Male Perpetrators of Violence Against Women:
An Attachment Theory Perspective

Peter Fonagy, Ph.D.

This paper proposes an attachment theory formulation of violent acts perpetrated by men against women, usually in the context of sexual relationships. ft is proposed that relationship violence may be seen as an exaggerated response of a disorganized attachment system. It is related to a disorganized attachment pattern in infancy coupled with a history of abuse and an absent male parental figure. The author proposes a theory based on a psychoanalytic understanding of the development of the sell; and highlights similarities between the clinical presentation of male perpetrators and those with borderline personality disorder.

KEY WORDS: domestic violence; aggression; attachment; disorganized attachment.

INTRODUCTION

Taking an historical perspective, violence against women is hardly a problem (Gordon, 1989). As long as society generally has construed women as property and explicitly promoted the social domination and privileges of men, family violence (bar its most extreme forms) has been condoned and legitimized. Yet relationships between intimate adult partners in which women are victims and husbands or boyfriends are perpetrators are now known to have devastating and long-term physical as well as psychological consequences, not only on the women involved but also on child witnesses (Osofsky, 1997). A study conducted in 1991 in a Boston City hospital found that 10% of the children who used outpatient psychiatric services had witnessed a knifing or a shooting by the age of 6 years, and the majority of these incidents occurred in the home (Taylor, Zuckerman, Hank, & Groves, 1994).

How social changes finally led to the criminalization of domestic violence (Pleck, 1987) is beyond the scope of this contribution. My aim here is to provide a model for understanding why, notwithstanding the late twentieth century changes in gender roles, abuse or assault by men of a wife or sexual partner remains relatively commonplace. I will restrict myself here to trying to understand the behavior of the men involved. About 12% of men have a lifetime incidence of serious violent acts toward women—acts such as punching, kicking, beating with a physical object, and rape; two-thirds of these individuals regularly repeat such assaults (Straus, 1979). The gravity of these offenses is indicated by their duration as well as their severity. In her seminal study of 120 battered women, Walker (1979) reported the typical duration of battering to be between 2 and 24 hours.

There are two categories of violent acts against women (Meloy, 1992). By far the most common are the Type I, impulsive attacks carried out as a consequence of minimal provocation but in a state characterized by the perpetrator as uncontrollable and overwhelming rage. A number of large-scale clinical surveys have identified a group of angry, jealous, depressed men frequently violent with their partner (Dutton, 1995a; Saunders, 1992). By contrast, there are premeditated attacks of "predatory" violence against women, Type II violence, where the perpetrator carefully plans the attack, attachment to the victim is based in fantasy, and pleas for temporarily diminished responsibility lack plausibility. I believe that both forms of violence are linked to attachment relationships, actual or imagined, but require somewhat different accounts. 1-lere we shall focus on the first type and do little more than acknowledge that the second form also richly deserves study.

There are several competing accounts of Type I male violence against women. An influential one is rooted in the tacit approval that society gives to abusive males. I quote: "Men who assault their wives are actually living up to cultural prescriptions that are cherished in Western society— aggressiveness, male dominance, and female subordination—and they are using physical force as a means to enforce that dominance" (Dobash & Dobash, 1979, p. 89). Domination of women is to some degree a cultural prescription and violence can serve as a desperate means to that end. However, this does not make a psychological account superfluous or exonerative, as some feminist writers imply (Goldner, Penn, Sheinberg, & Walker, 1990). There are also a number of observations that are inconsistent with a purely sociological feminist account (Dutton, 1995b). For example, why are abuse rates lower in more patriarchal cultures (Sorenson & Telles, 1991)? Why is there no simple relationship between the actual power a man has in a particular relationship and wife assault (Yllo & Straus, 1990)? Why are abuse rates relatively high in both male and female gay couples (Lie, Schilit, Bush, Montague, & Reyes, 1993)? Although I believe sociological factors to be at the root of the problem, such an enterprise is emphatically not exonerative—rather understanding is the preferred path to early prevention. A psychological account is required to address such issues.

Social learning theory accounts maintain that violence against women is a learned response to stress—supported by the immediate rewards (feelings of agency and control, the cathartic expression of anger, or the ending of an argument)—that violence may bring (Dutton, 1995a). Modeling on parental behavior may also be relevant (Bandura, 1986). Interparental violence certainly relates to male assaultiveness. In a 1-year prospective study, men who had been exposed to interparental physical assaults were 3 1/2 times more likely to hit their own wives (35% vs. 10%) (Straus, Gelles, & Steinmetz, 1980). It would be unwise, however, to accept a simple modeling account of this specific association. Observing fathers hitting mothers increases the likelihood of physical aggression on the part of wives as well as husbands, and parents of both genders toward their children (Kalmuss, 1984). Observing interparental aggression appears to prime all individuals to acts of violence. Further, interparental violence is not a complete explanation because many men who commit this type of act have no histories of witnessing violence between their parents (Walker, 1979).

ATTACHMENT THEORY

Our starting point here is the epidemiological observation that the vast majority of Type I violence, where the perpetrator is male and the victim female, involves intimate or attachment relationships (Dutton, 1995a). This most robust and reliable of observations, in a field otherwise littered with inadequately documented accounts, suggests that attachment theory may be helpful in constructing an explanation.

Attachment theory is the black sheep of the psychoanalytic family (Fonagy, in press). John Bowlby’s early writings (Bowlby, 1960) were the subject of severe criticisms from Anna Freud, Rene Spitz, and Max Schur as well as others who considered it mechanistic, nondynamic, and based on thorough misunderstandings of psychoanalytic theory (Freud, 1960; Schur, 1960; Spitz, 1960). Opposition to his views provided one small area of common ground for the followers of Anna Freud and Melanie Klein (Grosskruth, 1986), and for decades Bowlby was a relatively isolated figure in psychoanalysis.

Attachment theory is, however, experiencing something of a psychoanalytic renaissance (e.g., Lichtenberg, 1995). This is perhaps due 10 the increasing ecumenicalism of psychoanalysis and the general acceptance of an objeci relations perspective in the basic psychoanalytic model. According to attachment theory, the primary function of early object relationships is to provide the infant with a sense of security in environments that induce fear (Bowlby, 1973). Bowiby assumed that on the basis of the interactions between the infant and a caregiver, self—other representations develop (he termed these internal working models) that reflect the child’s cumulative experience of sensitivity on the part of that caregiver. This aspect of infant—caregiver relationships is present in most psychoanalytic formulations and clearly overlaps with some current uses of Bion’s containment concept (Bion, 1962), Winnicott’s holding environment (Winnicott, 1965), Kohut’s self-objects (Kohut, 1971), and Sandler’s concept of safety (Sandler, 1960). The child’s confidence in the caregiver’s capacity to appreciate his distress and to act on this understanding is reflected in the security of the bond between infant and mother (Ainsworth, Blehar, Waters, & Wall, 1978). Mary Ainsworth observed that insensitively parented 1-year-olds tended either to avoid the caregiver after a brief period of separation ("anxious-avoidant") or refuse to be comforted by her on her return ("anxious-resistant"). By contrast, sensitively parented children sought comfort from the caregiver following brief separation. These patterns are reliable across situations. Secure attachment predicts the healthy development of the child in terms of educational success, peer relationships, self-esteem, and identity formation. Bowlby (1973) assumed that the attachment system remains a central organizer of interpersonal behavior throughout the life span. Some infants, who may show any of these three patterns, also show signs of disorganization. On reunion with the caregiver they freeze, collapse, bang their heads, slap the caregiver, hide, or try to escape. These behaviors suggest deep dysfunction in the attachment system.

Bowlby (1969) claimed that the developmental task of the child is to establish a relationship with the caregiver that, through proximity, ensures protection. The parent’s caregiving system provides this proximity and contact to ensure protection and, ultimately, survival. Threat activates the attachment system that, in turn, triggers the caregiving system of the adult. But what of an abusive relationship? Maltreatment (a threat) causes an overactivation of the attachment system, often paradoxically in relation to the abuser who is at once the source of threat and the hope for its containment (Rajecki, Lamb, & Obmascher, 1978). This, in a nutshell, is at the heart of disorganized attachment.

Research using Mary Main’s Adult Attachment Interview (AAI) may have also contributed to the increasing acceptance of attachment theory ideas. The AAI represents an attempt to assess current mental representations by adults of their childhood attachment experiences (Main, Kaplan, & Cassidy, 1985). Several studies, including one by our team in London, have demonstrated that in as many as 80% of cases, infant attachment classification can be predicted on the basis of adult attachment classification made before the birth of the child (Fonagy, Steele, & Steele 1991b; Steele, Steele, & Fonagy, 1996). Parents with an insecure (incoherent) view of their childhood attachment experiences are highly likely to build an anxious attachment relationship with their infants (van IJzendoorn, 1995).

A number of writers working in the context of attachment theory have proposed that hurting one’s partner could be the exaggeration or perversion of attachment behavior (Dutton, 1995b; Dutton, Saunders, Starzomski, & Bartholomew, 1994; Mayseless, 1991; West & George, in press). Broadly speaking, the intensity and force of such abuse could be seen as reactions to attachment insecurity. They parallel behavioral patterns in highly insecurely attached infants and children. Let us explore this suggestion further.

RELATIONSHIP VIOLENCE AS EXAGGERATED
RESPONSE OF ATTACHMENT SYSTEM

Bowlby (1973) recognized that anger is the natural response of the child when the expectation of safety, close to the attachment figure, is jeopardized. In normal development, it has a signaling function of strengthening the relationship between child and caregiver. By the second year, and perhaps even earlier, anger also serves to maintain the integrity of the self-representation. Threats to the toddler’s capacity for self-determination normally lead to ferocious tantrums. Tantrums tend to subside with the emergence of a stable self-organization. The angry outburst of the toddler is not mainly a call for protection, it is also a self-protective response to insensitivity on the part of the caregiver, felt at the moment to have underniined the child’s nascent self-image (Fonagy, Moran, & Thrget, 1993). Normally this reaction provokes the immediate intensification of care-taking responses with a greater chance that the child’s intentionality is recognized. Toddler temper tantrums are common only in the attachment context.

The normal anger response, however, turns to aggression when insensitivity is pervasive. The defensive shield of anger is called for so frequently that the oppositional response becomes integrated with the child’s selfstructure. Self-assertion immediately yields aggression. Although anger has an important function within the attachment relationship, aggression is clearly dysfunctional because it threatens to break apart the attachment bond. In Bowiby’s formulation, aggression—or rather dysfunctional anger— lies at the root of anxious attachment.

Anxious attachment may thus be linked to violence but cannot provide sufficient explanation. Why? First, anxious attachment is far too common: In working class samples the majority of children are anxiously attached (Broussard, 1995), and most do not become violent. Second, anxious attachment is a relatively stable interpersonal strategy characterized in adulthood by the dismissing of relationships as unimportant (this tends to be the pattern of the avoidant individual), or as a preoccupation, admittedly sometimes quite angry, with past attachment relationships (the adult outcome of an anxious resistant attachment pattern) (Main, Kaplan, & Gassidy, 1985). Neither of these strategies resembles the uncontrolled behavior of violent males. Third, clinical evidence suggests that, far from dismissing the importance of relationships, such men sometimes go to extreme lengths to retain their dysfunctional relationships with their victims following violent episodes, expressing contrition, despair, and a commitment to change (Walker, 1979). Finally, but perhaps most important, aggression should not be confused with violence. Whether we conceive of aggression as inherent to human nature, or as a response to frustration, it is far too pervasive to be a sufficient cause of a physical attack on the body of another human being, which thankfully is less common. We would argue that violence involves an additional crucial component that predisposes such individuals to act on bodies rather than on minds. It is the inadequacy of their capacity to think about aggression in relation to attachment that pushes them into violent acts in intimate relationships.

ATTACHMENT AND THE CAPACITY TO REPRESENT MENTAL STATES

At the heart of the attachment theory account oI~ violence, is the caregiver’s peisistent failure to recognize the child’s subjective state in infancy. We have argued that without an intrinsic capacity for experiencing oneself as driven by wishes and desires, constrained by beliefs and expectations, the child depends on the attachment figure to discover his or hex own subjectivity (Fonagy & Target, 1995; Fonagy & Target, 1997). In my experience, few children wake up one morning and say "Hey! I think therefore I am." The child "finds himself" in the caregiver’s mind as an intentional being motivated by mental states, beliefs, and desires. This representation is internalized as the core of the psychological self. Thus, the realization of subjectivity might be more accurately stated: "My caregiver thinks of me as thinking, therefore I exist as a thinker."

There is some evidence consistent with this model: We have shown that the caregiver’s capacity to reflect on mental states in describing their own childhood predicts the child’s security of attachment (Fonagy, Steele, Moran, Steele, & Higgitt, 1991a). If the caregiver sees her or his own past in mental state terms—what mother and father thought and how they responded psychologically—they were two to three times more likely to have secure attachment to their children. The quality of early attachment is also a relatively good predictor of the child’s capacity to learn that other people’s behavior is best understood in terms of mental states—the so-called "theory of mind" (Premack & Woodruff, 1978). Children securely attached in infancy are more likely than their insecurely attached counterparts, at age 5 years, to pass cognitive tests of understanding affective states in others (Fonagy, 1997; Fonagy, Redfern, & Charman, 1997a). The child’s understanding of minds critically depends on a developmental opportunity to find himself or herself represented in the caregiver’s mind as a mentalizing individual. The parents’ mentalizing capacity is also a good predictor. Thus, a theory of mind is, first of all, a theory of self. Mentalization, the capacity to understand and interpret human behavior in terms of the putative mental states underpinning it, arises through the experience of having been so understood in the context of an attachment relationship. This, in my view, is a critical aspect of the transgenerational transmission of abuse.

THE ROLE OF ABUSE IN THE HISTORY OF VIOLENT MALES

We have already touched on the common observation that violent men often have a history of abuse in childhood (Kalmuss, 1984; Revitch & Schlesinger, 1981; van der Kolk & Fisler, 1994). In a prison-based study, we have systematically examined narratives of attachment histories of violent men (Levinson & Fonagy, 1998). Attachment interview narratives of violent males were characterized by a predominantly dismissing pattern, overtly denigrating or disavowing attachment relationships, and a high prevalence of early, unremitting trauma. Unlike the results in a personality disordered comparison group, unresolved trauma, as defined in the Adult Attachment Interview coding system, was not common in this group. Cognitive irregularities when trauma is discussed indicate unresolved trauma in the AAI. For example, slips of the tongue, confusions of past and present, prolonged pauses, or unexpected intrusions of the trauma into other contexts. This was not present in the violent men’s narratives. By far the most marked feature of these interviews was the refusal of almost all the violent offenders in the sample to either spontaneously, or in response to interrogation, comment on mental states in the context of attachment relationships—either their own or those of their caregivers.

For example, one man, in prison for (among other things) grievous bodily harm against his girlfriend, described in disturbing detail how his alcoholic father regularly, on returning home, emptied his bladder on him and his younger sister. When asked, as part of the interview protocol, "Why do you think your parents behaved as they did?" he responded: "You tell me. You are the f’ing psychologist." Throughout the rest of his interview, he mentioned mental states but a handful of times, and on each of these occasions it was in the context of what warders, fellow prisoners, or the police felt or thought. When discussing his wife, his lover, his children, or his parents, he addressed their social and economic circumstances, their physical environment, their personality, their psychopathology, and most commonly their observed behavior. "My wife, she does OK. She is on child support." "My father was right off his rocker. He’d belt us one, no trouble." He was remarkably insensitive about his wife: " She likes a bit of the rough stuff, y’know." The same man was, however, clearly sensitive to the warder’s mental states: "They ain’t bad the screws in this nick. They take a shine to you so long as you know what they are after. They ain’t saying—but you’ll know."

We suggest that violent acts against women are often committed by men with inadequate mentalizing capacities. We assume that they are frequently victims of childhood abuse who cope by refusing to conceive of their attachment figure’s thoughts and thus avoid having to think about their attachment figure’s wish to harm them. The lack of safety of their environment repeatedly triggers their attachment system, but the behavior of their caregiver destroys a mentalizing intentional stance. How could 5-or 6-year-old children begin to start thinking about the mental states that might have caused a parent, who claims to care, to regularly humiliate and degrade them? Under the pressure of the combined need to seek comfort, yet at the same time to escape, children’s minds compromise, accepting physical comfort yet creating mental distance. Unfortunately, in so doing, they are forced to continue to disn*pt the capacity to represent mental states in themselves and in others in the context of attachment relationships. They continue to operate on inaccurate or schematic impressions of thoughts and feelings that handicap them enormously when confronted with intimate relationships.

There are two propositions here: first, that individuals who experience childhood trauma are likely to defensively inhibit their capacity to mentalize and second, that some characteristics of violent males may be linked to the developmental pathology associated with this inhibition. I will address each one in depth.

THE IMPACT OF MALTREATMENT ON MENTALIZING

Accumulating evidence suggests that maltreatment impairs the child’s reflective capacities and sense of self. Schneider-Rosen and Cicchetti (1984, 1991) noted that abused toddlers showed less positive affect on recognizing themselves in the mirror than did controls. Beeghly and Cicchetti (1994) showed that these toddlers had a specific deficit in use of internal state words and that such language tended to be context-bound. Our Menninger study of maltreated 5- to-8-year-olds found specific deficits in tasks requiring mentalization, particularly among those referred for sexual or physical and sexual abuse. They could not solve puzzles requiring them to conceive of one person’s false beliefs concerning a second person’s false beliefs. These results suggest that maltreatment may cause children to withdraw from the mental world. Their attachment behaviors, their proximity seeking, is disorganized because they desperately seek physical closeness while trying to create mental distance.

Why should this happen? First, recognition of the mental state of the other can be dangerous to the developing self because the child who recognizes the hatred implied by the parent’s abuse is forced to see himself or herself as unlovable. Second, intentional states may be denied or distorted. Abusive parents commonly claim beliefs or feelings at odds with their behavior. The child cannot test or modify his or her own representations, which become rigid or inappropriate and may be abandoned. Third, the public world, where reflective function is common, may give rise to alternative models of experiencing the self that are rigidly kept separate from the attachment context. Finally, the dysfunction may occur, not because of the maltreatment, but of the family atmosphere that surrounds it. Authoritarian parenting, commonly associated with maltreatment, is also known to retard the development of mentalization (Astington, 1996). These youngsters and their mothers find it difficult to take a playful stance (Alessandri, 1992), so the social scaffolding for the development of mentalization may be absent in such families.

If lack of consideration for the child’s intentionality is pervasive, consequences may occur not just at the functional but also at the neurodevelopmental level. The work of Bruce Perry (1997) suggested that Romanian orphans, institutionalized shortly after birth and suffering severe neglect and maltreatment during most of the first year of their lives, show significant loss of cortical function in the orbital frontal areas. These areas have been independently shown to be involved with infemng mental states (Frith, 1996). At 4 years, those who had been adopted before 4 months showed far less frequent disorganized attachment than those adopted later (Fisher, Ames, Chishoim, & Savoie, 1997). It has been independently demonstrated that insecure, particularly disorganized, attachment is associated with a far slower return to baseline of separation-induced cortisol elevation (Spangler & Grossman, 1993). Chronic exposure to raised levels of cortisol associated with chronically insensitive caregiving may bring about neurodevelopmental anomalies that result in a mentalizing deficit. I believe that male violence often has neurobiological underpinnings.

MALE VIOLENCE AND DEFICIT IN MENTALIZING

So—to the second proposition—are some characteristics of violent males to be explained by a deficit of mentalization? This assumption is overly simplistic. There are variations across situations or types of relationships. Even incarcerated men may, as I have said, be aware of the mental states of prison officers.

Following the principles of Kurt Fischer’s "dynamic skills theory" of development (Fischer, Kenny, & Pipp, 1990), we may assume that maltreatment is associated with a "fractionation" or splitting of mentalization across tasks and domains. Just as the understanding of conservation of liquid does not generalize to conservation of area, reflective capacity in one domain of interpersonal interaction may not generalize to others. In violent men, the normal coordination of previously separate skills has not come about, fractionation has seemed adaptive to the individual and has continued to dominate over integration.

Certain types of violence are undoubtedly associated with a moral disengagement consequent on a more or less total failure of mentalization. There are at least four reasons why this is likely to be the case (Fonagy, Target, Steele, & Steele, 199Th; Fonagy et al., 1997c): (1) Individuals without a well-established sense of their own identity may more readily feel that they are not responsible for their actions because they genuinely lack a sense of agency; (2) mentalizing capacity is anticipating the psychological consequences of others; (3) a limited capacity to mentalize could cause the person to treat others like physical objects; and (4) limitations on mentalizing are likely to cause a certain fluidity within the mental representational system that can lead to liberal rationalizations. Certain violent men show a remarkable, self- serving capacity to reinterpret unacceptable conduct as acceptable.

Underlying these ideas is the pivotal notion that violent men extend their natural aggression into violence through a defensively generated deficit in sociocognitive functioning. There is good evidence that individuals with antisocial personality disorder fail to respond with a sense of shock to photographic images of human suffering (Blair, 1995), and Type II violence probably does involve some kind of pervasive dysfunction of what James Blair has called the "violence-inhibiting mechanism." By contrast, I believe that Type I violence, while entailing some of the same mechanisms, actually involves, as one of its essential components, an acute but grossly constricted sensitivity to the mental state of the victim. Let me explain what I mean.

DISORGANIZATION OF ATTACHMENT IN INFANCY AND 
THE VULNERABILITY TO LATER TRAUMA THAT THIS ENGENDERS

Why should emotionally charged interactions trigger a regression to nonmentalistic thinking? Schuengel (1997) has recently provided evidence for Main and Hesse’s (1990) hypothesis that caregivers of infants whose attachment is disorganized, frequently respond to the infant’s distress cues by frightening, frightened, or dissociated behavior. The infant’s emotional expression triggers dissociation, a temporary failure on the part of the caretaker to perceive the child as an intentional person. The child comes to experience his or her own arousal as a danger signal for abandonment. It should not surprise us then that fear of abandonment in such children can become a trigger for nonmentalizing functioning; it brings forth an image of the parent who withdraws from the child’s attachment cues in a state of anxiety or rage, to which the child reacts with a complementary dissociative response.

Thus far, we have skirted around the central implication of this model. I have explained that internalization of the caregiver’s image of the child as an intentional being is central in self-organization. If this image is accurate, the child’s self-representation will map on to what could be called a "constitutional self" (the child’s experience of an actual state of being, the self as it is). When the child feels anxious, the caregiver’s contingent reflection of this anxiety will be internalized and eventually serve as a symbol for the internal state (Gergely & Watson, 1996). The representation will be true to the child’s primary experience. Disorganized attachment precludes the development of such an organic self-image. Internal experience is not met by external understanding; it remains unlabeled and confusing, and the uncontained affect generates further dysregulation.

There is overwhelming pressure on the child to develop a representation for internal states. Winnicott (1967) warned us that failing to find his or her current state mirrored, the child is likely to internalize the mother’s own state as part of his or her self-structure. The child incorporates into his or her nascent self-structure a representation of the other (Fonagy & Target, 1995). When confronted with a frightened or frightening caregiver, the infant takes in as part of himself or herself the mother’s feeling of rage, hatred, or fear, and her image of the child as frightening or unmanageable. This painful image must then be externalized for the child to achieve a bearable and coherent self-representation. The disorganized attachment behavior of the infant, and its sequelae—bossy and controlling interactions with the parent—permit the externalization of parts of the self and limit further intrusion into the self-representation. The dissociated core of the self is an absence, rather than genuine psychic content. It reflects a breach in the boundaries of the self, creating an openness to colonization by the mental states of attachment figures. Disastrously, in the case of some children maltreated later in development, this will not be a neutral other but rather a torturing one. Once internalized and lodged within the self-representation, this "alien" representation will have to be expelled not only because it does not match the constitutional seir, but also because it is persecutory. The consequences for interpersonal relationships can be disastrous.

This constellation, we believe, is the root of disorganized attachment. Why do we think so? (1) The disorganized behavior of the infant is replaced, during the first 5 years of life, by brittle behavioral strategies of controlling the parent, through either aggression or age-inappropriate caregiving behavior (Cassidy & Marvin, 1992; Main & Cassidy, 1988). (2) The parents experience the child as taking control and themselves as increasingly immobilized, and failing to provide caregiving (George & Solomon, 1996; Solomon & George, 1996). (3) The mothers of disorganized children describe these children as replicas of themselves, and experience themselves as merging with the child. We assume that these experiences are explained by the children externalizing aspects of their self-representation that relate, not to the mother’s representation of the child, but the representation of the mother, now within the child’s self. (4) Precocious caregiving behavior (West & George, in press) is also consistent with the idea that the representation of the mother is internalized into the self. (5) Most follow-up studies also show an association between disorganized attachment and deviant levels of aggressive behavior (Lyons-Ruth, 1996).

The externalization of the image of the mother from within the self-representation makes this representation more coherent. Such externalization can work only if the mother is controlled sufficiently for the alien self-representation to be experienced as external. This strategy may be reinforced, in childhood, insofar as offensive or threatening behavior often compels the adult to resume a position of authority and thus reactivate the parent’s own caregiving system that the parent had temporarily abandoned (West & George, in press).

The mechanism described here is an example of the psychoanalytic notion of projective identification (Klein, 1946) or, more specifically, what Elizabeth Spillius (1994) has termed "evocatory projective identification." The disorganized pattern of attachment is rooted in a disorganized self. The individual, when alone, feels unsafe because of a torturing representation from which he or she cannot escape because it is experienced from within the self. Unless the relationship permits externalization, the individual feels at risk of disappearance, psychological merging, and losing all coherence and relationship boundaries.

In adulthood, disorganized self-representation still manifests as an overwhelming need to control the other. Violent men have to establish a relationship in which their partner acts as a vehicle for intolerable self states. They manipulate the relationship to engender the self-image in the other that they feel desperate to disown. They resort to violence at times when the independent mental existence of the other threatens this process of externalization. At these times, dramatic action is taken because of a terror that the coherence or the self will be destroyed by the return of what has been externalized.

The act of violence then performs a dual function: to recreate and reexperience the alien self within the other and to destroy it in the unconscious hope that it will be gone forever. Perceiving the terror in the eyes of their victim, they are reassured. Their subsequent pleas for forgiveness are genuine, because of their overriding need for a relationship where this externalization is possible. Let me conclude by considering in some detail the clinical presentation of men involved in partner abuse in terms of the theoretical framework proposed. This is based both on published clinical descriptions and our own interviews with men imprisoned because of their violence.

THE CLINICAL PRESENTATION OF MEN WHO ARE VIOLENT TOWARD WOMEN

There is evidence that such men find it intolerable to be alone: They report feeling vulnerable and abandoned (Dutton, in press). A man, unusually in prison ror raping his wife, described how his relationship with her disintegrated after he lost his job so she had to take part-time employment and he was forced to spend his days alone in the house. He complained of "having too much time to think," but his account made it clear that while he had been employed, he was able to think of his wife as fearful and totally dependent on him. Being at home, unemployed and confronted with evidence of her competence, he began to see her as successful and quite socially skilled. He described these ideas emotionally, as making him feel totally "gutted" (empty). He had no one available to be the vehicle for the frightened, dependent part of himself.

A second key feature of partner violence is escalating tension and angry outbursts on the part of the man, normally accompanied by recognition of wrong doing but also of being out of control (Walker, 1979). The impulsivity is, in our view, linked to their inability to coherently represent their emotional states. Fearing abandonment, they quickly resort to prementalistic physical action-centered strategies. They lose the capacity to influence their partner through changing their state of mind. A kind of mentalistic learned helplessness sets in. It is most readily triggered by the threatened loss of the vehicle for their self-representation. A younger man described to us how mild irritation could become uncontrollable rage. While he stayed out late drinking most evenings, his partner was required to return within seconds of the times they had agreed. Her lateness, whatever the cause, would lead to rapidly mounting anger, which he recognized to be unreasonable but was unable to contain. His phrase for describing the process was telling: "It was like holding onto hot metal, I just had to let go, I had to let her have it." The result may not be physical violence. It may be words that might sound like attempts at changing the other’s intentions but are in fact intimidation and coercion. The aim is physical control and possession of the woman’s body, her physical being. The oppressive jealousy of these men is not an expression of love or desire, but a crude manipulation to keep the partner captive and available to regulate self-states.

Why is the outburst of violence normally preceded by this intensification of abandonment anxiety, manifesting as controlling behavior? In our experience, these periods are generally marked by the woman’s increased autonomy. Often she is lulled into a false sense of security by her partner’s contrition following the most recent violent encounter. She may be puzzled by her partner’s excessive jealousy because, if anything, she feels more committed to him. What she fails to realize is that her relative sense of being at ease is an intolerable threat to her partner because oI~ his need to externalize an alien part of the self-representation.

What complicates matters in this phase is that the man appears to regress to a mode of thinking where only one version of reality is possible. The mere possibility of a false belief is denied (Fonagy & Target, 1996).  He loses his ability to "play with reality" and any attempt by the victim to present an alternative version of events is perceived as a provocative gesture or an attempt to drive him crazy. The insane possessiveness is aggravated by this singularity and rigidity of perspective. This primitive experience of psychic reality, which equates internal with external, signals his mounting anxiety associated with mental separation. She is no longer what he needs her to be. Delusional ideas that the partner had changed in character or appearance were frequently reported. To become "his" once more, she has to be forced into the role of the terrified infant and accept the unbearable image of the petrifying or petrified mother. Or as one of our subjects put it: "She has to be taught a lesson."

The trigger for the violent attack is often incidental, but, almost inevitably, it is a further indicator of the woman’s psychological separateness. Examples may be a new hairstyle, a passing remark that expresses concern about someone other than the perpetrator. The trigger is insignificant. It is inevitable that anything the woman does might remind the man that she is a person in her own right (as indeed she has always been and will always be). If she stays silent, or if she answers back, both can remind the batterer of her intentionality.

What ensues is a frenzied and often prolonged sadistic act where she is made to suffer. Such men do not lack empathy altogether. In fact they can be strikingly sensitive. It is essential that the man see the woman’s reaction—and within that something that he would otherwise experience as part of himself. The crucial aim is that it should be out there rather than in him. And therefore, the woman’s struggle, pleading, and suffering are vital features of his experience.

Reports of these attacks tend to support such a formulation. For example, the man’s rage is rarely "blind;" he remembers well how the woman reacted, and reports adjusting his actions to elicit what he needs ("I wanted to hear her scream, so I hit her harder"). There are subtle and specific links between the emotional states the man reports prior to the act of violence and the emotions he attributes to his victim during his attack. For example, a man who had been increasingly troubled by flashbacks to childhood trauma beat his wife wondering what kind of memories she would have of the experience. During the attack there tends to be a merger of subjectivities and the men perceive in the woman’s mind the feelings they try to shed, which tends to trigger an intensification of the attack. For example, the man we first described reported feeling mocked by his partner and he raped her principally to "take the smirk off her face."

It is hard to feel empathy for such actions. Yet, in a sense, it is only at these moments that such individuals feel coherent and real. There is undoubtedly an aspect of these tirades that at least superficially appears cathartic given the build-up of tension during the preceding period. Yet drive discharge seems to be an incidental part of the process. Men do not teport pleasure associated with these episodes—rather a restoration of an inner gestalt, a release of tension, the creation of an inner peace, an odd kind of tranquillity. I have often been puzzled by such accounts of violence where my perhaps understandable expectation was that turmoil and conflict would ensue, but descriptions were more consistent with the reinstatement of homeostasis, even if the context is chaos, like being in the eye of a hurricane. The calm reflects the successful destruction of the woman’s psychic independence. She is once again the mere vehicle of her partner’s pathological projective processes.

Inevitably, a dramatic and rapid shift may be observed in the relationship. In this capacity, the woman once again becomes uniquely important to his psychological well-being. The move from violence to contrition may at first seem puzzling. We know that the remorse is anything but genuine. Yet the need for the woman, which is expressed in obsequious self-abasing gestures, is genuine enough. Paradoxically, the man’s reaction frequently triggers the woman’s caretaking attachment system, as she senses his genuine need. The tragedy is that the man’s need is not for his partner as a real person, and the more she is reassured by his apparent remorse and protestations of commitment, the more she relaxes and permits her individuality to emerge, the closer she brings the next episode of terrible violence.

There is a further important aspect of the clinical picture that our approach helps to explain. One might search for hostility toward women born out of maternal rejection in the history of violent men. However, available evidence, both empirical and clinical, points to rejection by the father as a predictor of anger, domination, and emotional and physical abuse in the intimate relationship (Dutton, 1995a). We suggest that the father’s hostility plays an important role.

We suggest that a father has the potential to introduce the viewpoint of the third into a disorganized infant—caregiver relationship. The father’s absence and negativity cannot, in our view, cause male violence without the early attachment antecedents we have described, but his behavior may be a necessary contributing factor.

Beyond rejection, paternal behavior has been reported to have been characterized by various forms of shaming, insults, and random punishment. When a child’s mentalizing capacity is already vulnerable as a consequence of disorganized attachment and the associated disorganized self-structure, the child is unlikely to be able to withstand the paternal humiliation. Frightened and frightening caregiving in infancy creates a gap, an open window, within the self-structure that may then be filled by the father’s hostility toward the child. That hostility, mindless and terrifying, is now also experienced as within the self. It is hard for most of us to appreciate the desperation that such a predicament engenders.

CONCLUSIONS

There is a noticeable overlap between the formulation offered for men who are violent with women and one we have suggested could account for borderline personality organization (Fonagy, 1991). Is this because we tend to identify similar patterns whatever group of patients we study? Or are there genuine similarities between the two groups? I hope I am not being totally self-serving in opting for the latter alternative. In fact, the link be-. tween borderline phenomena and violence against women has been made by others (Dutton, 1995a, b), and there is a logic to this. The combination of troubled early attachment and maltreatment culminating in profound personality pathology follows different developmental paths in men and women (Fonagy & Target, 1995). Women with this kind of background often attempt suicide and deliberately harm themselves, having projected alien parts of the self into their own bodies. They develop disorganized violent relationships with their physical selves comparable to those of men with their sexual partner—hoping to destroy the "alien" self within. They rapidly acquire a borderline diagnosis. The social structure within which we live appears to lead men with similar problems down quite a different path. They become the abusers rather than the abused, tormenting others apparently without corresponding self-torment.

I believe both groups to be victims of a social structure that does not support families and the parenting function sufficiently to enable us to give adequate care to our young, that provides few alternative structures within which children could come to arrive at self-recognition through social interchange with a benevolent and dependable caretaker—a social structure that most cruelly places supreme emphasis on individuation, individuality, and the achievement of selfhood, the most vulnerable aspect of the individual with disorganized attachment. I believe that disorganized attachments and associated complications in self-organization can occur, at least in part, because society has relinquished its caretaking function, demolished its institutions for supporting emotional development, and shifted its priorities from the mental and emotional to the material. In the increasing violence around us, we may be seeing the casualties of this cavalier approach. The problem of male sexual violence cannot be adequately addressed without genuine commitment of our resources to prevention, to fully supporting parents and young children, and radically revising social priorities in the direction of a relationship-and-emotional-development-focused approach to education and socialization.

REFERENCES

Ainsworth, M. D S. Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale, NJ: Eribaum.

Alessandri, S. M. (1992). Mother-child interactional correlates of maltreated and nonmaltreated children’s play behavior. Development and Psychopathology, 4, 257.270.

Astington, J. (1996). What is theoretical about the child’s theory of mind?: A Vygotskian view of its development. In P. Carruthers & P. K. Smiih (Eds.), Theories of theories of mind (pp. 184-199). Cambridge: Cambridge University Press.

Bandura, A. (1986). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.

Beeghly, M., & Cicchetti, D. (1994). Child maltreatment, attachment, and the self system Emergence of an internal state lexicon in toddlers at high social risk. Development and Psychopathology, 6, 5-30.

Bion, W. R. (1962). Learning from experience. London: Heinemann.

Blair, R. J. R. (1995). A cognitive developmental approach to morality: Investigating the psychopath. Cognition, 57, 1-29.

Bowlby, J. (1960). Grief and mourning in infancy and early childhood. Psychoanalytic Study of the Child, 15, 3-39.

Bowlby, J. (1973). Attachment and loss, Vol. 2: Separation Anxiety and anger. London: Hogaith Press and Institute of Psycho-Analysis.

Broussard, E. R. (1995). Infant attachment is a sample of adolescent mother. Child Psychiatry and Human Development, 25, 211-219.

Cassidy, J., & Marvin, R. S. (1992). Attachment organization in preschool children. Coding Guidelines. Seattle: MacArthur Working Group on Attachment—unpublished coding manual.

Dobash, R. E., & Dobash, R. P. (1979). Violence against wives. A case against the patriarchy. New York: Free Press.

Dutton, D. G. (1995a). The domestic assault of women: Psychological and criminal justice perspectives (2nd ed.). Vancouver, BC: University of British Columbia Press.

Dutton, D. G. (1995b). Male abusiveness in intimate relationships. Clinical Psychology Review, 15, 567-581.

Dutton, D. G. (Ed.) (in press). Treatment of assaultiveness. New York: Guilford.

Dutton, D. G., Saunders, K., Starzomski, A., & Bartholomew, K. (1994). Intimacy-anger and insecure attachment as precursors of abuse in intimate relationships. Journal of Applied Social Psychology, 24, 1367-1386.

Fischer, K. W., Kenny, S. L., & Pipp, S. L. (1990). How cognitive processes and environmental conditions organize discontinuities in the development of abstractions, In C. N. Alexander E. J. Langer & R. M. Oetzel (Eds.), Higher stages of development (pp. 162.187). New York: Oxford University Press.

Fisher, L., Ames, E. W., Chisholm, K, & Savoie, L (1997). Problems reported by parents of Romanian orphans adopted to British Columbia. International Journal of Behavioral Development, 20, 67.82.

Fonagy, P. (1991). Thinking about thinking: Some clinical and theoretical considerations in the treatment of a borderline patient. International Journal of Psycho-Analysis, 72, 639-656.

Fonagy, P. (1997). Attachment and theory of mind: Overlapping constructs? Association for Child Psychology and Psychiatry Occasional Papers, 14, 31-40.

Fonagy, P. (in press). A review of psychoanalytic theory from the point of view of attachment theory and research. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment theory and research. New York: Guilford Publications.

Fonagy, P., & Target, M. (1995). Towards understanding violence: The use of the body and the role of the father International Journal of Psycho-Analysis, 76, 487-502.

Fonagy, P., & Target, M. (1996). Playing with reality: I. Theory of mind and the normal development of psychic reality. International Journal of Psycho-Analysis~ 77, 217-233.

Fonagy, P., & Target, M. (1997). Attachment and reflective function: Their role in self-organization. Development and Psychopathology, 9, 679-700.

Fonagy, P., Moran, G. S., & Target, M. (1993). Aggression and the psychological self. International Journal of Psycho-Analysis, 74, 471-485.

Fonagy, P., Steele, H., Moran, G., Steele, M., & Higgitt, A. (1991a). The capacity for understanding mental states: the reflective self in parent and child and its significance for security of attachment. Infant Mental Health Journal, 13, 200-217.

Fonagy, P., Steele, H., & Steele, M. (1991b). Maternal representations of attachment during pregnancy predict the organization of infant-mother attachment at one year of age. Child Development, 62, 891-905.

Fonagy, P., Target, M., Steele, M., & Steele, H. (199Th). The development of violence and F crime as it relates to security of attachment. In J. D. Osofsky (Ed.), Children in a violent society (pp. 150-177). New York: Guilford Press.

Fonagy, P., Redfein, S., & Charman, T. (1997a). The relationship between belief-desire reasoning and a projective measure of attachment security SAT). British Journal of Developmental Psychology, 15, 5 1-61.

Fonagy, P., Target, M., Steele, M., Steele, H., Leigh, T., Levinson, A., & Kennedy, R. (1997c). Morality, disruptive behavior, borderline personality disorder, crime and their relationship to security of attachment. In L. Atkinson & K. J. Zucker (Eds.), Attachment and psychopathology (pp. 223-274). New York: Guilford Press.

Freud, A. (1960). Discussion of Dr. Bowlby’s paper (Grief and mourning in infancy and early childhood). In The writings of Anna Freud (pp. 167-186). New York: International University Press, 1969.

Filth, C. (1996). Neuropsychology of schizophrenia. What are the implications of intellectual and experiential abnormalities for the neurobiology of schizophrenia? In E. C. Johnstone (Ed.), Biological psychiatry London, England: Royal Society of Medicine Press Limited.

George, C., & Solomon, J. (1996). Representational models of relationships: Links between C caregiving and attachment. In C. George & J. Solomon (Eds.), Defining the caregiving system (Infant Mental Health Journal; Volume 17). New York: Wiley.

Gergely, G., & Watson, J. (1996). The social biofeedback model of parental affect-mirroring. International Journal of Psycho-Analysis, 77, 1181-1212.

Goldner, V., Penn, P., Sheinberg, M., & Walker, G. (1990). Love and violence: Gender paradoxes in volatile attachments. Family Process, 29, 343-364.

Gordon, L (1989). Heroes of their own lives: The politics and history of family violence Boston, 1880-1960. New York: Penguin.

Grosskruth, P. (1986). Melanie Klein: Her world and her work New York: Knopf.

Kalmuss (1984). The intergenerational transmission of marital aggression. Journal of Marriage and the Family, 46, 31-39.

Klein, M. (1946). Notes on some schizoid mechanisms. In M. Klein, P. Heimann, S. Issacs, & .J. Riviere (Eds.), Developments in psychoanalysis (pp. 292.320). London: Hogarth Press.

Kohut, H. (1971). The analysis of the self. New York: International Universities Press.

Levinson, A., & Fonagy, P. (1998). Attachment classification in prisoners and psychiatric patients. Manuscript submitted for publication.

Lichtenberg, J. D. (1995). Can empirical studies of development impact on psychoanalytic theory and technique? In T. Shapiro & R. N. Emde (Eds.), Research in process, development, outcome (pp. 261-276). New York: International Universities Press.

Lie, G., Schilit, R., Bush, R., Montague, M., & Reyes, I. (1993). Lesbians in currently aggressive relationships: How frequently do they report aggressive past relationships? Violence and Victims, 6, 121-135.

Lyons.Ruth, K (1996). Attachment relationships among children with aggressive behavior problems: The role of disorganized early attachment patterns. Journal of Consulting and Clinical Psychology, 64, 32-40.

Main, M., & Cassidy, J. (1988). Categories of response to reunion with the parent at age 6: Predictable from infant attachment classifications and stable over a 1-month period. Developmental Psychology, 24, 415-426.

Main, M., & Hesse, E. (1990). Parents’ unresolved traumatic experiences are related to infant disorganized attachment status: Is frightened and/or frightening parental behavior the linking mechanism? In M. Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment in the preschool years: Theory research and intervention (pp. 161-182). Chicago: University of Chicago Press.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. Monographs of the Society for Research in Child Development, 50(1-2), 66-104.

Mayseless, 0. (1991). Adult attachment patterns and courtship violence. Family Relations, 40, 21-28.

Meloy, R. J. (1992). Violent attachments New Jersey: Jason Aronson.

Osofsky, J. D. (1997). Children in a violent society. New York: Guilford Press.

Perry, B. (1997). Incubated in terror: Neurodevelopmental factors in the "cycle of violence" In J. Osofsky (Ed.), Children in a violent society (pp. 124-149). New York: Guilford Press.

Pleck, E. (1987). Domestic tyranny: The making of social policy against family violence from colonial times to the present. New York: Oxford University Press.

Premack, D, & Woodruff, G. (1978). Does the chimpanzee have a "theory of mind"? Behavioral and Brain Sciences~ 1, 515-526.

Rajecki, D. W., Lamb, M., & Obmascher, P. (1978). Toward a general Iheory of infantile attachment: A comparative review of aspects of the social bond. The Behavioral and Brain Sciences, 3, 417-464.

Revitch, E., & Schlesinger, L. B. (1981). Psychopathology of homicide. Springfield, IL: Charles Thomas.

Sandier, J. (1960) The backgivund of safety, from safety to superego. Selected papers of Joseph SandIer (pp. 1-8) London: Karnac, 1987.

Saunders, D. G. (1992). A typology of men who battei. Three types derived from cluster analysis. American Orthopsychiatiy, 62, 264-275.

Schneider-Rosen, K., & Cicchetti, D. (1984). The relationship between affect and cognition in maltreated infants: Quality of attachment and the development of visual self. recognition. Child Development; 55, 648-658.

Schneider-Rosen, K., & Cicchetti, D. (1991). Early self-knowledge and emotional development: Visual self-recognition and affective reactions to mirror self.image in maltrealed and non-maltreated toddlers. Developmental Psychology, 27, 481-488.

Schuengel, C. (1997). Attachment, loss, and maternal behavior. A study on inteigenerational transmission Leiden, The Netherlands: University of Leiden Press.

Schur, M. (t960). Discussion of Dr. John Bowlby’s paper. The Psychoanalytic Study of the Child 15, 63-84.

Solomon, J., & George, C. (1996). Defining the caregiving system: Toward a theory of caregiving. In C. George & J. Solomon (Eds.), Defining the caregiving system (Infant Mental Health Journal, Volume 17). New York: Wiley.

Sorenson, S. R., & Telles, C. A. (1991). Self-reports of spousal violence in Mexican· American and non-Hispanic white population. Violence and Victims, 6, 3-15.

Spangler, G., & Grossman, K. E. (1993). Biobehavioral organization in securely and insecurely attached infants. Child Development 64, 1439-1450.

Spillius, E. B. (1994). Developments in Kleinian thought: Overview and personal view. Psychoanalytic Inquiry, 14, 324-364.

Spitz, R. A. (1960). Discussion of Dr. John Bowlby’s paper. The Psychoanalytic Study of the Child, 15, 85-94.

Steele, H., Steele, M., & Fonagy, P. (1996). Associations among attachment classifications of mothers, fathers, and their infants. Child Development, 67, 541-555.

Straus, M. A. (1979). Measuring family conflict and violence: The conflict tactics scale. Journal of Marriage and Family, 41, 75-88.

Straus, M. A., Gelles, R. J., & Steinmetz, S. (1980). Behind closed doors: Violence in the American family. Garden City, NY: Anchor Press/Doubleday.

Taylor, L., Zuckerman, B., Hank, V., & Groves, B. (1994). Witnessing violence by young children and their mothers. Journal of Developmental and Behavioral Pediatrics, 15, 120-123.

van der Kolk, B., & Fisler, R. (1994). Childhood abuse and neglect and loss of self-regulation. Bulletin of the Menninger Clinic. 58 145-168.

van IJzendoom, M. H. (t995). Adult attachment representations, parental responsiveness, and infant attachment: A meta-analysis on the predictive validity of the Adult Attachment Interview. Psychological Bulletin, 117, 387-403.

Walker, J. E. (1979). The battered woman. New York: Harper & Row.

West, M., & George, C. (in press). Abuse and violence in intimate adult relationships: New perspectives from attachment theory. In D. G. Dulton (Ed.), Treatment of assaultiveness. New York: Guilford.

Winnjcott, D. W. (1965). The maturational pmcess and the facilitating environment London: Hogarth Press.

Winnicott, D. W. (1967). Mirror-role of the mother and family in child development. In P. Lomas (Ed.), The predicament of the family: A psycho-analytical symposium (pp. 26-33). London: Hogarth.

Yllo, K., & Straus, M. (1990). Patriarchy and violence against wives: The impact of structural and normative factors. In M. Straus & R. Celles (Eds.), Physical violence in American families (pp. 38 3-399). New Brunswick, NJ: Transaction Publishers.