Britisb Journal of Medical Psycbology (1997), 70, 249-263 Printed in Great Britain 249

0 1997 The British Psychological Society

Adult attachment processes:

Individual and couple perspectives
Kim Bartholomew“
Department Of PJyChOhgY, Simon Fraser University, Bumaby, BC, VSAlS6 C a n a b

This paper overviews a new approach to understanding the range of difficulties

experienced in close attachment relationships in adulthood. Drawing on the work of
Bowlby, four prototypic adult attachment patterns are defined in terms of the
intersection of two underlying dimensions, the positivity of the person’s self-image
and the positivity of the person’s image of others (Bartholomew, 1990; Bartholomew &
Horowitz, 1991).The distinct interpersonal difficulties associated with each attachment
pattern are described. Findings are presented indicating that individual differences in
attachment have implications for the quality of adults’ romantic relationships, and that
attachment theory may be helpful in understanding violent spousal relationships. Five
current issues in the study of adult attachment are addressed: the stability of attachment
patterns, the associations between attachment and general personality factors, the
relative merits of categorical and prototype assessments of attachment, the identification
of multiple attachments in adulthood, and the specificity of adult attachment patterns.
It is suggested that the four-category model of adult attachment is especially sensitive to
the range and complexity of attachment-related difficulties experienced in adulthood.

My interest in adult attachment stems from an obvious, but perplexing, observation. On

the one hand, people are highly motivated to have satisfying intimate relationships with
their family members, friends and sexual partners. And yet, despite this motivation,
many people have difficulties in establishing and maintaining these relationships. For
example, consider Elizabeth Taylor. A few years ago she got married for the eighth time,
and, even more astounding, she cried during the ceremony-presumably with happiness,
not sadness. As I’m sure all of us would have predicted, that relationship is now over. The
question is: How can we explain this sort of behaviour?
I will be giving an overview of one approach to understanding the range of difficulties
experienced in close attachment relationships in adulthood. First, I will describe the four-
category model of adult attachment I have been working with (Bartholomew, 1990;
Bartholomew & Horowitz, 1991). I will then present some findings from studies
applying this model to couple relationships. Finally, I will discuss some general issues
that have arisen in adult attachment research.
Attachment theory, to quote Bowlby, is ‘a way of conceptualizing the propensity of
human beings to make strong affectional bonds to particular others’ (1977, p. 201). Over
the course of childhood and adolescence, children are hypothesized to internalize their
experiences with caretakers to form schwas or what Bowlby refers to as working models of

*Requests for reprints.

250 Kim Bartholomew
close attachment relations. And he postulates two dimensions underlying these working
models: ‘(a)whether or not the attachment figure is judged to be the sort of person who in
general responds to calls for support and protection; [and’) (b) whether or not the self is
judged to be the sort of person towards whom anyone, and the attachment figure in
particular, is likely to respond in a helpful way’ (1973, p. 204). These working models of
other people and of the self become integrated into the personality structure and thereby
provide the prototype for later social relations.
Let us now review how previous attachment research can be interpreted in terms of
models of the self and other (see Fig. 1). In the standard childhood attachment paradigm,
three patterns of attachment are identified: secure, anxious ambivalent, and anxious
avoidant (Ainsworth, Blehar, Waters & Wall, 1978). Although a few studies have
examined working models in early childhood, our knowledge of these patterns is largely
limited to behavioural measures, and I think it is too early to come to any conclusions
about the working models associated with these patterns.
However, two major lines of work have looked at attachment patterns in adults.’
First, Hazan & Shaver (1987) extended the childhood paradigm to adult love relation-
ships, and they developed a brief self-report measure to assess adult parallels of the three
childhood patterns. In a second independent line of work, Mary Main (Main, Kaplan &
Cassidy, 1985) was interested in how adults’ representations of their childhood
experiences in the family may impact upon their child rearing practices. She also defined
three parallel adult patterns as assessed by a semi-structured interview, the Adult
Attachment Interview. Trained coders judged individuals’ attachment patterns based
on descriptions of childhood experiences growing up in the family, with particular
attention given to bow individuals discuss their childhoods, rather than the content of their
descriptions.

CHILDHOOD 11 ADULTHOOD

Child to Love Family Close Adult

Parent Relationships Representations Relationships
(Ainsworth) (Hazan (Main) (Bartholomew)
& Shaver)

Secure Secure Secure Secure

Ambivalent Ambivalent Preoccupied Preoccupied

Avoidant Avoidant Fearful

Dismissing Dismissing

Figure 1. Conceptual correspondence of childhood and adult attachment patterns.

‘There are other lines of adult attachment work that I have not covered, notably the work of West & Sheldon (1987) and
Sperling & Berman (1991). Also, I have not included subsequently identified fourth patterns of infant attachment and a
fourth pattern identified in the Adult Attachment Interview (e.g. Crittenden, 1988; Main & Hesse, 1990).
 Adult attachment processses 25 1
In both streams of research, there is evidence that secure individuals have positive
conceptions of both themselves and others. The ambivalent and preoccupied also seem to
have a positive Orientation to others, as evidenced by a valuing of love relationships, a
tendency to fall in love readily, a tendency to go to others for support, and a preoccupation
with close relationships. But in contrast to the secure, they describe themselves in quite
negative terms. Finally, self-described avoidant individuals report a negative orientation
to others, saying they are distrustful of others and not inclined to fall in love readily. The
dismissing also show a negative orientation to others, especially in downplaying the
importance of close relationships. However, individuals in the two avoidant groups differ
in their self-conceptions. The avoidants identified by Hazan & Shaver’s self-report
measure describe themselves in relatively negative terms, whereas the dismissing
identified by the Adult Attachment Interview show a relatively positive conception of
the self. For reviews of these findings, see Bartholomew (1990, 1993), Bartholomew &
Shaver (in press), Shaver & Hazan (1993) and Shaver & Clark (1994).
In summary, there have been two very different approaches to applying the childhood
attachment classification to adults. These approaches differ in method of assessment (self-
report vs. interview), focus on content versus structure, and content domain (love vs.
family relationships). Though these different approaches have resulted in somewhat
consistent findings, the two groups conceptually tied to the infant avoidant pattern
appear to be characterized by different self-models.

T h e four-category model of adult attachment

I have proposed an expanded model of adult attachment which conceptualizes adult
attachment in intimate peer relations-along the lines pioneered by Hazan & Shaver-
but distinguishes between two forms of adult avoidance. I would argue that the Hazan &
Shaver (1987) measure tends to capture adults who avoid intimacy because of a fear of
rejection (who I labelfea$uI), and the Adult Attachment Interview identifies adults who
dismiss or take a detached stance towards close relationships (who I label dismissing). More
generally, I have come to the conclusion that to understand adult attachment patterns it is
important to take into consideration both individuals’ perceptions of themselves, how
they feel about themselves in relation to others, and their expectations of others and their
approach to others. I therefore looked to Bowlby’s conceptual analysis of internal models
of the self and cther to provide a framework for looking at the potential range of adult
attachment patterns (Bartholomew, 1990; Bartholomew & Horowitz, 1991). Within this
new framework, four prototypic attachment patterns are defined in terms of the
intersection of two underlying dimensions-how positive to negative models of the
self are and how positive to negative models of hypothetical others are (see Fig. 2).
The self-model dimension is associated with the degree of emotional dependence on
others for self-validation. Thus, a positive self-model reflects an internalized sense of self-
worth that is not dependent on ongoing external validation, and a negative self-model is
associated with anxiety regarding acceptance and rejection in close relationships. The
other-model dimension reflects expectations of others’ availability and supportiveness.
Positive other-models thus facilitate actively seeking out intimacy and support in close
relationships, and negative other-models lead to avoidance of intimacy. Each combination
of self- and other-models defines a prototypical attachment pattern, or a particular
252 Kim Bartholomew

Positive
Model of Other

SECURE PREOCCUPIED

Positive Negative
Model of Model of
Self Self

DISMISSING FEARFUL

Negative
Model of Other
Figure 2. Four-category model of adult attachment

strategy of regulating felt security within close relationships (Bartholomew, Cobb &
Poole, 1997).
First, prototypically seczlre individuals are characterized by a positive image of the self
and positive images of others. Consistently responsive caretaking in childhood is
hypothesized to have facilitated the development of both an internalized sense of self-
worth and a trust that others will generally be available and supportive. The secure are
thus high on both autonomy and intimacy, and they are comfortable using others as a
source of support when needed.
Preocctlpied individuals are characterized by a negative self-model and a positive model
of others. Inconsistent parenting, particularly if accompanied by messages of parental
devotion, may lead children to conclude that they are to blame for any lack of love from
caretakers. Preoccupied individuals are preoccupied with their attachment needs and
actively seek to have those needs fulfilled in their close relationships. The result is an
overly dependent style in which personal validation is sought through gaining others’
acceptance and approval.
In contrast, individuals showing both of the avoidant patterns avoid close contact with
others, especially under conditions of stress. Presumably due to a history of rejecting or
 Adult attachment processses 253
unresponsive attachment figures, they have learned not to turn to other people as a source
of security. But individuals in the two avoidant groups have come to very different
conclusions about their own self-worth. On the one hand, the fearful have concluded both
that others are uncaring and unavailable, and that they themselves are unlovable.
Although they desire acceptance by others and are aware of their attachment needs,
the fearful avoid becoming close out of a fear or expectation of being rejected. The
dismissing, in contrast, have managed to maintain a positive self-image by distancing
themselves from attachment figures and developing a model of the self as self-reliant and
invulnerable to the potential rejection of others.
Thus, the preoccupied and fearful are alike in their dependence on others’ acceptance,
in their attachment anxiety, but they differ in their willingness to approach others for
support. In contrast, the fearful and dismissing both avoid seeking support from others,
but differ in their emotional reliance on others’ acceptance.
Any model is invariably a simplification of reality, and this one is no exception.
Bowlby’s concept of positivity of self and other models is a convenient organizing
framework. In practice, the two dimensions interact, such that the attachment patterns
are better described in terms of self-in-relation-to-other-models (Griffin & Bartholomew,
19946). In contrast to research employing the Adult Attachment Interview, the four-
category model conceptualizes working models that are more or less consciously held
(though they tend to operate automatically). However, I presume that at some
unconscious level prototypical dismissing individuals do feel negatively about them-
selves, and their adoption of a detached stance toward others is a way of defending a
fragile sense of self from potential hurt by others (Bartholomew, 1990). Similarly, the
positive model of the preoccupied masks a less conscious negative model of others, with a
tendency to idealize others acting as a defense against acknowledging that significant
others are, at least at times, uncaring and unavailable. The disparate self and other models
of the dismissing and preoccupied can be understood in terms of Bowlby’s conception of
multiple models (1973).
My first step in assessing the four-category model was to develop reliable measures of
the four attachment patterns proposed. The primary measure of the four-category model
is the Peer Attachment Interview (Bartholomew & Horowitz, 1991), a semi-structured
interview exploring participants’ experiences in close friendships and romantic relation-
ships. Particular attention is given to those aspects of peer relationships most relevant to
attachment-such as use of others for emotional support. Based on these interviews,
trained coders assess each participant’s fit with the four attachment patterns. Although
my primary research interest is adult intimate relationships, this model is also applicable
to representations of family relationships (along the lines of the Adult Attachment
Interview). A Family Attachment Interview (Bartholomew & Horowitz, 199 1 ) assesses
the four attachment patterns in this domain. These two interviews have been used with a
range of samples, including college students (Bartholomew & Horowitz, 1991), men and
women in violent spousal relationships (Saunders, 1992), adolescents diagnosed with
conduct disorder (Scharfe, 1996), and women undergoing treatment for breast cancer
(Bellg, 1995). I have also developed two paper-and-pencil attachment measures that can
be filled out by the target individual or knowledgeable informants such as close friends
and romantic partners (e.g. Bartholomew & Horowitz, 1991; Griffin & Bartholomew,
1994a; Scharfe & Bartholomew, 1994a).
254 Kim Bartholomew
Regardless of the assessment method, there are three ways that ratings based on the
four-category model can be treated. First, each method yields a continuous rating of each
of the four attachment prototypes, referred to as prototype ratings. These four ratings
provide an attachment profile for each individual. For example, an individual who is quite
secure, but with preoccupied and fearful tendencies, may receive a profile (on nine-point
scales ranging from no fit to perfect fit with prototype) of 7 on security, 3 on preoccupation,
2 on fearfulness, and 1 on dismissingness. Second, linear combinations of the four
prototype ratings can be used to derive ratings of the two underlying dimensions,
positivity of the self-model and positivity of the other-model (e.g. Griffin & Bartholomew,
19946;Scharfe & Bartholomew, 1994a). Scores on the two dimensions could then be used
to plot each individual in the two dimensional space shown in Fig. 2. Finally, the highest
of the four prototype ratings can be used to derive a best-fitting category membership. In
non-clinical samples we generally find that about 50 per cent of participants are
predominantly secure, with the rest distributed across the insecure patterns.
Much of my research over the past few years has been directed toward testing the
validity of this model. Various colleagues and I have demonstrated that the intercorre-
lations among ratings of the four patterns conform to the structure implied by the model,
that measures based on the model show adequate convergent and discriminant validity,
and that the two attachment dimensions are related to theoretically relevant outcome
variables (Bartholomew & Horowitz, 1991; Bartholomew & Shaver, in press; Griffin &
Bartholomew, 1994a,b).
Although these analyses are important to validate the model, my primary goal in this
work is to understand the attachment-related problems that individuals experience in
their close relationships. I would expect the interaction of differing self and other models
to give rise to distinct interpersonal patterns and problems. I have used the circumplex
model of interpersonal behaviour (also referred to as the interpersonal circle) to look at
this question. From this perspective, interpersonal behaviours are seen as being jointly
defined by two dimensions: a vertical dimension of control (dominance to submission)
and a horizontal dimension of affiliation (warmth to coldness or distance) (e.g. Kiesler,
1983; Wiggins, 1982). Maladaptive interpersonal behaviour is characterized by a lack of
flexibility in moving around the circle in response to situational demands. The
instrument I have used to measure behaviour on the circumplex is the Inventory of
Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureno & Villasenor, 1988). A
circumplex scoring procedure for the IIP yields eight problem subscales, one for each
octant of the interpersonal circle (Alden, Wiggins & Pincus, 1990). Profiles of
interpersonal problems for individuals or groups can then be constructed by looking at
mean levels across the eight subscales.
Figure 3 presents the interpersonal profiles for the four attachment groups from a
sample of 77 college students (see sample 1 , Bartholomew & Horowitz, 1991). The solid
line indicates self-reports and the dotted line indicates the reports of close same-sex
friends. The secure group’s profile of problems, though somewhat elevated on the warm
side of the interpersonal space, is not distinctive (i.e. no subscale scores were extreme). In
contrast, the dismissing group’s profile is centered on the cold side of the circumplex,
according to both self and peer reports. Typical items endorsed by the dismissing are ‘It is
hard for me to feel close to others’ and ‘A lot of people are not worth getting to know’.
Thus, the dismissing show interpersonal problems associated with distance from and
 Adult attachment processses 255
coldness toward others. The fearful group’s profile shows elevations on introversion and
subassertiveness. Typical items assessing these forms of interpersonal problems are ‘I am
too afraid of other people’ and ‘It is hard for me to confront people with problems that
come up’. The preoccupied group shows a very different profile, characterized by
dominant warmth, consistent with these individuals’ active strategies to gain the support
of others. Typical items endorsed by this group are ‘It is hard for me to spend time alone’
and ‘I am overly disclosing’.
In summary, each attachment group demonstrates a distinct and theoretically
consistent profile of interpersonal problems. Of particular note are the different profiles

SECURE PREOCCUPIED
*ulocntic
-
AutocratIc

Introverted

A
Subassertive
-
SubasSertlVe

DISMISSING FEARFUL

Autocratlc Autocratlc

Subarsertlve

Figure 3. Self-report and friend-report profiles of interpersonal problems across attachment groups.
256 Kim Bartholomew
of the two avoidant groups (the fearful and dismissing), and of the two groups with
negative self-models (the fearful and preoccupied). These patterns have been replicated
across a number of samples, for self-reports and reports of romantic partners (e.g.
Bartholomew & Scharfe, 1 9 9 4 ~ ) .

Attachment in couple relationships

Attachment-related models are expected to guide expectations about close relationships,
feelings and behaviours in close relationships, interpretations of experiences in close
relationships, and perhaps even our choice of relationship partners. Working models are
thereby expected to be self-perpetuating over time (e.g. Bartholomew, 1993; Collins &
Read, 1994; Main et a/., 1985). I will selectively present some findings from a study
conducted by Elaine Scharfe and myself examining how attachment patterns may have
implications for the quality of romantic relationships. The sample consisted of 78 young
couples with a mean relationship length of just over four years and a mean age of 24.6
years (see Scharfe & Bartholomew, 1994a, 1995).
As shown in Table 1, attachment ratings (in this case self-report ratings) were
associated with the relationship satisfaction of both partners. Consistent with past
findings (e.g. Collins & Read, 1990; Simpson, 1990), these associations were somewhat
gender linked: Female partners’ ratings indicating high attachment anxiety (fearful and
preoccupied) predicted low relationship satisfaction for both partners, while male
partners’ ratings of dismissing attachment were uniquely associated with low satisfaction
for both partners. Additional analyses suggest that, to some degree, both partners’
attachment ratings independently contribute to their joint relationship satisfaction.
Table 1. Correlations between relationship-specific self-report attachment ratings and relation-
ship satisfaction
Relationship satisfaction
Female Male
Female attachment
Secure rating .33** .38**
Fearful rating -.24* -.36**
Preoccupied rating -.25* -.35**
Dismissing rating -.18 -.17
Male attachment
Secure rating .15 .24*
Fearful rating .03 -.11
Preoccupied rating -.20 -.27*
Dismissing rating -.47** -.51**
*p < .05; **p < .01; ***p < ,001.

Choice of complementary social partners may be one means through which attachment
patterns are perpetuated over time. We therefore looked at partner matching on
attachment ratings. As shown in Table 2 (this time using interview attachment ratings),
there were some reliable associations between partners’ ratings. Again consistent with
previous findings (e.g. Senchak & Leonard, 1992), the strongest association was between
 Adult attachment processses 257
Table 2. Partner marching on interview attachment ratings
Male partner
~ ~ ~

Female partner Secure Fear Preoc Dism

Secure .47*** -.I8 .11 -.34**
Fear -.04 .02 .01 .04
Preoc -.36** .23* .10 .12
Dism -.18 -.07 -.08 .20
*p < .05; **p < .01; ***p < ,001.
Note. Abbreviations refer to the fearful (Fear), preoccupied (Preoc) and dismissing (Dism) attachment ratings.

the security ratings of the two partners. Consistent with the satisfaction findings, security
of women was negatively associated with the degree to which their male partners were
rated as dismissing, and security of males was negatively associated with preoccupation in
their female partners. However, it is important to be cautious in interpretation of these
cross-sectional findings. For example, the latter finding may indicate that secure women
tend not to select dismissing men as romantic partners (or dismissing men do not select
secure women), that secure women tend to select themselves out of relationships with
dismissing men over time (or dismissing men do not stay with secure women), that
secure females tend to help their partners to become less dismissing over time (or
dismissing men drive their partners to become less secure), or, most likely, some complex
combination of the above. Only long-term longitudinal work will be able to sort out
these alternate interpretations of the findings.
We also explored whether attachment security was associated with the degree to which
couples share similar views of each other’s interpersonal difficulties. An important aspect
of security is the ability to realistically evaluate our own interpersonal needs and
behaviours, as well as those of our relationship partners. We therefore expected that
security would predict agreement between partners on their interpersonal problems. In
fact, we did find that security of both partners predicted agreement, independently of
level of problems (Bartholomew & Scharfe, 1994a).
Finally, we looked at whether attachment patterns were predictive of break-ups in this
sample over a two-year period (Bartholomew, Scharfe & Henderson, 1996). We are still in
the process of looking at this data, but so far we have found no evidence that attachment
security is predictive of relationship maintenance. In fact, fearfulness of both partners
appears to be weakly predictive of relationship maintenance. From our experiences in
coding attachment interviews, it comes as no surprise that security does not necessarily
facilitate the maintenance of long-term romantic relationships. Secure individuals are
able to end unsatisfying relationships and even to feel comfortable without romantic
relationships for extended periods. Conversely, we tend to assume that insecurity will
undermine the ability of individuals to maintain long-term relationships. However,
insecure individuals are also at risk for getting involved in and being unable to extradite
themselves from problematic relationships-as dramatically shown in the samples of
abusive and abused spouses that we have worked with.

Abusive spousal relationshipj

The puzzle of how and why individuals become involved in and remain in abusive
258 Kim Bartholomew
relationships becomes more understandable when one considers Bowlby’sproposition that
the strength of attachment bonds is unrelated to the quality of the attachment relationship
(1973). Bowlby believed that strong bonds are formed in conditions of threat which
activate the attachment system and lead the threatened individual to seek proximity to an
attachment figure, even when that attachment figure may be the source of threat.
Toni Henderson and I studied a sample of 63 abused women who were followed up
over six months, once shortly after having left their abusive relationship and again six
months later (Henderson, Bartholomew & Dutton, 1997). We investigated whether
attachment, based on an interview assessment at Time 1, would predict how these women
were coping with the separation from their abusive partners six months later. Separation
resolution was assessed by a collection of variables, including continued emotional
involvement with the partner, desire for reconciliation with the partner, and contact with
the partner during the separation. Nor surprisingly, 87 per cent of these women had a
predominant attachment pattern with a negative self-model, either fearful or preoccu-
pied. Though in general the fearful pattern is most strongly associated with poor mental
health (e.g. Carnelley, Pietromonaco & Jaffe, 1994), in this setting preoccupation was
predictive of the most problematic separation resolution. Preoccupied women tend to
idealize romantic relationships, tend to have a history of multiple leavings and reunions
with romantic partners (Kirkpatrick & Hazan, 1994), and may be especially responsive to
the so-called contrition phase of abusive relationships.
Initial studies applying an attachment framework to understanding why men are
violent to their intimate partners have also been fruitful. The stereotype of the abuser is of
a distant, unyielding, and callous individual, typical of the dismissing pattern. However,
from an attachment perspective, violent men’s assaultive episodes may be a form of
protest behaviour precipitated by a real or imagined threat of rejection or abandonment
by the attachment figure. The attachment system is thus activated in order to regain the
attachment figure. Consistent with this proposition, we found that fearfulness and
preoccupation were elevated (as assessed by self-reports and attachment interviews) in a
sample of 120 men undergoing treatment for battering their female partners (Dutton,
Saunders, Starzomski & Bartholomew, 1994; Saunders, 1992). Self-report ratings of
fearfulness and preoccupation were also predictive of severity of abusiveness, anger,
jealousy, and trauma symptoms. It is noteworthy that attachment patterns with a negative
self-model predominate in both the victims and perpetrators of abuse. Future research
needs to examine both partners in abusive relationships to begin to understand how the
attachment patterns of each individual affect the dynamic interaction of the two partners.

Current issues in the study of adult attachment

Many, or more accurately, most, of the important questions regarding adult attachment
remain to be addressed. I will briefly touch on five outstanding issues.

1. How stable are adult attachment patterns over time?

If adult attachment patterns are conceived of as an individual difference variable, they
should be at least moderately stable over time. Elaine Scharfe and I assessed the stability
of attachment patterns in the sample of young couples previously described (Scharfe &
 Adult attachment processses 259
Bartholomew, 1994a). We found high levels of stability for interview ratings over eight
months (with stability coefficients ranging from .72 to .85), and moderate stability of
self-report and partner-report ratings (averaging .5 1). We are still analysing two-year
stabilities for this sample, but preliminary findings indicate equally high levels of
stability over two years as over eight months (Scharfe & Bartholomew, 19946). This result
suggests that the true stability of attachment patterns may be very high over this period,
but that the observed stabilities are attenuated by unreliability of the measures.
Consistent with this interpretation, Kirkpatrick & Hazan (1994) have shown comparable
levels of stability over four years for a diverse community sample using the Hazan &
Shaver (1987) self-report measurement of attachment. It should be noted, however, that
these initial studies of the stability of attachment patterns in adulthood do not speak to
the longer-term stability of attachment patterns or to the continuity of patterns from
childhood to adulthood.

2. Are attachment dimensions reducible t o general personality factors?

The five-factor model of personality (commonly referred to as the Big Five) is a widely
used and well-validated frame of reference within which new constructs such as
attachment can be located (e.g. Digman, 1990; John, 1990; McCrae & Costa, 1987).
The Big Five dimensions are neuroticism, extraversion, openness to experience, agree-
ableness, and conscientiousness. Dale Griffin and I looked at the associations between the
two dimensions underlying the four-category model of attachment and the Big Five
personality dimensions. We found that 48 per cent of the variance in the latent self-model
dimension of the four-category model overlaps with the Big Five (and neuroticism in
particular), and 27 per cent of the variance in the latent other-model dimension overlaps
with the Big Five (with low associations with extraversion and agreeableness) (Griffin &
Bartholomew, 19946). These findings suggest that although there are meaningful
associations between attachment and personality, attachment ratings are far from
reducible to general personality factors (see Shaver & Brennan, 1992, for a similar
demonstration). There is also evidence that attachment ratings predict important
components of relationships functioning (interpersonal dependency in one study and
relationship satisfaction in another) independently of their associations with personality
dimensions (Griffin & Bartholomew, 19946; Shaver & Brennan, 1992).

3. Categories uersus prototypes

In both the childhood and adult attachment fields it has been common to conceptualize
individual differences in attachment in terms of three or more discrete categories.
However, we have shown across a range of studies that such approaches overlook
meaningful variation within the attachment categories and that a prototype approach
to assessing attachment may more accurately reflect the nature of individual differences in
this domain (Griffin & Bartholomew, 19946). Individuals do not fit neatly into simple
categories: Most individuals show a complex profile across attachment patterns, and there
is considerable variation among individuals within best-fitting categories. For example,
in the couples study previously described we found that individuals categorized as secure
(based on interview ratings) reported more satisfying relationships than those categorized
260 Kim Bartholomew
as insecure. We then looked at the associations between degree of attachment security and
satisfaction within the secure and insecure groups. As shown in Table 3 , ratings of
attachment security are predictive of satisfaction, even within the secure and insecure
groups (Bartholomew & Scharfe, 1994s). This variation is lost with simple grouping
approaches to assessing adult attachment.

Table 3. Correlations between interview ratings of attachment security and relationship

satisfaction by security categorization
Relationship satisfaction
Female Male
Secure subsamples
Female security .58** .40**
Male security .20 .08
Insecure subsamples
Female security .30* .29*
Male security .46** .29*
*p < .05;**p < .01; ***p < ,001.

Theoretically, two individuals with the same primary attachment patterns will have
different interpersonal experiences if they have different secondary or even tertiary
strategies. For example, a secure individual who has a mix of secure and preoccupied
tendencies will look very different from a secure-dismissing individual. In research
practice, this means that even if we are faced with a special sample that does not include
the whole range of attachment groups, we can still expect the four prototype ratings to
yield meaningful results (e.g. Dutton et al., 1994). In terms of applying such a model to
individuals, for example as an aid to conceptualizing therapy clients’ interpersonal
difficulties, it would also be critical to consider attachment profiles rather than groupings.

4. Do adults have multiple attachment relationships?

There has been a tendency in the field of adult attachment to focus on romantickexual
relationships as the important and defining attachment relationship in adult life.
Although there is evidence that romantic partners become the primary or preferred
attachment figure for adults in long-term sexual relationships (Hazan & Zeifman, 1994),
it may be that most adults have a network of attachment relationships. Shanna Trinke and
I investigated this question in a sample of 223 young adults (Trinke & Bartholomew, in
press). We found that most participants reported three to five relationships that met the
criteria for attachment relationships and that these relationships appeared to be arranged
in a hierarchy. Romantic partners (for those in relationships) and mothers were most
commonly judged to be at the top of the attachment hierarchy, followed by fathers,
siblings, and close friends. Our findings emphasize the continued importance of family
members for the fulfillment of attachment needs, at least through young adulthood. In
ongoing work, we are examining whether the composition of adults’ attachment
networks have implications for their psychological well-being.
 Adult attachment processses 261
5. Do adults have general attachment ‘styles’ or is there specificity across domains?
Many attachment researchers talk of attachment ‘styles’ as if people possess general
internal models or strategies that are applied across all relationships and situations. And
yet researchers typically use only one measurement approach (an interview or self-report
focusing on peer, romantic, or family relations) that by its very nature may be limiting
the construct of interest (Bartholomew, 1994).Findings from various studies suggest that
there is moderate associations between domains, suggesting some generality. But it has
also become clear that different domains are conceptually and empirically distinct, and
especially the family and peer domains (Bartholomew & Shaver, in press). Future work
needs to address how these various attachment-related models are organized within
individuals and under what conditions varying models are activated (Collins & Read,
1994).

Conclusions
In summary, this four-category models extends and integrates previous research in adult
attachment in a number of ways. First, it encompasses the different domains and methods
of measurement that previously have been studied in isolation. Second, it indicates that
there may be two distinct forms of avoidance and thereby helps explain inconsistent
findings from prior research. As well, in contrast to previous attachment work dealing
with groups with no clear relations to one another, this model proposes that there is an
underlying structure that specifies the relations between various attachment patterns. And
the model applies a prototype conception of attachment patterns that is more sensitive to
the complexity of individual differences than are traditional grouping approaches. There
are still many unresolved issues, both with this model of adult attachment and with this
approach to studying relationships in general. However, I hope that I have demonstrated
that the proposed model is a promising theoretical framework for understanding the
range of difficulties that all too often arises in close relationships.

Acknowledgement
This paper is based on a talk given at Attachment and Psychotherapy, the inaugural joint meeting between
The Psychotherapy Section of the British Psychological Society and The Psychotherapy Section of the Royal
College of Psychiatrists ar the Royal Society of Medicine, London, 10 November 1995.

References
Ainsworth, M. D. S., Blehar, M. C., Waters, E. &Wall, S. (1978). Patterns ofAttachment: A PsychologicalStudy
of the Strange Situation. Hillsdale, NJ: Erlbaum.
Alden, L. E., Wiggins, J. S. & Pincus, A. L. (1990). Construction of circumplex scales for the Inventory of
Interpersonal Problems. Journal of Personality Assessment, 5 5 , 521-536.
Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal
Relationships, 7, 147- 1 78.
Bartholomew, K. (1993). From childhood to adult relationships: Attachment theory and research. In S. Duck
(Ed.), Understanding Relationship Processes: Learning about Relationships, vol. 2 , pp. 30-62. Beverly Hills, CA:
Sage.
Bartholomew, K. (1994). The assessment of individual differences in adult attachment. Psychological Inquiry,
5, 23-27.
262 Kim Bartholomew
Bartholomew, K., Cobb, R. J. & Poole, J. A. (1997). Adult attachment patterns and social support processes.
In G. Pierce, B. Lakey, I. Sarason & B. Sarason (Eds), Social Support and Personality: Structure, Process, and
Change, pp. 359-378. New York: Plenum.
Bartholomew, K. & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category
model. Journal of Personality and Social Psychology, 61, 226-244.
Bartholomew, K. & Scharfe, E. (1994~).Adult attachment patterns and interpersonal problems in young
adults. Paper presented at the American Psychological Association Convention, Los Angeles, California,
August.
Bartholomew, K. & Scharfe, E. (1994b). Attachment processes in young couples. Paper presented at the
Seventh International Conference on Personal Relationships, Groningen, The Netherlands, July.
Bartholomew, K., Scharfe, E. & Henderson, A. J. 2. (1996). Longitudinal study of attachment processes in
young couples. Paper presented at the International Conference on Personal Relationships, Banff, Canada,
August.
Bartholomew, K., & Shaver, P. (in press). Methods of assessing adult attachment: Do they converge? In J. A.
Simpson & W. S. Rholes (Eds), Attachment Theory and Close Relationships. New York: Guilford.
Bellg, A. J. (1995). Adult attachment and adjustment to breast cancer. Unpublished doctoral dissertation,
University of Rochester, New York.
Bowlby, J. (1973). Attachment and loss: vol. 2: Separation: Anxiety and Anger. New York: Basic Books.
Bowlby, J. (1977). The making and breaking of affectional bonds. British Journal of Psychiatry, 130,
201-210.
Carnelley, K., Pietrornonaco, P. & Jaffe, K. (1994). Depression, working models of others, and relationship
functioning. Journal of Personality and Social Psychology, 66, 127-140.
Collins, N. L. & Read, S. J. (1990). Adult attachment, working models, and relationship quality in dating
couples. Journal of Personality and Social Psychology, 58, 644-663.
Collins, N. L. & Read, S. J. (1994). Cognitive representations of attachment: The structure and function of
working models. In K. Bartholomew & D. Perlman (Eds), Advances in Personal Relationships, vol. 5,
Attachment Processes in Adulthood, pp. 53-90. London: Jessica Kingsley.
Crittenden, P. (1988). Relationships at risk. In J. Belsky & T. Nezworski (Eds), Clinical lmplzcations of
Attachment, pp. 136-174. Hillsdale, NJ: Erlbaum.
Digman, J. M. (1990). Personality structure: Emergence of the five-factor model. Annual Review of Psychology,
41,417-440.
Dutton, D. G., Saunders, K., Starzomski, A. J. & Bartholomew, K. (1994). Intimacy-anger and insecure
attachment as precursors of abuse in intimate relationships. Journal of Applied Social Psychology, 24, 1367-
1386.
Griffin, D. & Bartholomew, K. (1994~).Models of the self and other: Fundamental dimensions underlying
measures of adult attachment. Journal of Personality and Social Psychology, 67, 430-445.
Griffin, D. W. & Bartholomew, K. (1994b). The metaphysics of measurement: The case of adult attachment.
In K. Bartholomew & D. Perlman (Eds), Advances in Personal Relationships, vol. 5, Attachment Processes in
Adulthood, pp. 17-52. London: Jessica Kingsley.
Hazan, C. & Shaver, P. R. (1987). Romantic love conceptualized as an attachment process. Journal of
Personality and Social Psychology, 5 2 , 5 11-524.
Hazan, C. & Zeifman, D. (1994). Sex and the psychological tether. In K. Bartholomew & D. Perlman (Eds),
Advances in Personal Relationships, vol. 5, Attachment Processes in Adulthood, pp. 151-178. London: Jessica
Kingsley.
Henderson, A. J. Z., Bartholomew, K. & Dutton, D. G. (1997). He loves me; he loves me not: Attachment
and separation resolution of abused women. Journal of Family Violence, 12, 169-191.
Horowitz, L. M., Rosenberg, S. E., Baer, B. A,, Ureno, G. & Villasenor, V. S. (1988). Inventory of
Interpersonal Problems: Psychometric properties and clinical applications. Journal of Consulting and
Clinical Psychology, 56, 885-892.
John, 0. P. (1990). The ‘Big Five’ factor taxonomy: Dimensions of personality in the natural language and in
questionnaires. In L. Pervin (Ed.), Handbook of Personality Theory and Research. New York: Guilford.
Kiesler, D. J. (1983). The 1982 interpersonal circle: A taxonomy for complimentarity in human transactions.
Psychological Reuiew, 90, 185-214.
Kirkpatrick, L. A. & Hazan, C. (1994). Attachment styles and close relationships: A four-year prospective
study. Personal Relationships, 1, 123- 142.
 Adult attachment processses 263
McCrae, R. R. & Costa, P. T. Jr (1987). Validation of the five-factor model of personality across instruments
and observers.Journal of Personality and Social Psychology, 52, 81-90.
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. T.
Greenberg, D. Cichetti & 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.
Saunders, K. D. (1992). The links between patterns of male attachment and the development of wife assault.
Unpublished doctoral dissertation, University of British Colombia, Vancouver, BC, Canada.
Scharfe, E. A. (1996). A test of Bartholomew’s four-category model in a clinical sample of adolescents.
Unpublished doctoral dissertation, Simon Fraser University, Burnaby, BC, Canada.
Scharfe, E. & Bartholomew, K. (1995). Accommodation and attachment representations in young couples.
Journal of Social and Personal Relationships, 12, 389-401.
Scharfe, E. & Bartholomew, K. (1994~).Reliability and stability of adult attachment patterns. Personal
Relationships, 1, 23 -43.
Scharfe, E. & Bartholomew, K. (19946). Stability of adult attachment representations: A two-year follow-up.
Paper presented at the American Psychological Association Convention, Los Angeles, California, August.
Senchak, M. & Leonard, K. E. (1992). Attachment styles and marital adjustment among newlywed couples.
Journal of Social and Personal Relationships, 9, 5 1-64.
Shaver, P. R. & Brennan. K. A. (1992). Attachment styles and the ‘big five’ personality traits: Their
connections with each other and with romantic relationship outcomes. Personality and Social Psychology
Bulletin, 18, 536-545.
Shaver, P. R. & Clark, C. L. (1994). The psychodynamics of adult romantic attachment. In J. M. Masling & R.
F. Bornstein (Eds), Empirical Perspectives on Object Relations Theory, pp. 105-1 56. Washington, DC:
American Psychological Association.
Shaver, P. R. & Hazan, C. (1993). Adult romantic attachment: Theory and evidence. In D. Perlman & W. H.
Jones (Eds), Advances in Personal Relationships, vol. 4, pp. 29-70. London: Jessica Kingsley.
Simpson, J. A. (1990). The influence of attachment styles on romantic relationships. Journal of Personality and
Social Psychology, 59, 97 1-980.
Sperling, M. B. & Berman, W. H. (1991). An attachment classification of desperate love. Journal of Personality
Assessment, 56, 45-55.
Trinke, S. & Bartholomew. K. (in press). Attachment hierarchies in young adults.Journal of Social and Personal
Relationships.
West, M. & Sheldon, A. (1987). An approach to the delineation of adult attachment: Scale development and
reliability. Journal of Nwvous and Mental Disease, 175, 738-741.
Wiggins, J. S. (1982). Circumplex models of interpersonal behavior in clinical psychology. In P. C. Kendall &
J. N . Butcher (Eds), Handbook ofResearch Methods in Clinical Psychology, pp. 183-221. New York: Wiley.

Received 9 May 1996; revised version received 24 October 1996