifestations often result in significant intrapersonal and interpersonal consequences, there are cur- rently no criteria for the diagnosis of an anger disorder (Diagnostic and Statistical Manual of Mental Disorders, 4th ed., text revision; DSM-IV-TR; American Psychiatric Association [APA], 2000). In the context of the cur- rent psychiatric nosology, anger serves only as a contributing characteristic for numerous mental disorders (APA, 2000), such as posttraumatic stress disorder (PTSD), mood disorders, impulse control disorders, and borderline personality disorder. Due in part to the absence of diagnostic criteria of its own, to date, there have been no epidemiological investigations of prob- lematic anger (DiGiuseppe & Tafrate, 2003). However, there is a large empirical base to support its serious clinical relevance. In recent investiga- tions, 36% of patients meeting criteria for major depressive disorder, 61% of patients meeting criteria for Bipolar II disorder (Benazzi, 2003), and 48% of patients meeting criteria for PTSD (Murphy et al., 2004) reported substantial difficulties with anger. Frueh, Henning, Pellegrin, and Chobot (1998) found that measures of anger were significantly elevated in patients experiencing combat-related PTSD and were highly related to occupational impairment even when controlling for PTSD severity. In addition, Erwin, Heimberg, Schneider, and Liebowitz (2003) found that individuals with social anxiety disorder experienced greater elevations of anger and expressed their anger in more problematic ways than did nonanxious con- trols. This same study further suggested that high levels of anger predicted premature termination from treatment and a generally less satisfactory response to an empirically supported treatment for social anxiety disorder. Similarly, exposure therapy was shown to be less effective for individuals

898 Behavior Modification

with high levels of pretreatment anger in patients meeting criteria for PTSD (Foa, Riggs, Massie, & Yarczower, 1995). In yet another study, Fava and Rosenbaum (1998) found that approximately one third of depressed outpa- tients presented with anger attacks, which were defined as the sudden and intense experiencing of anger, accompanied by symptoms indicative of the activation of autonomic arousal (such as rapid heart rate, sweating, hot flashes, and chest tightness).

Previous Theoretical Considerations of Anger and Aggression

Although there are clear comorbidities between anger and other disor- ders, cognitive-behavioral conceptual models of anger have actually been sparse. Thus, prior to a discussion of the development of our Anger Avoidance Model, a description of the theoretical model that has dominated to date is warranted.

To date, the most commonly discussed model for explaining anger dyscontrol and its consequences is the cognitive content specificity model. This model, in its various forms, postulates that specific cognitions, most likely related to unrealistic demands, expectations, or assumptions about the behavior of others, result in the emotional experience of anger and are in turn associated with physiological changes (Kassinove & Tafrate, 2002). To describe the relationship between anger and aggressive/violent behavior, the model suggests that heightened anger results in socially constructed overt behavioral responses in the form of aggression/violence and these behavioral responses essentially function to discharge the anger. In turn, this process is typically reinforced by immediate interpersonal outcomes (i.e., getting what one wants). Although this position is central to the cog- nitive content specificity model of anger and aggression, recent research strongly suggests that aggressive behavior does not in fact reduce the level of experienced anger. In fact, to the contrary, empirical findings suggest that venting aggressive behavior (i.e., hitting a punching bag) actually leads to an increase in the experience of anger (Bushman, 2002).

In essence, the cognitive content specificity model of anger and aggres- sion views (distorted) cognitive content as the central feature in anger, as it is suggested to be in other emotional states (Beck, Brown, Steer, Eidelson, & Riskind, 1987; Kassinove & Tafrate, 2002). This position fundamentally places the misappraisal of environmental events as central to the under- standing of and ultimately the treatment for problematic anger and views

Gardner, Moore / Clinical Anger 899

behavioral excesses or deficits such as aggressive/violent behavior as a direct result of the emotion of anger. From this perspective, anger has been defined as “a felt state that is negative and, thus, generally to be avoided” (Kassinove & Tafrate, 2006, p. 5).

Although the cognitive content specificity model has been the primary model for understanding high levels of anger and its associated conse- quences, the model appears to have several flaws: (a) Because anger is a natural human emotion, it is problematic to view anger as something to be avoided; (2) the anger treatments that have been based on this model have demonstrated questionable efficacy, thus calling into question the theoreti- cal model from which they have been developed (Santanello, Gardner, & Moore, 2008); and (c) the model has not integrated more contemporary research on emotion and emotional disorders.

The Anger Avoidance Model: Integrating New Developments

Given the previously noted presence of anger within other clinical pop- ulations, the paucity of consideration of anger in the scientific literature, and the difficulties with the cognitive content specificity model, we have developed a new model for understanding clinical anger and its behavioral manifestations. This model, which we have termed the Anger Avoidance Model, draws on several key developments in the study of other emotional disorders (see Figure 1 for a schematic representation of the entire Anger Avoidance Model).

Anger and the Emotional Disorders

In recent years, as greater emphasis has been placed on developing a more comprehensive understanding of the processes involved in emotional disorders, findings in experimental psychopathology have highlighted the significant overlap between anxiety and mood disorders (Barlow, Allen, & Choate, 2004; Brown, Campbell, Lehman, Grisham, & Mancill, 2001; Mineka, Watson, & Clark, 1998). In fact, it has been suggested that emo- tional disorders share foundational psychosocial and biological diatheses and the expression of particular symptom clusters are in fact minor varia- tions of a broader syndrome. This has been referred to as negative affect syndrome (NAS; Barlow et al., 2004). Using structural equation modeling as a base methodological approach, Barlow et al. (2004) developed a model

900 Behavior Modification

for understanding anxiety and mood disorders (referred to as emotional dis- orders) that confirms earlier descriptions of a tripartite model of emotional disorders described by Clark and Watson (1991).

Gardner, Moore / Clinical Anger 901

Aversive History (Maltreatment, Neglect)

+ Biological Vulnerability

Hostile Anticipation (External Scanning, Attentional Biases,

Attributions of Personal Violation)

Specific Life Event

Impaired Emotional Processing & Emotion Dysregulation

Internalized Avoidance (Hostile Rumination)

Externalized Avoidance (Aggressive Behavior)

Experience of Anger (Including Heightened Physiological Arousal)

Figure 1 Model of Clinical Anger

In the tripartite model, emotional disorders are distinguished by varying combinations of negative affect, positive affect, and autonomic arousal. When viewing emotional disorders from this conceptualization, the com- monalities across the varying DSM-defined emotional disorders are more significant than the variations. For example, negative affect and autonomic (somatic) arousal are highly (positively) correlated with panic disorder, which in turn is not correlated with positive affect. Both depression and social anxiety, on the other hand, are highly correlated with negative affect (positive correlation) and positive affect (negative correlation) yet are min- imally correlated with autonomic arousal. With this in mind and given the numerous empirical findings regarding the comorbidity of anger with a variety of current diagnostic categories, rather than suggest the need for a new diagnostic category for “anger disorders” (DiGiuseppe & Tafrate, 2003; Kassinove & Sukhodolsky, 1995), clinical anger may best be viewed as a specific variation of NAS. In fact, Mineka et al. (1998) suggested that “it now is obvious that this general Negative Affect dimension is not con- fined solely to mood and anxiety disorders, but is even more broadly related to psychopathology” (pp. 397-398). From this perspective, we suggest that anger is positively correlated with negative affect, is negatively correlated with positive affect, and is positively correlated with somatic arousal. In essence, this would place anger as a mixed emotional disorder manifesting the heightened distress found in both anxiety and mood disorders, the heightened autonomic arousal most often seen in panic disorder and gener- alized anxiety disorder, and the absence of positive feelings such as joy and happiness (anhedonia) most typical of depression and social anxiety disor- der. This is graphically presented in Table 1.

Etiological Considerations

The triple vulnerabilities model proposed by Barlow et al. (Barlow, 2002; Barlow et al., 2004) has been helpful in describing the development of other emotional disorders, and similarly, we suggest that it is useful for conceptualizing the development of clinical anger and its behavioral mani- festations. In the triple vulnerabilities model, three important diatheses interact in a synergistic manner to result in the ultimate development and presentation of an emotional disorder.

The first diathesis is biological and reflects the myriad of research demonstrating the biological and genetic contribution to basic tempera- ments (Barlow, 2002). Specifically relevant to the conceptualization pre- sented herein, Barlow (2002) has suggested that individuals with high

902 Behavior Modification

levels of anger and those with panic disorder appear to be similar in their biological vulnerabilities. In fact, recent research offers some preliminary support for the proposition that individuals experiencing anger and/or anx- iety do in fact experience similar levels of physiological arousal (Wenzel & Lystad, 2005). Yet although an individual’s biological vulnerability func- tions as a type of inherent risk factor for the development of an emotional disorder, a biological vulnerability does not appear to predict the later development of specific emotional disorders. Rather, it sets the stage for the ultimate manifestation of these disorders by resulting in a general overre- sponsiveness to stress and challenge.

The second vulnerability has been described as the general psychologi- cal diathesis and reflects the impact of specific life experiences on the indi- vidual. It is these experiences that have been postulated to contribute to the development of anxiety, depression, and related affective states by produc- ing a general sense of uncontrollability (Barlow, 2002). Yet with respect to anger, Barlow (2002) has suggested that patients experiencing clinically relevant anger differ from patients experiencing anxiety in that rather than the perception of uncontrollability experienced by anxious patients, angry patients may experience an exaggerated sense of mastery and control. However, our observations from working with clinical anger patients in our anger specialty clinic do not support the proposition that angry individuals experience an exaggerated sense of mastery and control. Rather, we suggest that angry patients actually exhibit as great a sense of uncontrollability/ vulnerability as anxious patients but choose a fight (as opposed to flight) response based on both biological factors noted earlier and specific early learning histories and modeling experiences. It is important to note that in our conceptualization of clinical anger (Anger Avoidance Model), percep- tions of uncontrollability are as problematic for the clinical anger patient as they are for the anxious patient. However, as angry and anxious patients typically encounter very different early learning histories, they experience different affective and behavioral responses to their perceptions of uncon- trollability. Erwin et al.’s (2003) findings suggesting a strong relationship

Gardner, Moore / Clinical Anger 903

Table 1 Anger and Negative Affect Syndrome

Emotional Disorder Positive Affect Negative Affect Arousal

Anxiety Moderate to high Moderate to high Moderate to high Depression Low Moderate to high Low Anger Low Moderate to high Moderate to high

between social anxiety and trait anger provide some indirect, albeit tenta- tive, support for the idea that anger does not reflect an exaggerated sense of mastery or control but instead reflects a different response to an extreme sense of uncontrollability and vulnerability. We suggest that the early aver- sive histories noted among angry individuals culminate in feelings of vul- nerability, which in turn become stimuli for the experience of anger.

It has already been demonstrated that the biologically based tempera- ments of anxious and angry patients share some clear similarities (Wenzel & Lystad, 2005). Yet although some similarities between anxious and angry patients do of course exist, there appear to be several significant ways that the learning histories of anxious and angry patients differ, thus resulting in different general psychological vulnerabilities. Patients with anxiety disor- ders frequently describe parents/caretakers as less supportive, less socially engaged, exerting greater control over the decision making of their children, and generally more enmeshed in their children’s lives (Barlow, 2002). This does not appear to be the case with patients experiencing clin- ical anger. In fact, recent findings from our lab suggest that the early expe- riences of angry patients are actually more similar to those found in chronically depressed patients (meeting criteria for dysthymic disorder) as described by McCullough (2000). This early environment is often chroni- cally aversive and is characterized by (a) a harsh and punitive environment (often including a history of neglect or mistreatment), (b) being subjected to a hazardous social and/or familial environment where caretakers and/or family members cannot protect or even hurt their children as well as each other, and (c) the experience of chronic physical and/or emotional pain. Together, these variables culminate in an intense emotional environment in which physical and psychological safety and survival become most impor- tant. We suggest that individuals who confront such early histories experi- ence intense emotion along with a pervasive sense of powerlessness, which culminates in a level of anger/rage that is subjectively overwhelming. In fact, we have found that in both clinical (n = 60) and nonclinical (n = 250) samples, early aversive history does in fact predict levels of trait anger, as measured by the State Trait Anger Expression Inventory (STAXI; Spielberger, 1988). In turn, individuals exposed to a chronic early aversive history manifest significantly higher levels of trait anger and anger reactiv- ity than individuals who have not been exposed to such a harsh early history (Gardner, Moore, Wolanin, Alm et al., 2006; Gardner, Moore, Wolanin, Deutsch, & Marks, 2006).

One question may be suggested by our recent empirical finding con- firming the etiological similarities between patients experiencing clinical

904 Behavior Modification

anger and those experiencing chronic depression (Gardner, Moore, Wolanin, Alm et al., 2006). That is, because the early learning histories of both clinical anger patients and chronically depressed patients are largely characterized by chronically harsh and punitive early environments, why do clinical anger patients not then develop chronic depressive disorders instead? Referring back to the triple vulnerabilities model, in response to the simi- lar early environments, we hypothesize that patients who are more biolog- ically prone to heightened levels of autonomic arousal are likely to respond with clinical anger, whereas those with a different biologically based tem- perament are likely to respond with lower levels of arousal and as such demonstrate overt and covert manifestations consistent with a chronic depressive disorder. Although this hypothesis is an open empirical question, it is consistent with our conceptualization of anger as sharing tripartite model characteristics with both depression and anxiety.

Information Processing Biases

The third synergistic vulnerability for emotional disorders has been labeled the specific psychological diathesis (Barlow, 2002), which is an integration of the individual’s biological temperament and his or her early social learning history from the first two diatheses. In the case of dysfunc- tional anger, the specific psychological diathesis often displays as cognitive (information processing) biases for violation-relevant stimuli. We suggest that in response to the highly aversive early social-familial environments that are common for patients with dysfunctional anger, a functional survival- based tendency evolves in which the individual with clinical anger scans the external environment for early signs of potential personal (physical or psychological) violation. We have created the term hostile anticipation to represent the angry patient’s perseverative hypervigilance for signs of hos- tile intent and personal violation. We view the hypervigilance for signs of external danger in angry patients as analogous to the hypervigilance for signs of internal danger often noted in anxious patients.

Yet although anxious patients often present with excessive self-focused attention and see physical sensations, social evaluation, and unacceptable thoughts as a sign of potential danger, Barlow (2002) has suggested that rather than the excessive self-focused attention most often seen in anxious patients, the angry patient is likely to manifest a nearly exclusive external focus of attention. Consistent with this hypothesis, a number of recent stud- ies have demonstrated specific information processing biases in angry indi- viduals. One recent study found that those individuals in a college student

Gardner, Moore / Clinical Anger 905

sample scoring high on trait anger did in fact make more global and exter- nal attributions than anxious or nonangry/anxious controls (Langton & Wenzel, 2004). Similarly, Aquino, Martinko, and Douglas (2004) found that hostile attributions mediated the relationship between external events and reported anger in an organizational context. In yet another study, Cohen, Eckhardt, and Schagat (1998) found that individuals high in trait anger demonstrated attentional biases toward anger-related cues following insult. Finally, Wenzel and Lystad (2005) found that angry subjects rated the likelihood of angry explanations for ambiguous events higher than anx- ious or nonanxious/nonangry controls. Taken together, these studies do in fact suggest that specific information processing biases exist among patients experiencing clinical levels of anger. The consequence of these information processing biases is that an inordinate number of otherwise innocuous life events will instead be interpreted as threatening or danger- ous, thus triggering the emotion of anger.

Emotion Regulation and Avoidance

Based on our ongoing experience with clinical anger patients and recent laboratory findings, it appears that patients manifesting clinical anger also manifest significant difficulties with emotion processing, which has been defined as “the way in which an individual processes stressful life events” (Baker, Holloway, Thomas, Thomas, & Owens, 2004, p. 1272; see also Rachman, 2001). When such events are not fully integrated and processed, a chronic reliance on avoidance as a coping strategy is likely to occur. We suggest that when these individuals experience levels of affect that are per- ceived to be intolerable (i.e., anger), the avoidance typically takes the form of overt aggressive or violent behavior, or covert cognitive processes such as hostile rumination. Studies from our laboratory, recently presented else- where, indicate that difficulties in emotion regulation do in fact mediate the previously noted relationship between early aversive history and trait anger and anger reactivity (Moore, Gardner, & Wolanin, 2006).

Behavioral avoidance. Recent studies suggesting that aggressive behav- ior serves an affect regulation function (Bushman, Baumeister, & Phillips, 2001) and interpersonal violence in particular may function as a means of regulating emotion (Jakupcak, Lisak, & Roemer, 2002), are consistent with, and can be explained by our laboratory finding that emotion regulation dif- ficulties mediate the relationship between early aversive history and trait anger and anger reactivity (Moore et al., 2006). Thus, in the Anger

906 Behavior Modification

Avoidance Model proposed herein, ineffectively processed anger becomes a cue for cognitive avoidance in the form of hostile rumination and/or behavioral avoidance in the form of overt aggressive behavior (verbal or physical). Seen in this light, aggressive/violent behavior is conceptualized as an overt avoidance or escape response and functions to reduce the full experience of anger (and the fear that may have been the initial stimulus for the anger). Consequently, it is negatively reinforced. To expand, overt aggressive/violent behavior serves as an escape from the stimulus that led to the anger response by either eliminating the stimulus itself (i.e., the indi- vidual who has “wronged” the angry client withdraws from the situation) or by changing the form of the stimulus (i.e., the individual acts differently toward the angry client).

Anecdotal clinical findings also suggest that aggressive/violent behavior may serve an avoidant function in many clients who seek psychological treatment. Clinicians working with violent offenders frequently note that patients with clinical anger often report “feeling nothing” both during and immediately after an aggressive outburst (DiGiuseppe, Fuller, & Fountain, 2006; Gardner, Moore, Ronkowski, & Wolanin, 2006) and tend to score extremely high on the anger suppression subscale of the STAXI (Gardner, Moore, Wolanin, Alm et al., 2006). According to the Anger Avoidance Model, these psychometric findings would appear to reflect the attenuation (avoidance) of the emotion of anger and not simply a motivationally based reluctance to admit to experiencing anger or engaging in aggression.

Although the avoidant function of aggressive/violent behavior seen in many clients may be confused with instrumental aggressive/violent behav- ior, it is important to note that the theoretical position presented herein does not suggest that all acts of violence come from an effort to avoid the expe- rience of anger. Although the form of the behavior is the same in both groups, the functions of the behaviors differ. In one group, the function of the behavior is the reduction or elimination of affect, whereas in the other group, the function of the behavior is appetitive and seeks to achieve an end by way of control or manipulation. In fact, we suggest that there are actu- ally two subgroups within the population of individuals manifesting aggres- sive/violent behavior. The first subgroup includes individuals who use aggressive behavior in an instrumental fashion as a means of achieving con- trol over others or as a means of making others, feel, think, or act differ- ently. These individuals report no clear pattern of chronic early aversive history, demonstrate no significant difficulty in emotion regulation, and experience little (if any) discernable anger. We have termed this group

Gardner, Moore / Clinical Anger 907

instrumental aggressors. In contrast, the second subgroup includes individu- als whose aggressive/violent behavior functions as a form of overt avoidance/ escape from the experience of anger. This subtype is defined by an early aversive history, extensive experiential avoidance, and difficulties in emo- tion regulation, and it displays significantly elevated scores on measures of trait anger and anger reactivity. It is this second subgroup, which we have termed avoidance/escape aggressors, that appears to use aggressive/violent behavior as a form of avoidance or escape, as conceptualized by the Anger Avoidance Model. Recent studies from our laboratory support this con- tention within a clinical population court mandated following violent offenses (Gardner, Ronkowski, Wolanin, & Moore, 2006).

Cognitive avoidance. As previously stated, ineffectively processed anger becomes a cue for behavioral avoidance in the form of overt aggres- sive behavior and/or cognitive avoidance in the form of hostile rumination. In terms of cognitive avoidance, hostile rumination differs from hostile anticipation (introduced earlier) in that hostile rumination functions as a manifestation of experiential avoidance by repetitive and recurrent think- ing about perceived historical and/or present violations and consequences, whereas hostile anticipation is a future-oriented cognitive appraisal of both the likelihood and consequences of personal violation. With this distinc- tion in mind, the avoidant function of hostile rumination would be expected to follow the same pattern seen in worry (Borkovec, 1994). The negatively reinforced (i.e., avoidant) function of rumination has been shown in numerous studies demonstrating reduced emotional and physio- logical arousal and poorer emotional processing during rumination (Clark & Collins, 1993; Pennebaker, 1997; Teasdale, 1999). As such, hostile rumination would appear to function in a manner similar to worry in patients with generalized anxiety disorder by attenuating the full experi- ence of emotion. This occurs because rumination is typically past or future oriented, and thus, the individual does not have an in-the-moment experi- ence of anger. The emotion-attenuating function of ruminative processes is thus negatively reinforced and results in an increase in frequency (Borkovec, Ray, & Stober, 1998). Recent findings support the suggestion that rumination is a maladaptive effort to cope with affect and is associated with cognitive biases, sustained negative mood states, and impaired prob- lem solving (Joorman, Dkane, & Gotli