Chapter 1
Introduction
D'you Know What I Mean?
Everyone wants to be understood. Indeed, these words pepper the discourse of the least clear speakers as if their best efforts were doomed to failure—and they knew it.
In this volume, we offer a way to understand the meanings behind unclear communication. More than that, we offer a theoretical model of distorted communication and dysfunctional behavior. It is a simple, intuitive model that can be understood by troubled adults. It is also a sophisticated and complex model that can guide clinicians to select effective treatments and that can help researchers to test hypotheses about the intricate and interwoven pathways to maladaptation. We combine genetic potentials, history of exposure to danger, mental processing of information about danger, and communication about danger into a functionally coherent model of psychological processes, the Dynamic-Maturational Model of attachment and adaptation (DMM; Crittenden, 1995,2008).
The DMM addresses normal thinking and behavior with the same principles that are used to describe dysfunctional thinking and behavior. The difference is that exposure to danger increases the probability of dysfunction. In this volume, we describe a wide range of human
2
Assessing Adult Attachment
adaptation in ways that foster understanding of maladaptation. We also recommend an assessment tool to elicit the crucial information about danger and adaptation to danger and a method for extracting information from the assessment. The tool is the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1985,1996). The method is the Dynamic-Maturational Model of discourse analysis for the AAI, as presented in this book.
Our goal is to describe maladaptation in a way that connects the experiences of threatened and threatening people with the skills of mental health professionals so that professionals can reduce distress and maladaptive behavior more successfully. We also seek to guide researchers to fine-grained, theory-based hypotheses that can refine our understanding of developmental processes in a way that multifactorial modeling alone cannot. The key, we think, is understanding. When we understand the meanings that distressed people struggle to communicate, a pathway to safety and comfort will be opened.
Our ideas build on a century of effort to understand and ameliorate psychological suffering. In both DMM theory and our method for analyzing AAIs, we have sought to retain the best ideas from all theories of psychological disorder while reframing and recombining these ideas to reflect the most current developmental and neurocognitive knowledge. The result addresses problems in assessment, diagnosis, and treatment of disorder in a fresh manner.
We chose the AAI as the assessment tool because it lets people tell their stories in their own way, thus preserving their reality. At the same time, the questions elicit crucial information without obscuring it with irrelevancies. We think these two things—the speakers' own words and pertinent questions, together with a method of discourse analysis that promotes understanding of what speakers mean—can be the bridge that connects people who dare not be clear with caring healers who understand with clarity.
THE ROOT OF THE PROBLEM
Defining Psychological Disorder
Psychological disorder has stubbornly resisted understanding and treatment for more than a century. The lack of understanding has led to fragmentation among those clinicians who cling almost religiously to a single theory of treatment (e.g., psychoanalytic, behavioral, cognitive,
2
Introduction
3
adaptation in ways that foster understanding of maladaptation. We also recommend an assessment tool to elicit the crucial information about danger and adaptation to danger and a method for extracting information from the assessment. The tool is the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1985, 1996). The method is the Dynamic-Maturational Model of discourse analysis for the AAI, as presented in this book.
Our goal is to describe maladaptation in a way that connects the experiences of threatened and threatening people with the skills of mental health professionals so that professionals can reduce distress and maladaptive behavior more successfully. We also seek to guide researchers to fine-grained, theory-based hypotheses that can refine our understanding of developmental processes in a way that multifactorial modeling alone cannot. The key, we think, is understanding. When we understand the meanings that distressed people struggle to communicate, a pathway to safety and comfort will be opened.
Our ideas build on a century of effort to understand and ameliorate psychological suffering. In both DMM theory and our method for analyzing AAIs, we have sought to retain the best ideas from all theories of psychological disorder while reframing and recombining these ideas to reflect the most current developmental and neurocognitive knowledge. The result addresses problems in assessment, diagnosis, and treatment of disorder in a fresh manner.
We chose the AAI as the assessment tool because it lets people tell their stories in their own way, thus preserving their reality. At the same time, the questions elicit crucial information without obscuring it with irrelevancies. We think these two things—the speakers' own words and pertinent questions, together with a method of discourse analysis that promotes understanding of what speakers mean—can be the bridge that connects people who dare not be clear with caring healers who understand with clarity.
THE ROOT OF THE PROBLEM
cognitive-behavioral, cognitive-analytic, interpersonal, family systems). Because each theory describes the problems of suffering individuals differently and none is more effective than the others at easing psychological suffering, we think that none reflects the meaning of suffering well enough. The names of the treatments suggest what is needed: an integration of meaningful ideas, from all the theories, that goes beyond individual pathology to place people in their context of relationships and families. Add to that a focus on danger and the psychological andi behavioral effects of exposure to unprotected and uncomforted danger) and one has the rudiments of a new and integrative theory of the development and treatment of dysfunction.
Instead, by side-stepping differences in theory, two similar and explicitly nontheoretical systems for diagnosing disorders were developed: the Diagnostic and Statistical Manual of Mental Disorders {DSM, North America) and International Classifications of Disease (ICD, World Health Organization). This was a major step toward describing behavior accurately. The previous focus on theory biased observation toward theory-based expectations. The focus on describing symptoms turned attention to accurate observation of the behavior of people who suffer. But we lost the meanings that theory had! With good observation, the number of disorders increased, comorbidity increased, and diagnoses of "not otherwise specified" increased (Angold & Costello, 2009; Goldberg, 2010). It seems we know in detail what troubled people do, but we do not know how to cluster them, how their behavior functions, what it means to them, or its implications for treatment.
If we are to understand people who suffer—and whose suffering distresses and sometimes harms other people—we need both sound observations and a language of shared meanings. This language needs to express the experiences of those who suffer in ways that they can understand while, at the same time, uniting clinicians from different theoretical backgrounds in ways that lead to more effective treatment. Further, this language needs to guide researchers to discover and address the crucial inconsistencies, distortions, and omissions in theory that can clarify both developmental and rehabilitative processes.
Defining Psychological Disorder A New Model ivith Familiar Roots
Psychological disorder has stubbornly resisted understanding and The Dynamic-Maturational Model (DMM) of attachment and adapta-
treatment for more than a century. The lack of understanding has led to turn developed out of Bowlby's integration of psychoanalytic theory
fragmentation among those clinicians who cling almost religiously to a with general systems theory (including emerging family systems the-
single theory of treatment (e.g., psychoanalytic, behavioral, cognitive, ory), ethology, and the cognitive neurosciences (Bowlby, 1969/1982,
4
Assessing Adult Attachment
1973,1980). Among theories of disorder, the DMM has the unique advantage of being prospective in that it is based on empirical evidence regarding developmental processes beginning in infancy and progressing forward to adixlthood (Ainsworth, 1989; Crittenden & Ainsworth, 1989). Other models of adult psychopathology are based on the behavior and recalled history of adult patients—and this, we think, is not understood well enough because of errors in thought and communication.
The DMM assumes that all people seek to understand their experi- \ ences, but some dangerous experiences cannot be understood early in life and, sometimes, cannot be understood at all. When danger occurs early, children need protection and comfort from someone they trust. If that is absent, misunderstanding, miscommunication, and maladapta-tion become more likely. Tracking the process of miscommunication and misunderstanding, especially around issues of danger and comfort, is central to the DMM.
Three central points define the DMM:
1. Dysfunction is a response to intolerable threats that often occur early in development when the child is not protected and comforted (Bowlby, 1969/1982,1973,1980).
2. Psychological processing is transformed developmentally in a progressive attempt to understand and protect the self from that which is neither apparent nor explained (Crittenden, 1995, 2008).
3. Maladaptive behavior, that is, psychological disorder, is the individual's best attempt to apply what he or she learned about danger while growing up to the adult tasks of self-protection, reproduction, and protection of children.
Beyond that, the DMM is built on insights from all the major theories of adaptation and disorder and these will feel familiar to most professionals. It is also built on sound developmental findings and emerging cognitive, neurological, and genetic (including especially epigenetic) findings. Intrinsic and ongoing ties to empirical research will keep the DMM honest and relevant. Its unique feature is that the DMM organizes the ideas: therefore it is more than a collection of ideas. The organization is developmental and functional and, thus, suited to the needs of clinicians who must identify deviancy and select suitable treatment responses. The DMM also meets the needs of researchers who must select, from among the almost limitless number of possible variables, those hypotheses that are crucial to understanding.
Assessing Adult Attachment^
1973, 1980). Among theories of disorder, the DMM has the unique advantage of being prospective in that it is based on empirical evidence regarding developmental processes beginning in infancy and progressing forward to adulthood (Ainsworth, 1989; Crittenden & Ainsworth, 1989). Other models of adult psychopathology are based on the behavior and recalled history of adult patients—and this, we think, is not understood well enough because of errors in thought and communication.
The DMM assumes that all people seek to understand their experiences, but some dangerous experiences cannot be understood early in life and, sometimes, cannot be understood at all. When danger occurs early, children need protection and comfort from someone they trust. If , that is absent, misunderstanding, irriscommunication, and maladapta- ( tion become more likely. Tracking the process of miscommunication and misunderstanding, especially around issues of danger and comfort, is central to the DMM.
Three central points define the DMM:
1. Dysfunction is a response to intolerable threats that often occur early in development when the child is not protected and comforted (Bowlby, 1969/1982,1973,1980).
2. Psychological processing is transformed developmentally in a progressive attempt to understand and protect the self from that which is neither apparent nor explained (Crittenden, 1995, 2008).
3. Maladaptive behavior, that is, psychological disorder, is the individual's best attempt to apply what he or she learned about danger while growing up to the adult tasks of self-protection, reproduction, and protection of children.
Beyond that, the DMM is built on insights from all the major theories of adaptation and disorder and these will feel familiar to most professionals, ft is also built on sound developmental findings and emerging cognitive, neurological, and genetic (including especially epigenetic) findings. Intrinsic and ongoing ties to empirical research will keep the DMM honest and relevant. Its unique feature is that the DMM organizes the ideas: therefore it is more than a collection of ideas. The organization is developmental and functional and, thus, suited to the needs of clinicians who must identify deviancy and select suitable treatment responses. The DMM also meets the needs of researchers who must select, from among the almost limitless number of possible variables, those hypotheses that are crucial to understanding.
Introduction
What We Offer in This Volume
We offer the rudiments of DMM theory of individual differences in psychological adaptation, together with a means of extracting meaning from the interpersonal discourse of adults (Ainsworth, 1979; Bowlby, 1980; Crittenden, 2008; Crittenden & Ainsworth, 1989). The method of discourse analysis forms the bulk of this volume and expands on the work of Main and Goldwyn (M&G; 1994). It differs from the M&G method of discourse analysis in having a wider range of constructs, identifying a wider range of self- and progeny-protective organizations, and specifying more precisely how attempts to adapt can fail.
A crucial difference is in the search for meaning in communication. Main and her colleagues state explicitly that their method looks at the dysfluence, not behind or through it, to assess the extent of speakers' coherence, or lack of coherence, of mind (Main, Hesse, & Goldwyn, 2008). In contrast, the DMM method seeks the underlying or obscured meaning that speakers cannot articulate fully or clearly. Rather than focusing on lack of coherence, we want to understand how speakers protect themselves and, to the extent possible, how they learned that psychological and behavioral strategy. This can help us to understand how the now maladaptive behavior was once protective. This understanding can promote respect by professionals for the children who our clients and patients had once been, and self-respect by distressed peo- ! pie who can find in the DMM a positive way of understanding them- -selves.
In the sections that follow, we introduce the AAI, explore the constructs underlying the DMM, and present a model of individual differences in adaptation based on an expansion of the work of Bowlby and Ainsworth (Ainsworth, 1973; Ainsworth, Blehar, Waters, & Wall, 1978; Bowlby, 1969/1982, 1973, 1980; Crittenden, 1995, 2008; Crittenden & Ainsworth, 1989).
WHY DO WE NEED A NEW METHOD?
Problems with the Main and Goldwyn Method
Even among its advocates, there are mixed opinions about the M&G-AAI (Slade, 2007). At its best, 22% of the variance in infants' attachment can be accounted for by knowing mothers' classifications on the AAI (van IJzendoorn, 1995); this relation is stronger among middle-class
6
Assessing Adult Attachment
and stable families than among at-risk families. This raises questions regarding the remaining 78% of variance, especially regarding less affluent and advantaged families. More recent work applying the M&G-AAI to a broader range of circumstances has found far more modest outcomes than expected (Bakermans-Kranenburg & van IJzendoorn, 2009). Specifically, analysis of 10,000 AAIs indicated that the M&G-AAI differentiated normative and at-risk samples, but with substantial overlap; security was found even in cases of maltreatment, psychiatric hospitalization, and violent criminality and the categories of very insecure attachment applied to some normative individuals. Moreover, some findings were contradictory or counter to theory. Further, little or no differentiation within at-risk groups was revealed, suggesting that the ABC+D model of attachment had little to offer that could guide differential treatment. We detail these problems in Chapters 15 and 16, but here it is worth noting that, after two decades and several hundred studies, the outcomes of the M&G-AAI, as well as ABC+D theory in general, are disappointing. Unfortunately, in spite of its wide usage, the M&G-AAI has remained under development, with neither the interview nor the manual being published for almost three decades. On the other hand, the conceptual underpinnings of the M&G method have remained unchanged since 1996 when "Cannot Classify" was introduced as a new classification (Hesse, 1996; Main et al, 2008).
We think six issues limit the current use of the M&G-AAI. First, it has been very difficult for trained coders to achieve reliability on the M&G classifications. Second, it is unclear what the M&G-AAI assesses; this problem is compounded by the unavailability of a manual for review by researchers and clinicians who have not taken the training course. Such an investment of time, prior to viewing the instrument, exceeds the resources and motivation of many potentially interested professionals. Third, the M&G classificatory method is based on work in the cognitive sciences from the late 1970s. Knowledge of mental functioning derived since Bowlby completed his 1980 volume has not been incorporated into the classificatory process. Fourth, the adult clas-i sificatory system is based on Ainsworth's infant classificatory system and does not account fully for the complexity of adult behavior. Fifth, the Ainsworth system was developed from observations of middle-income, low-risk samples of American infants and their parents. Thus, it may not adequately reflect the range of diversity found in other cultures and in risk populations. Finally, the notion of disorganization lacks empirical support and conceptual coherence (Rutter, Kreppner, &
Assessing Adult Attachment
and stable families than among at-risk families. This raises questions regarding the remaining 78% of variance, especially regarding less affluent and advantaged families. More recent work applying the M&G-AAI to a broader range of circumstances has found far more modest outcomes than expected (Bakermans-Kranenburg & van IJzendoorn, 2009). Specifically, analysis of 10,000 AAIs indicated that the M&G-AAI differentiated normative and at-risk samples, but with substantial overlap; security was found even in cases of maltreatment, psychiatric hospitalization, and violent criminality and the categories of very insecure attachment applied to some normative individuals. Moreover, some findings were contradictory or counter to theory. Further, little or no differentiation within at-risk groups was revealed, suggesting that the ABC+D model of attachment had little to offer that could guide differential treatment. We detail these problems in Chapters 15 and 16, but here it is worth noting that, after two decades and several hundred studies, the outcomes of the M&G-AAI, as well as ABC+D theory in general, arc disappointing. Unfortunately, in spite of its wide usage, the M&G-AAI has remained under development, with neither the interview nor the manual being published for almost three decades. On the other hand, the conceptual underpinnings of the M&G method have remained unchanged since 1996 when "Cannot Classify" was introduced as a new classification (Hesse, 1996; Main et al., 2008).
We think six issues limit the current use of the M&G-AAI. First, it has been very difficult for trained coders to achieve reliability on the M&G classifications. Second, it is unclear what the M&G-AAI assesses; this problem is compounded by the unavailability of a manual for review by researchers and clinicians who have not taken the training course. Such an investment of time, prior to viewing the instrument, exceeds the resources and motivation of many potentially interested professionals. Third, the M&G classificatory method is based on work in the cognitive sciences from the late 1970s. Knowledge of mental functioning derived since Bowlby completed his 1980 volume has not been incorporated into the classificatory process. Fourth, the adult clas- j I sificatory system is based on Ainsworth's infant classificatory system I and does not account fully for the complexity of adult behavior. Fifth, the Ainsworth system was developed from observations of middle-income, low-risk samples of American infants and their parents. Thus, it may not adequately reflect the range of diversity found in other cultures and in risk populations. Finally the notion of disorganization j lacks empirical support and conceptual coherence (Ruttcr, Kreppner, &
Introduction
Sonuga-Barke, 2009). With disappointing empirical findings from the M&G-AAI method, conceptual limitations to the ABC+D model, and lack of implications for treatment, we think it is time to move forward.
New Solutions
These six issues are addressed in this volume, thus, for the first time, making a classificatory method widely available for examination by the scientific and professional communities. As a consequence, both researchers and mental health professionals will be better able to evaluate this powerful, but complex, instrument. By anticipating the concerns listed above, the DMM has evolved to reflect practical solutions to the complex problems described so well by Bowlby.
First, by framing some of Main and Goldwyn's ideas in terms of functional mental processes, rather than sometimes vague or language-specific discourse markers, the coding guidelines have been made more explicit and achievement of reliability by trained coders has become more feasible. Second, by focusing on the role of danger in initiating psychological organization, the DMM clarifies the importance of assessing the adaptation of individuals to their contexts. Third, current empirical science, particularly emerging cognitive/neurological theory and research, has modified our understanding of the constructs upon which the M&G-AAI was based. The DMM-AAI incorporates new findings to generate new theory and methods that enhance and expand the scope of the AAI. As a consequence, the expanded classificatory system and DMM-AAI discourse method better fit the range of mental functioning seen in adults living in different cultures and conditions. A particular affvMISpr6f"lKe"ěčpaivdeci methodis that it addresses Row 1 disSrbě^híívicluals uselňTóřmation, rather than simply finding that they are not integrated or cannot be classified. Fourth, the developmental array of self- and progeny-protective strategies has been expanded through the preschool (end of the second year of life to approximately 5 years of age) and school years (approximately 6 years of age to puberty) to adolescence and on to adulthood, as envisioned by Ainsworth (Ainsworth, 1989; Crittenden & Ainsworth, 1989). Fifth, the DMM-AAI fits a wider range of adult functioning than the M&G-AAI does. Specifically, questions have been added to address imaged (i.e., perceptual) memory, anger from and toward parents, dangerous experiences and abuse, adolescence, adult attachment, sexuality, and use of one's parents as models for raising one's own children. Moreover, the interview
8
Assessing Adult Attachment
has been applied in substantial numbers to speakers who differ in either risk or culture or both. Finally, the notion of disorganization and the category of "Cannot Classify" have been eliminated.
Rather than describing what threatened people fail to do, the DMM-AAI gathers information about alternative ways of making meaning, specifically meanings tied to threat and the need to protect oneself and one's children from danger. We keep in mind that one of Bowlby's notions that excluded him from psychoanalytic theory was his belief in the reality of danger. In the DMM, the emphasis on the effects of exposure to danger, rather than the benefits of security, directs attention to those people who most need the advantage of protection, good theory, and compassionate treatment.
Problems with Psychiatric Diagnoses
Validating the new DMM classifications requires information external to the AAI. We have often used child protection and psychiatric status as defining variables. The intent, however, is not merely to accurately find differences betiveen normative and maladaptive categories. Instead, with the DMM-AAI we seek to add to the information generated by child protection or diagnostic authorities. That is, we want (1) to expand the array of possibilities from the four offered in the Main and Goldwyn method and (2) to do so in a way that is informative, above and beyond assessment of child abuse or neglect or psychiatric diagnosis, to mental health personnel.
Seeking validity of our differential classifications within troubled populations has led to using diagnoses as a criterion. This is problematic in many ways. Many people have noted that comorbidity is the norm, rather than the exception (Angold & Costello, 2009; Goldberg, 2010). Which diagnosis shall we use for validation—or more accurately, how will we cluster adults who vary in their array of diagnoses? This is tied to the issue of a growing number of "not elsewhere classified" diagnoses and increasingly complex diagnoses (Goldberg, 2010). Where shall we cluster these adults or do we just leave them out? Similarly, the two major diagnostic systems, the DSM and ICD, seem unable to agree on exactly which diagnoses there are or on what the basic premises for diagnoses should be (First, 2009; Frances, 2010). How can one validate against disputed criteria?
An important observation is that most diagnoses are made on the basis of a clinical interview and most clinical interviews are poor diag-nostically because they do not account for biases in human information
Assessing Adult Attachment^
Introduction
has been applied in substantial numbers to speakers who differ in either risk or culture or both. Finally, the notion of disorganization and the category of "Cannot Classify" have been eliminated.
Rather than describing what threatened people fail to do, the DMM-AAI gathers information about alternative ways of making meaning, specifically meanings tied to threat and the need to protect oneself and one's children from danger. We keep in mind that one of Bowlby's notions that excluded him from psychoanalytic theory was his belief in the reality of danger. In the DMM, the emphasis on the effects of exposure to danger, rather than the benefits of security, directs attention to those people who most need the advantage of protection, good theory, and compassionate treatment.
Problems with Psychiatric Diagnoses
Validating the new DMM classifications requires information external to the AAI. We have often used child protection and psychiatric status as defining variables. The intent, however, is not merely to accurately find differences betweeii normative and maladaptive categories. Instead, with the DMM-AAI we seek to add to the information generated by child protection or diagnostic authorities. That is, we want (1) to expand the array of possibilities from the four offered in the Main and Goldwyn method and (2) to do so in a way that is informative, above and beyond assessment of child abuse or neglect or psychiatric diagnosis, to mental health personnel.
Seeking validity of our differential classifications within troubled populations has led to using diagnoses as a criterion. This is problematic in many ways. Many people have noted that comorbidity is the norm, rather than the exception (Angold & Costello, 2009; Goldberg, 2010). Which diagnosis shall we use for validation-—or more accurately, how will we cluster adults who vary in their array of diagnoses? This is tied to the issue of a growing number of "not elsewhere classified" diagnoses and increasingly complex diagnoses (Goldberg, 2010). Where shall we cluster these adults or do we just leave them out? Similarly, the two major diagnostic systems, the DSM and ICD, seem unable to agree on exactly which diagnoses there are or on what the basic premises for diagnoses should be (First, 2009; Frances, 2010). How can one validate against disputed criteria?
An important observation is that most diagnoses are made on the basis of a clinical interview and most clinical interviews are poor diag-nostically because, they do not account for biases in human information
processing (Angold & Costello, 2009; Dozier & Lee, 1995). That is where the DMM-AAI shines. It is based on current understanding of biases in information processing (i.e., transformation of information based on past experience).
Our goals in using child protection and psychiatric diagnoses as validating criteria are to demonstrate that the DMM classifications have evidence of external validity by being significantly, but not perfectly, related to diagnostic information and also to bring conceptual coherence to the diagnostic process. We think the DMM can do the latter by focusing the grouping of individuals around the basic survival issues of self-protection, reproduction, and protection of progeny and by focusing on transformations of information as the means to organizing protection strategies. By framing maladaptation in terms of danger and sexuality, we highlight the functional aspects of maladaptive behavior, thus, working from a strengths approach rather than a deficit model. We also focus on what is most crucial in human life (safety and sex) and bring coherence to a somewhat confusing array of psychiatric diagnoses. Finally, our information processing approach brings the DMM in line with current scientific understanding of the basis for behavior and positions it for continued modification as the human sciences generate new understandings.
Our hope is that with a clearer and more scientifically based conceptual framework, tied to evolutionary processes, we will be able to reduce the number of outcome categories, compared to psychiatric diagnoses, while increasing the relevance of the categories for designating appropriate treatment. The Dynamic-Maturational Model of attachment and adaptation has, we think, passed "first muster"; that is, it has shown validity in terms of adaptation in a few, small comparative studies and relevance to treatment in several published case studies (see Chapter 15). We think it is time to make the process available to the scientific and professional communities both for application and to foster the input that will promote growth of the theory and methods.
INTRODUCTION AND ORIENTATION TO THE AAI
Adult attachment is complex and fascinating. Adults have the neurological maturity, experience, and need to formulate complex plans that require integration of conflicting information. Consequently, they can exercise greater flexibility in matching strategy to context adaptively
10
Assessing Adult Attachment
than is possible at younger ages. Complexity in response to complex conditions and change in the face of changing circumstances are the hallmarks of adult adaptation. In this book, we describe an array of adult protective strategies, consider the information processing that underlies the strategies, and outline a method for discerning the strategies in individual adults.
The Dynamic-Maturational Model of attachment and adaptation defines attachment as three entwined components: (1) relationships focused on protection and comfort, (2) patterns of mental processing of information about danger and sexual opportunity, and (3) strategies for self-protection, reproduction, and protection of progeny. Not surprisingly, however, the quality of adult relationships, the strategies, and the transformations of information underlying the strategies can be difficult to discern and correspondingly difficult to study. This is especially true because adults can dissemble, both knowingly and without awareness. As a consequence, their functioning is much less obvious and transparent to observers than is that of infants.
We present an approach to understanding and assessing adult attachment based on analysis of adults' spoken language about protection and comfort and, to a lesser extent, about sexuality and caregiving. The method of discourse analysis that we describe assumes that adults use both preconscious and explicit transformations of information and behavior to protect themselves, attract reproductive partners, and protect their children.
Attachment in Adulthood
Although infant attachment has received far more attention than adult attachment, adult attachment is more crucial to our survival as a species. A child who does not attach to his or her parents can survive through the efforts of the parents, but if adults do not attach to their infant, the infant's attachment will be useless. To promote the survival of the species, adults must protect themselves, have babies, and protect the babies to their reproductive maturity. Viewed this way, attachment refers to both oneself and one's progeny and includes sexuality. It also refers to the tie to one's reproductive partner.
Attachment as a relationship
Adults live in families marked by an array of attachment relationships: reciprocal attachment to a partner including sexual behavior and moti-
10
Assessing Adult Attachment
Introduction
11
than is possible at younger ages. Complexity in response to complex conditions and change in the face of changing circumstances are the hallmarks of adult adaptation. In this book, we describe an array of adult protective strategies, consider the information processing that underlies the strategies, and outline a method for discerning the strategies in individual adults. (
The Dynamic-Maturational Model of attachment and adaptation defines attachment as three entwined components: (1) relationships focused on protection and comfort, (2) patterns of mental processing of information about danger and sexual opportunity, and (3) strategies for self-protection, reproduction, and protection of progeny. Not surprisingly, however, the quality of adult relationships, the strategies, and the transformations of information underlying the strategics can be difficult to discern and correspondingly difficult to study. This is especially true because adults can dissemble, both knowingly and without awareness. As a consequence, their functioning is much less obvious and transparent to observers than is that of infants.
We present an approach to understanding and assessing adult attachment based on analysis of adults' spoken language about protection and comfort and, to a lesser extent, about sexuality and caregiving. The method of discourse analysis that we describe assumes that adults use both preconscious and explicit transformations of information and behavior to protect themselves, attract reproductive partners, and protect their children.
Attachment in Adulthood
Although infant attachment has received far more attention than adult attachment, adult attachment is more crucial to our survival as a species. A child who does not attach to his or her parents can survive through the efforts of the parents, but if adults do not attach to their infant, the infant's attachment will be useless. To promote the survival of the species, adults must protect themselves, have babies, and protect the babies to their reproductive maturity. Viewed this way, attachment
I refers to both oneself and one's progeny and includes sexuality. It also
' refers to the tie to one's reproductive partner.
Attachment as a relationship
Adults live in families marked by an array of attachment relationships: reciprocal attachment to a partner including sexual behavior and moti-
vation, changing attachment to parents, and being the attachment figure for children during the children's different stages of development. Adults must manage the integration of these different attachment relationships with other functions of the same relationships, for example, learning, working, and playing. This involves the maximum complexity of functioning in the life cycle. A hierarchically organized meta-model of relationships and functions is needed to manage the tasks successfully. When integration has not been accomplished and the context is unsupportive or threatening, behavior can become maladaptive. In that case, it is likely that self-protection, reproduction, or protection of progeny (in any combination) will be compromised and that oneself or one's family members will suffer psychological distress and, sometimes, physical harm.
Attachment as information processing
Information processing underlies all behavior. That is, all information is simply sensory stimulation that is given meaning by the brain; therefore, it is transformed. Adults transform incoming sensory stimulation in different ways to yield representations of the relation of self to context. These transformations enable adults to generate strategic plans of action even when information is missing, ambiguous, or false. This promotes adaptation and survival even as it makes one-to-one relations between observed behavior and psychological processes less direct. The behavior that one observes does not directly imply the underlying psychological process. Instead, the same behavior could result from different processes, and the same processes, under different contextual circumstances, could yield different behavior. Consequently, the meth- j ods employed to discern patterns of information processing must go ' beyond simply observing behavior.
Attachment as a self-protective and child-protective strategy
Representation creates a disposition to act. When there is threat or actual danger to the self, the disposition is a self-protective strategy. When there is sexual opportunity or threat to such opportunity, the representation disposes sexually motivated behavior (but not necessarily sexual behavior). When the threat is to one's child, the representation disposes child-protective strategy. Three points are crucial.
First, individuals often have more than one dispositional representa-
12
Assessing Adult Attachment
tion (DR). When they tend to rely on a particular pattern of information processing to guide behavior, that pattern reflects their typical self- or child-protective strategy. Of course, all individuals vary in strategy from time to time, but as threat increases, so does uniformity of strategy-Second, integration is needed to reconcile competing DRs. If the
usual strategy involves integrating past and current information, behavior is likely to vary from occasion to occasion, reflecting variation in the context. If the strategy does not usually include psychological integration, behavior will be more similar across occasions and reflect an overreliance on information about past experience. If the current con-| text is unlike that in the past, the resulting behavior will often be mal-' adaptive. Because integration is a slow cortical process, endangered people usually do not take time to reflect, thus, they have less opportunity to learn to integrate and are more likely to act precipitously in the future.
Third, for adults who are parents, the situation is even more complex. In the face of some threats, the self-protective and child-protective strategy may not only be different, but may also be—or seem to be-mutually exclusive. These situations are very difficult for parents.
Assessment of attachment in adulthood should reflect the full range of adult behavior and also the salient issues of adulthood (i.e., self-regulation regarding protection and reproductive opportunities, spousal functioning, and parental functioning). Further, it should address the complexity of information processing available to adults.
The Adult Attachment Interview: Its Original Intent and Current Applications
The Adult Attachment Interview was developed in the mid-1980s by Main and her colleagues to explore the relation between infants' quality of attachment at 12 months of age and their mothers' state of mind with regard to attachment (George et al., 1985; Main, Kaplan, & Cas-sidy, 1985). Since that time, it has been used in many studies, some of which replicate Main's findings and others of which explore additional uses of the AAI, particularly its application to clinical populations.
It has become clear that the AAI is relevant to adults other than mothers, to relationships other than the mother-infant relationship, and to functioning outside of the normal range. Indeed, it appears that the AAI is relevant to examination of the strategies used by adults to solve
I
12
Assessing Adult Attachment
Introduction
13
I
3
■i
tion (DR). When they tend to rely on a particular pattern of information "processing to guide behavior, that pattern reflects their typical self- or child-protective strategy. Of course, all individuals vary in strategy from time to time, but as threat increases, so does uniformity ol strategy- , . Second, integration is needed to.reconcile competing. DRs. If the
usual strategy involves intcr.r.iiir.e pas: ami current tnlonr.atior.. behavior is likely to vary from occasion to occasion, reflecting variation in the context. If the strategy does not usually include psychological integration, behavior will be more similar across occasions and reflect an overreliance on information about past experience. If the current context is unlike that in the past, the resulting behavior will often be maladaptive. Because integration is a slow cortical process, endangered people usually do not take time to reflect, thus, they have less opportunity to learn to integrate and are more likely to act precipitously in the future.
Third, for adults who are parents, the situation is even more complex. In the face of some threats, the self-protective and child-protective strategy may not only be different, but may also be—or seem to be— mutually exclusive. These situations are very difficult for parents.
Assessment of attachment in adulthood should reflect the full range of adult behavior and also the salient issues of adulthood (i.e., self-regulation regarding protection and reproductive opportunities, spousal functioning, and parental functioning). Further, it should address the complexity of information processing available to adults.
The Adult Attachment Interview: Its Original Intent and Current Applications
The Adult Attachment Interview was developed in the mid-1980s by Main and her colleagues to explore the relation between infants' quality of attachment at 12 months of age and their mothers' state of mind with regard to attachment (George et al., 1985; Main, Kaplan, & Cas-sidy, 1985). Since that time, it has been used in many studies, some of which replicate Main's findings and others of which explore additional uses of the AAI, particularly its application to clinical populations.
It has become clear that the AAI is relevant to adults other than mothers, to relationships other than the mother-infant relationship, and to functioning outside of the normal range. Indeed, it appears that the AAI is relevant to examination of the strategies used by adults to solve
problems in relationships and problems related to self-protection from danger. These issues, in turn, are central to marital relations, family functioning, individual mental health, and possibly even issues of professional or employment success. The focus on strategies for mentally ^JJrocessmg information relevant to dGngeFar^^xual opportunity is" •^.^peaaTtyjwnner^^
ior. The DMMirpproacTnoTTRTftXl goes beyondTVoTlTwith mothers to acTdress the full range of adult concerns regarding danger, reproduction, and threat to one's children.
Thus, the DMM-AAI has the potential to inform researchers regarding developmental processes occurring during adulthood (i.e., from early adulthood to old age) and to identify those processes that promote adaptation, given particular life circumstances. Further, the DMM-AAI can throw light on the types of mental functioning associated with maladaptation and mental/emotional dysfunction. Although up to now, the AAI has primarily been used to yield one of four categories (i.e., secure, dismissing, preoccupied, and unresolved/Cannot Classify), it could be analyzed in terms of mental and developmental processes. For this reason, the DMM approach to the AAI may be informative regarding the process of psychological treatment both in theory and in specific cases. In particular, with the addition of a wider range of strategies, the DMM-AAI has the potential to become an efficient assessment at the beginning of psychotherapy, a part of the treatment itself, and also an evaluative tool during and at the completion of therapy. These possible uses make understanding the potential of the DMM-AAI relevant to researchers, theorists, and psychotherapists.
Wliat Is the AAI?
The AAI is a semi-structured interview, usually lasting for 1 to IVi hours. It consists of a specified series of questions about childhood relationships with attachment figures (usually parents), together with interviewer-generated follow-up questions. Although it is usually treated as a single entity, the AAI has four distinct components:
1. an embedded theoretical perspective;
2. a specific set of interview questions that query, in a systematic manner, about childhood experiences and adult perspectives on these;
3. a classificatorv method; and
14
Assessing Adult Attachment
4.    a classificatory system derived from Ainsworth's system for infants. (Ainsworth, 1979)
These four components are usually referred to, as a whole, as the Adult Attachment Interview. Nevertheless, they can be considered separately.
Theory
Various theoretical perspectives on attachment can be applied to the interpretation of the AAI. Main and her colleagues developed the AAI based on a version of attachment theory that assumed that (1) by adulthood most adults had a single representation of attachment relationships, (2) this relationship reflected one of the Ainsworth patterns of infant attachment, (3) these patterns were transmitted from mother to child across generations, and (4) frightening circumstances disrupted the organizational process, leading to a state of disorganization in infancy or lack of resolution of the frightening circumstances in adulthood (Main & Hesse, 1990; Main & Goldwyn, 1984; Main et al, 1985).
In the Dynamic-Maturational Model of attachment and adaptation (Crittenden, 1994, 1995, 2000a, 2000b, 2000c, 2008) none of these four assumptions are made. To the contrary, the DMM approach to attachment theory presumes that adults have multiple dispositional representations, each unique to the information processing underlying it. Second, the array of strategies is developmentally expanded from its roots in infancy, with endangered individuals most often using the later developing and more complex strategies (Crittenden & Ainsworth, 1989).
Third, it is understood that each individual constructs his or her own dispositional representation from his or her own experience. Sometimes this will reflect similarities to the parent's dispositional representation, but, especially in cases of parental disturbance or inadequacy, children will often organize the opposite strategy from the parent. Moreover, in families with several children, older children's strategies are expected to influence the strategies of younger children. For example, if an older child takes care of the parent in order to be protected, a younger child may find it more adaptive to be demanding of the parent. In such a case, the older child might use a different strategy from the mother and the younger child might use a different strategy than either the mother or the sibling.
14
Assessing Adult Attachment
Introduction
15
4.    a classificatory system derived from Ainsworth's system for infants. (Ainsworth, 1979)
These four components are usually referred to, as a whole, as the Adult Attachment Interview. Nevertheless, they can be considered separately.
Theory
Various theoretical perspectives on attachment can be applied to the interpretation of the AAI. Main and her colleagues developed the AAI based on a version of attachment theory that assumed that (1) by adulthood most adults had a single representation of attachment relationships, (2) this relationship reflected one of the Ainsworth patterns of infant attachment, (3) these patterns were transmitted from mother to child across generations, and (4) frightening circumstances disrupted the organizational process, leading to a state of disorganization in infancy or lack of resolution of the frightening circumstances in adulthood (Main & Hesse, 1990; Main & Goldwyn, 1984; Main et al., 1985).
In the Dynamic-Maturational Model of attachment and adaptation (Crittenden, 1994, 1995, 2000a, 2000b, 2000c, 2008) none of these four assumptions are made. To the contrary, the DMM approach to attachment theory presumes that adults have multiple dispositional representations, each unique to the information processing underlying it. Second, the array of strategies is developmentally expanded from its roots in infancy, with endangered individuals most often using the later developing and more complex strategies (Crittenden & Ainsworth, 1989). ' Third, it is understood that each individual constructs his or her mvn dispositional representation from his or her own experience. Sometimes this will reflect similarities to the parent's dispositional representation, but, especially in cases of parental disturbance or inadequacy, children will often organize the opposite strategy from the parent. Moreover, in families with several children, older children's strategies are expected to influence the strategies of younger children. For example, if an older child takes care of the parent in order to be protected, a younger child may find it more adaptive to be demanding of the parent. In such a case, the older child might use a different strategy from the mother and the younger child might use a different strategy than either the mother or the sibling.
Finally, exposure to danger is assumed to be the essential condition that elicits attachment behavior, and, across repeated experiences, leads to organized self-protective strategies. Furthermore, adapting to dangerous circumstances often takes more skill than infants have. For this reason, infants are extremely dependent upon parents for protection; a central function of their development is learning to identify danger and protect themselves from it. When dangers are unusually prevalent, difficult to discern, or hard to avoid, a threatened person will need complex psychological processes and sometimes misleading or deceptive behavior to stay safe. In the DMM, danger leads to both organization and more complex organization than does safety.
A crucial aspect of DMM theory is that it poses two basic psychological transformations of information—cognition and affect—which constitute, thebasic input to repre transformed informa-
tion is combined using two basic psychological processes: association (putting together) and disassociation (keeping apart). These processes then generate dispositional representations. Together the transformations and processes generate the three basic patterns of attachment (i.e., Ainsworth's ABC patterns of attachment). Type A is a "cognitively" organized strategy that disassociates affect, Type C is "affectively" organized with association of sometimes unrelated affective information, and Type B integrates both transformations of information using both associative and disassociative processes. Within Type B, some patterns are more cognitive (Bl-2) whereas others are more affective (B4-5). Thus, the model itself involves two basic transformations and two basic processes that are combined and transformed after infancy to generate more complex strategies.
The terms classification, pattern, and strategy each have a unique, albeit related, meaning in the DMM. Classification refers to the outcome of coding and is constrained with reference to reality the way data are always constrained. Pattern refers to the clusters of constructs used by speakers; these tend to co-occur (rather than operating independently) and are the basis for classification. Strategy is what the classification is meant to identify. That is, we are Interested in strategies, but use a patterning of constructs to define a classification as the proxy for the strategy.
These four differences between the work of Main and colleagues and the DMM in underlying theory, together with the notions of cognition and affect, yield a new, developmentally differentiated, and complex way of thinking about the output of an AAI.
16
Assessing Adult Attachment
The interview
A different interview could be used to assess attachment or self-protective strategies; the interview could then be analyzed using the discourse analysis procedure (e.g., see Crittenden, Partridge, & Claussen, 1991 for a parenting interview analyzed with Main and Goldwyn's method). In addition, the original George, Kaplan, and Main interview could be revised or modified (George et al., 1985,1996). Indeed, we recommend modification of this interview for clinical applications and for exploring more fully developmental pathways and a wider range of interpersonal strategies. Although transcripts using the original and revised versions of the AAI can be analyzed and classified using the DMM method described in this volume, the modified AAI (Crittenden, 2007) provides the detail needed in clinical cases.
A particular feature of the interview by George and colleagues is its replication in discourse of the process of regulating stress so as to elicit the individual's self-protective strategy that is embedded in Ains-worth's Strange Situation Procedure (SSP). That is, the questions (1) begin comfortably, (2) become a bit more challenging, but still under the control of the speaker, then (3) shift to minor forms of threat introduced by the interviewer and then to (4) fairly substantial threats, including (5) loss of attachment figures. The interview draws to a close by asking (6) a graduated series of integrative questions. In the Crittenden modification, the array of dangers is broader and the closing is (7) returned to the control of the speaker. In both versions of the AAI, it is the discourse, not the content of the answers to the questions, that is crucial............................
Often, interest is expressed in selecting some of the AAI questions and delivering them to clients outside of the context of the interview. This seems to imply that the power of the AAI lies in the questions themselves. Excerpting questions from the AAI and popularizing them outside the AAI format dilutes the power of the interview to surprise the mind of the speaker and transforms the questions to ordinary questions, forcing equally ordinary, content-based interpretation.
Instead, the AAI process of regulating stress and probing different representations depends on the order of the questions and the interpersonal context in which they are delivered. The interpersonal process can be seen as a conversation with a stranger about sensitive and personal topics, addressed in surprising ways that require new, on-the-spot thinking. This unusual and special conversation with a com-
16
Assessing Adult Attachment
The interview
A different interview could be used to assess attachment or self-protective strategies; the interview could then be analyzed using the discourse analysis procedure (e.g., see Crittenden, Partridge, & Claussen, 1991 for a parenting interview analyzed with Main and Goldwyn's method). In addition, the original George, Kaplan, and Main interview could be revised or modified (George et al, 1985,1996). Indeed, we recommend modification of this interview for clinical applications and for exploring more fully developmental pathways and a wider range of interpersonal strategies. Although transcripts using the original and revised versions of the AAI can be analyzed and classified using the DMM method described in this volume, the modified AAI (Crittenden, 2007) provides the detail needed in clinical cases.
A particular feature of the interview by George and colleagues is its replication in discourse of the process of regulating stress so as to elicit the individual's self-protective strategy that is embedded in Ains-worth's Strange Situation Procedure (SSP). That is, the questions (1) begin comfortably, (2) become a bit more challenging, but still under the control of the speaker, then (3) shift to minor forms of threat introduced by the interviewer and then to (4) fairly substantial threats, including (5) loss of attachment figures. The interview draws to a close by asking (6) a graduated series of integrative questions. In the Crittenden modification, the array of dangers is broader and the closing is (7) returned to the control of the speaker. In both versions of the AAI,it... is the discourse, not the content of the answers to the questions, that is crucial.
Often, interest is expressed in selecting some of the AAI questions and delivering them to clients outside of the context of the interview. This seems to imply that the power of the AAI lies in the questions themselves. Excerpting questions from the AAI and popularizing them outside the AAI format dilutes the power of the interview to surprise the mind of the speaker and transforms the questions to ordinary questions, forcing equally ordinary, content-based interpretation.
Instead, the AAI process of regulating stress and probing different • representations depends on the order of the questions and the interpersonal context in which they are delivered. The interpersonal process can be seen as a conversation with a stranger about sensitive and personal topics, addressed in surprising ways that require new, on-the-spot thinking. This unusual and special conversation with a com-
_Introduction___17
passionately available interviewer sets up a pathway to questions that can trigger integrative functioning, setting the mind of the speaker in motion in potentially new ways.
In this special conversation, the interviewer's role is quite precise. Its function is to create an opportunity for the speaker to be clear to himself and to the interviewer. Speakers seem intuitively aware of this. Every time they ask "D'you know what I mean?" or "You know?" or "Am I making sense?" speakers are asking whether the function of communication is being fulfilled. Ironically, interviewers are extremely likely to choose this moment to be reassuring as opposed to informative. It is our observation, from reading several thousand AAIs, that the less clear the speaker is, the more reassuring the interviewer is. "Yes, I understand." "Uh huh." "Right." And speakers learn nothing about themselves or the process of communication. It is far better for an interviewer to listen more attentively, asking herself repeatedly whether she did indeed understand the speaker and, when asked by the speaker, had answered in a way that promoted clarity. "Yes, somewhat, but maybe you could say a bit more." "Not quite, I didn't quite understand ... [quoting the speaker's words]." Although there are many skills needed to give a proper AAI and these are offered in the AAI training course, this way of communicating stands out because il is so directly connected to the interpersonal process of revealing meaning.
The classificatory method
Main and Goldwyn (1984, 1994) constructed the original discourse analysis for the AAI. Other classificatory procedures could be applied to it, as has been done by Grossmann, Fremmer-Bombik, Rudolph, and Grossmann (1988). This volume describes a method that has been developed over almost two decades. Its roots are in the groundbreaking work of Main and Goldwyn, augmented by ideas from Bandler and Grinder (1975), Bateson (1972), Grinder and Bandler (1975), and Watzlawick, Beavin and Jackson (1967), as well as others. Many of this method's constructs have been drawn from AAIs themselves when reconciling the discourse with the history of the speaker-required tools (i.e., transformations of information and attributions of meaning) that were not in the method at that point. The process has been developmental: using existing tools to meet new challenges that produce the discrepancy of a mismatch and call for integrative thinking that yields a new construct or new organization of existing constructs. The new
18
Assessing Adult Attachment
tool is then applied to future transcripts, sometimes yielding an unexpected mismatch and initiating the whole integrative process again.
Unlike the Main and Goldwyn method that was developed on a sample for upper middle-class American parents, the DMM method was honed from AAIs from more than 20 countries, including AAIs with normative adults, adults in outpatient treatment, adults in psychiatric hospitals, and prisoners in correctional facilities. The DMM approach was first published in 1999 but has continued to undergo refinement up to now. We expect it to continue to change as it becomes more widely applied
The classificatory system
Just as different versions of attachment theory, of the AAI, and of discourse analysis can be used, the classificatory system can vary. It could retain the Ainsworth patterns of infancy or be revised to reflect adult organizations that coalesce only after infancy (see Crittenden & Ainsworth, 1989). Describing such organizations is at the heart of the DMM method. From the three Ainsworth ABC patterns of infant attachment, an expanded DMM set of classifications is offered to address organizations of thought and behavior beyond the range described by Ainsworth. These classifications identify strategies that infants cannot yet organize. The DMM set of classifications was derived from theory (from both Bowlby's and Crittenden's theories) and observed in the transcripts of adults from many cultures and conditions in life, including, in particular, adults who were endangered in their childhood or who display various forms of psychopathology in their adulthood. The full array of Dynamic-Maturational Model patterns is displayed in Figure 1.1.
The major classifications reflect Ainsworth's ABC trichotomy and are expanded here to include patterns observed only after infancy. In Main and Goldwyn's model, Ainsworth's ABC patterns were relabeled as dismissing (Ds = A), free /autonomous (F = B), and preoccupied (E = C). In deference to Ainsworth and in order to clarify developmental relations, the original ABC terminology is retained in the Dynamic-Maturational Model.
The major ABC classifications are described below. In Chapter 2, these descriptions are extended to a brief overview of each of the classifications. In Chapter 3, the basis in information processing for the patterns is sketched. In Chapter 4, the array of constructs used in this
18
Assessing Adult Attachment
Introduction
tool is then applied to future transcripts, sometimes yielding an unexpected mismatch and initiating the whole integrative process again.
Unlike the Main and Goldwyn method that was developed on a sample for upper middle-class American parents, the DMM method was honed from AAIs from more than 20 countries, including AAIs with normative adults, adults in outpatient treatment, adults in psychiatric hospitals, and prisoners in correctional facilities. The DMM approach was first published in 1999 but has continued to undergo refinement up to now. We expect it to continue to change as it becomes more widely applied
the classif1catory system
Just as different versions of attachment theory, of the AAI, and of discourse analysis can be used, the classificatory system can vary. It could retain the Ainsworth patterns of infancy or be revised to reflect adult organizations that coalesce only after infancy (see Crittenden & Ainsworth, 1989). Describing such organizations is at the heart of the DMM method. From the three Ainsworth ABC patterns of infant attachment, an expanded DMM set of classifications is offered to address organizations of thought and behavior beyond the range described by Ainsworth. These classifications identify strategies that infants cannot yet organize. The DMM set of classifications was derived from theory (from both Bowlby's and Crittenden's theories) and observed in the transcripts of adults from many cultures and conditions in life, including, in particular, adults who were endangered in their childhood or who display various forms of psychopathology in their adulthood. The full array of Dynamic-Maturational Model patterns is displayed in Figure 1.1.
The major classifications reflect Ainsworth's ABC trichotomy and are expanded here to include patterns observed only after infancy. In Main and Goldwyn's model, Ainsworth's ABC patterns were relabeled as dismissing (Ds = A), free/autonomous (F = B), and preoccupied (E = C). In deference to Ainsworth and in order to clarify developmental relations, the original ABC terminology is retained in the Dynamic-Maturational Model.
The major ABC classifications are described below. In Chapter 2, these descriptions are extended to a brief overview of each of the classifications. In Chapter 3, the basis in information processing for the patterns is sketched. In Chapter 4, the array of constructs used in this
Figure 1.1 The Dynamic-Maturational Model of attachment strategies in adulthood. Integrated True Information
True Cognil ;in
Distorted Cognition Omitted Negative After
I a3-*
False Positive /
\ ccinpfum
Denied Negative Aff Delusional Cognitu,,
I Me Negative Affect
listorted Negative Affect ii.™i-sX &0mitted Cognition
as-6 * co!it|mlb:.£!y \prO(SiEC.E.^ \sbh-Rcti3n!/A7-S
False Cognition
ß-W c <
Denied True Cognition; \  Delusional Affect
Integrated Transformed Information--
method is described, including the distinction between history and discourse. In Chapters 5 to 12, each classification is described in detail in terms of discourse and the function of the discourse, and each pattern is also described in terms of the mental strategies evident in the discourse and of the developmental history associated with the pattern. The process of this manual, in other words, is one of increasing differentiation of theory and classification through a steady accretion of concepts and detail.
THE MAJOR PATTERNS OF ATTACHMENT
Type B: Balanced Attachment
When attachment figures were both protective and comforting, adults are usually balanced with regard to processing information and managing relationships. In some cases, however, even adults who were endangered as children are able to achieve mental and emotional clarity with regard to their experiences and to function, in adulthood, in an "earned" (i.e., reorganized) balanced manner. In such cases, it is not necessary that the adult be "secure" (i.e., safe from danger or worry about danger). Indeed, security and comfort may be impossible to
20
Assessing Adult Attachment
achieve in outer reality because of forces outside of the control of the individual (e.g., conditions of poverty or war). Nevertheless, all adults have a possibility of psychological balance.
Speakers who are classified as balanced (B) tend to describe their childhood experiences using both sources of information: (1) cognition (i.e., temporal and causal order, realistically identifying complex causal relations) and (2) affect, including both positive and negative feeling states. They also use both associative and disassociative processes in a judicious manner. In addition, balanced speakers describe their relationships with their parents in terms of varied attributes and provide evidence (in the form of recalled episodes) to support these generalizations. As adults, they are able to look back and recall their own childhood perspectives, construct an understanding of their parents' probable perspectives, and describe their current understanding of events. This current understanding contains conclusions that are complex in that they (1) acknowledge that people and relationships change over time, (2) portray the self and attachment figures as varying in behavior and being less than perfect (or unredeemingly terrible), (3) reflect the interactive effects of self and others, and (4) differentiate appearance from reality.
Type A: Dismissing of the Self Types Al-2: Dismissing attachment in the context
of physical safety (i.e., the low-numbered
Type A classifications)
When attachment figures fail to protect or comfort children, defensive processes may be used. If the child is actually safe, but only partially comforted by attachment figures, only a mild distortion is expected (i.e., some mild disassociation of positive and negative characteristics). When, in addition, lack of comfort is accompanied by rejection of the child's unnecessary attachment behavior (e.g., display of anger, fear, and desire for comfort), a simple defense against negative affect is often used. In this case, the good and bad qualities of the parent are split and only the good is acknowledged and display of negative affect is inhibited (i.e., disassociated).
Consequently, in the AAI, the Al-2 speaker describes the rejecting parent in positive, idealized terms (whereas Type B speakers are accurate with regard to both desired and unpleasant qualities of their par-
20
Assessing Adult Attachment
Introduction
21
achieve in outer reality because of forces outside of the control of the individual (e.g., conditions of poverty or war). Nevertheless, all adults have a possibility of psychological balance.
Speakers who are classified as balanced (B) tend to describe their childhood experiences using both sources of information: (1) cognition (i.e., temporal and causal order, realistically identifying complex causal relations) and (2) affect, including both positive and negative feeling states. They also use both associative and disassociative processes in a judicious manner. In addition, balanced speakers describe their relationships with their parents in terms of varied attributes and provide evidence (in the form of recalled episodes) to support these generalizations. As adults, they are able to look back and recall their own childhood perspectives, construct an understanding of their parents' probable perspectives, and describe their current understanding of events. This current understanding contains conclusions that are complex in that they (1) acknowledge that people and relationships change over time, (2) portray the self and attachment figures as varying in behavior and being less than perfect (or unredeemingly terrible), (3) reflect the interactive effects of self and others, and (4) differentiate appearance from reality.
Type A: Dismissing of the Self Types Al-2: Dismissing attachment in the context
of physical safety (i.e., the low-numbered
Type A classifications)
When attachment figures fail to protect or comfort children, defensive processes may be used. If the child is actually safe, but only partially comforted by attachment figures, only a mild distortion is expected (i.e., some mild disassociation of positive and negative characteristics). When, in addition, lack of comfort is accompanied by rejection of the child's unnecessary attachment behavior (e.g., display of anger, fear, and desire for comfort), a simple defense against negative affect is often used. In this case, the good and bad qualities of the parent are split and only the good is acknowledged and display of negative affect is inhibited (i.e., disassociated).
Consequently, in the AAI, the Al-2 speaker describes the rejecting parent in positive, idealized terms (whereas Type B speakers are accurate with regard to both desired and unpleasant qualities of their par-
ents). In order to maintain this distortion, memories of instances in which the child was safe, but not comforted, are set aside and not recounted during the AAI, even when they are directly requested. Alternatively, episodes in which desired comfort was not given are truncated and the lack of comfort may be dismissed as trivial. Expression of affect is largely absent from interviews of speakers classified as Types Al-2. The Al and A2 patterns are rarely associated with serious, life-threatening physical danger. To the contrary, the danger is the psychological \ discomfort of having attachment behavior rejected by the attachment I figure (when the child was actually safe). i
Types A3-8: Dismissing in the context of danger (i.e., the high-numbered, compulsive Type A classifications)
When parents are a source of danger or fail to protect children from danger and if the danger is predictable and preventable, children learn to do what is necessary to increase their safety.
In such cases, the threats are generally well remembered and recounted as episodes. Therefore, idealization is not possible; instead, in the AAI, adults who use a Type A strategy make excuses for their parents (exonerate them), take the parents' perspectives, and deny their own attachment needs and feelings, both as children and also as adults. Often there is some form of compulsion. There may be compulsive _compliance if violence was the threat (A4), compulsive caregiving (role reversal) if neglect was the threat (A3), compulsive self-reliance if nothing except escape protected the child from the parents' dangerous behavior (A6), or compulsive seeking of intimacy with strangers (A5). Compulsively self-reliant children protect themselves by isolating themselves from dangerous parents, but, to do so, they give up access to parental protection and to their own feelings. Frequently, the isola-, tion of the A6 strategy is associated with promiscuity, including sexual'. promiscuity, in casual relationships (i.e., A5). j
The two most extreme strategies result from serious endangerment, beginning early in life and extending across developmental periods. In the case of some contact with attachment figures, adults may delusion-ally idealize dangerous figures, denying their negative experiences and delusionally transforming them into positive ones, thus protecting themselves in recall from danger in childhood (A7). Others who had no figure to turn to (or a series of changing figures) develop an externally constructed self (A8).
22
Assessing Adult Attachment
The high-numbered strategies are associated with increased rates of psychological distress and are observed in greater proportion in the transcripts of adults in psychotherapy than in the normative population (in relatively safe societies).
Type C: Preoccupied with the Self Types Cl-2: Preoccupied with relationships in the
context of unpredictable care (i.e., the low-numbered
Type C classifications)
Type C speakers attempt to coerce comfort and protection from attachment figures by using exaggerated and alternating displays of anger and desire for comfort with some fear. In most cases, attachment figures were affectively available, but their unpredictable responses provided children with no confidence that they would be protected.
Such attachment figures are inconsistent, but vary regarding protection from danger. Because there is little predictability, children do not learn to attend to temporal order as a reliable source of information and are unable to draw sound causal conclusions. They also do not learn to inhibit display of negative affect. In such cases, children experience their parents as indecisively loving and are unable to explain why they continue to feel uneasy. Their response is (1) to become dependent, angry, or fearful and (2) to wonder whether, in the event of danger, the attachment figure might fail to protect them or whether they might have to rely on themselves. Their strategy is to focus on (1) feelings that signal danger and (2) threatening events and details about such events that could possibly enhance identification of threat in the future. In the AAI, mildly self-preoccupied speakers use an associative process when recalling the critical fragments of past experiences, slipping easily from past to present and back again (blurring the boundaries of time), confusing people (blurring the boundaries between people), failing to draw reasoned conclusions about their childhoods (failing to make accurate causal attributions because they have disassociated cognitive information), and showing affective arousal (e.g., giggling). In addition, they often seek confirmation that interviewers are paying attention. Temporal and causal ordering, thus, are minimized, whereas affect is exaggerated.
The CI and C2 subpatterns are rarely associated with physical danger. To the contrary, the danger is the psychological discomfort of being uncertain when and how attachment figures would respond.
22
Assessing Adult Attachment
ntroduction
23
The high-numbered strategies are associated with increased rates of psychological distress and are observed in greater proportion in the transcripts of adults in psychotherapy than in the normative population (in relatively safe societies).
Type C: Preoccupied with the Self Types Cl-2: Preoccupied with relationships in the
context of unpredictable care (i.e., the low-numbered
Type C classifications)
Type C speakers attempt to coerce comfort and protection from attachment figures by using exaggerated and alternating displays of anger and desire for comfort with some fear. In most cases, attachment figures were affectively available, but their unpredictable responses provided children with no confidence that they would be protected.
Such attachment figures are inconsistent but vary regarding protection from danger. Because there is little predictability, children do not learn to attend to temporal order as a reliable source of information and are unable to draw sound causal conclusions. They also do not learn to inhibit display of negative affect. In such cases, children experience their parents as indecisively loving and are unable to explain why they continue to feel uneasy. Their response is (1) to become dependent, angry, or fearful and (2) to wonder whether, in the event of danger, the attachment figure might fail to protect them or whether they might have to rely on themselves. Their strategy is to focus on (1) feelings that signal danger and (2) threatening events and details about such events that could possibly enhance identification of threat in the future. In the AAI, mildly self-preoccupied speakers use an associative process when recalling the critical fragments of past experiences, slipping easily from past to present and back again (blurring the boundaries of time), confusing people (blurring the boundaries between people), failing to draw reasoned conclusions about their childhoods (failing to make accurate causal attributions because they have disassociated cognitive information), and showing affective arousal (e.g., giggling). In addition, they often seek confirmation that interviewers are paying attention. Temporal and causal ordering, thus, are minimized, whereas affect is exaggerated.
The CI and C2 subpatterns are rarely associated with physical danger. To the contrary, the danger is the psychological discomfort of being uncertain when and how attachment figures would respond.
Types C3-8: Preoccupied with relationships in the context of danger (i.e., the high-numbered, obsessive Type C classifications)
In more severe cases, dangerous events occurred and parents did not protect or comfort children. Some such children become very anxious and escalate their efforts to elicit protective responses from the parents (C3-4). Depending upon the extent to which the parent deceives the child about the danger, the child may become more (C7-8) or less (C5-6) suspicious of feeling comfortable.
In the AAI, uncertainty regarding temporal contingencies appears as the lack of logical/rational conclusions, plus irrational, magical or deceptive conclusions (i.e., disassociated cognition and transformed cognition). Distortions of affect are displayed as intense affect of one sort (e.g., anger) that is present in the interview nonverbally or in affectively intense language, while display of other incompatible affects (e.g., fear and desire for comfort) is inhibited, then the displays are reversed. For example, intense anger may be displayed without evidence of fear or desire for comfort (C5). In most cases, the speaker appears unable to tell his or her story alone and the interviewer finds himself or herself subtly pulled into the interview as an ally or opponent of the speaker and, thus, into the family conflict. Like the compulsive classifications, these high-numbered Type C classifications are associated with psy-chopathology (in relatively safe societies).
WHAT DOES THE AAI ASSESS?
The Adult Attachment Interview appears to be a straightforward interview about childhood experiences. The first result is a narrative that provides content on the history of the speaker and which is relevant to experience with attachment figures and with protection from danger. There is no way to know, within the interview, if what the speaker says is accurate about his or her history: It is, nevertheless, information
about :i-e speaker's perspective. ........ " —
When administered and interpreted properly, the AAI addresses the , question of how the speaker uses his or her perspective on past experi-i ence to predict when and where danger is likely to occur in the future | and how best to prepare for it.Th^includes_such issues as_how prob-jj able JGHiger is pCTCMvea^ryWr'ivhg ce(n be trusted (and under what > conditions), and what one canjio_to_protect oneself. Put another way,
Assessing Adult Attachment
the Adult Attachment Interview explores what the speaker has learned from past experience that is applied to the future and what he or she believes is specific and unique to the past (and therefore not relevant to future conditions). In particular, the AAI considers how information is transformed to give it meaning in terms of future protection of self, attachment figures, and progeny.
Framed this way, the AAI assesses more than the individual's state of mind in regard to attachment. It assesses how the mind processes in- _ formation, in terms of how different dispositional repre&entatiaas are activated in response to specific stimulation, how integration draws these different representations together, and how complete and usable are the products of integrative processing.
Thus the AAI assesses the pattern of attachment of the speaker both in terms of holvlilll^Tsh^ dangerous topics
~ aliiarTn^r^erlyTng basis of this behavior in mentarprocessing or7nfoi> , malioh7This~assessment is"r5ased on a coder's interpretation of how the
,jk ------- —■
behavioral evidence is patterned strategically. Like many projective ] tests, the AAI permits fine-grained observation of mental and interper- I sonal processes that often are not in the respondents' conscious awareness. The AAI can collect evidence of behaviors based on implicit, preconscious mental representations. Like more objective measures, it also permits specification of what is being assessed, why it is important, and how it contributes to the summary result. Behavior is recorded by transcribing the interview, its function is assigned to specific categories called memory systems (see Chapter 3), and these contribute in specified ways to the overall strategic pattern.
Classification of an AAI takes the form of ascribing (1) a basic attachment strategy, with the possibility of modification by (2) unresolved losses or traumas, or (3) a relatively pervasive condition, such as depression. Because each of these components of a classification is based on specified instances of behavior assigned to specific memory systems, the reasons for assignment to a classification (with its functional implications) are clear to appropriately trained coders.
FOR WHOM IS AN AAI CLASSIFICATION OF INTEREST?
Viewed as a functional description of how a person is likely to behave when threatened, an AAI classification can be used as descriptive data
24
Assessing Adult Attachment
Introduction
25
the Adult Attachment Interview explores what the speaker has learned from past experience that is applied to the future and what he or she believes is. specific and unique to the past (and therefore not relevant to future conditions). In particular, the A AI considers how information is transformed to give it meaning in terms of future protection of self, attachment figures, and progeny.
Framed this way, the AAI assesses more than the individual's state of mind in regard to attachment. It assesses how the mind processes in- . formation, in terms of how different di_sjwsiüun^ are activated in response to specific sUmiüation, how integration draws these different representations together, and how complete and usable are the products olintegrative processing.'.'
Tims the AAI assesses the pattern of attachment of the speaker both in tenrislJhövv'TTe TSr^erJgKäveTwMi"considering dangerous topics
ImcTthlTulrirffl .
"matiör\7TEs~Mses^e of how the
behavioral evidence is patterned strategically. Like many projective tests, the AAI permits fine-grained observation of mental and interpersonal processes that often are not in the respondents' conscious awareness) The AAI can collect evidence of behaviors based on implicit, preconscious mental representations. Like more objective measures, it also permits specification of what is being assessed, why it is important, and how it contributes to the summary result. Behavior is recorded by transcribing the interview, its function is assigned to specific categories called memory systems (see Chapter 3), and these contribute in specified ways to the overall strategic pattern.
Classification of an AAI takes the form of ascribing (1) a basic attachment strategy, with the possibility of modification by (2) unresolved losses or traumas, or (3) a relatively pervasive condition, such as depression. Because each of these components of a classification is based on specified instances of behavior assigned to specific memory systems, the reasons for assignment to a classification (with its functional implications) are clear to appropriately trained coders.
-.~r re?.eSS^ ffiA as. a diagnostic assessment in clinical settings where the main interest is planning interventions. More recently, its suitability for settings where a judgment on the adaptive qualities of an individual is needed (e.g., parenting assessment in child protection, forensic settings) is being explored.
For these reasons, the AAI has attracted the interest of researchers studying human development, theorists constructing models of human adaptation, clinicians working with adults with psychological disorders or maltreating parents, and experts testifying for courts.
PLAN FOR THE BOOK
This volume contains general guidelines for classifying normative transcripts that correspond to Main and Goldwyn's basic AAI system. In addition, it contains guidelines for an expansion of this system in the direction of wider adult variation, including both cultural variation and maladaptation, from child protection to psychopathology and criminal behavior. However, this is not a manual that can replace direct training on AAI transcripts and the feedback that accompanies training. Nevertheless, we hope that the availability of a guide will encourage others to explore the instrument, become trained and reliable on the procedure, contribute to the growing pool of findings from the AAI, and, over time, adapt the interview and discourse analysis further to varied circumstances, populations, and applications.
The book is divided into three parts. Part I addresses the theory, history, and concepts that are necessary for understanding the Adult Attachment Interview. Part II provides the details of each classification. Part III concludes the volume with a guide to the applications of the DMM method for the AAI, an overview of its possible uses, a review of the validity studies available on the DMM-AAI, and future directions for using the AAI to further our understanding of human adaptation and ways to reduce psychological suffering.
FOR WHOM IS AN AAI CLASSIFICATION OF INTEREST?
Viewed as a functional description of how a person is likely to behave when threatened, an AAI classification can be used as descriptive data