You are on page 1of 16

Review of General Psychology Copyright 2000 by the Educational Publishing Foundation

2000, Vol. 4, No. 2, 111-131 1089-2680/00/$5.00 DOI: 10.1037//1089-2680.4.2.U1

Adult Romantic Attachments:


A Developmental Perspective on Individual Differences
Jude Cassidy
University of Maryland

Attachment theory has, since its inception, contained the proposition that long-term
adult romantic relationships (i.e., relationships described by ethologists as "pair
bonds") are generally attachments. In this article, the possibility that individual differ-
ences in adult romantic attachments may emerge, in part, from individual differences
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

in childhood attachments is discussed. The article begins with an examination of the


developmental precursors of individual differences in two of the behavioral systems
prominent in adult romantic relationships: the attachment system and the caregiving
system. For each of these behavioral systems, theory is discussed and the empirical
literature is reviewed. The remainder of the article addresses the mechanisms of both
continuity across development (i.e., factors that may account for the influences of early
attachments on later romantic relationships) and discontinuity (i.e., factors that may
account for change in the quality of attachments from childhood to adulthood).

Attachment theory has, since its inception, elaborated the reasons for viewing these rela-
contained the proposition that long-term adult tionships as attachments, describing points of
romantic relationships (i.e., relationships de- congruence between infant-caregiver bonds
scribed by ethologists as "pair bonds") are gen- and adult romantic bonds: "[These] include the
erally attachments and that these attachments nature of physical contact that typifies and dis-
provide important benefits to both partners tinguishes attachment bonds; the factors that
(Ainsworth, 1989; Bowlby, 1979). Although at- influence the selection of attachment figures;
tachment has been studied mostly in infants and reactions to attachment disruption and loss; and
mothers, Bowlby made it clear that attachment the role of attachment in biological and psycho-
is important "from the cradle to grave" (1979, p. logical fitness" (p. 341).
129) and that people of all ages do best when Although most long-term adult romantic re-
they have one or a few trusted attachment fig- lationships are considered attachments, theorists
ures to whom they can turn in times of trouble. have pointed out that the "attachment behav-
Ainsworth (1989) listed six criteria for attach- ioral system" (a biologically based system of
ments across the life span, all of which typically behaviors that serves to maintain proximity or
characterize adult romantic relationships: (a) contact between an individual and an attach-
persistence, (b) specificity to a particular indi- ment figure) is not the only behavioral system
vidual, (c) emotional significance, (d) desire for central to these relationships. Ainsworth (1989),
proximity or contact, (e) distress at involuntary as well as Shaver, Hazan, and Bradshaw (1988),
separation, and (0 security and comfort seeking. described these relationships as characterized
More recently, Hazan and Zeifman (1999) have by a complex intertwining of (principally) three
behavioral systems: the attachment system, the
caregiving system, and the sexual system. The
attachment system involves behavior organized
This article was written with the support of grants from around maintaining proximity to or contact with
the National Institute of Mental Health (RO1MH50773 and
RO1MH58907) and the National Institute for Child Health an attachment figure, particularly in times of
and Development (RO1HD36635). I am grateful to Jon trouble; the attachment figure serves as a "haven
Mohr for his thoughtful comments on a previous version of of safety" during such times. The caregiving
the article. system involves the organization of behaviors
Correspondence concerning this article should be ad- that provide care to another, again, particularly
dressed to Jude Cassidy, Department of Psychology, Uni-
versity of Maryland, College Park, Maryland 20742. in times of trouble. The sexual system consists

111
112 CASSIDY

of behaviors related to sexual activity (Bowlby, viewed as increasing reproductive fitness in at


1969/1982). least two ways: Caregiving of an adult partner
There are individual differences in the rela- increases the likelihood that the partner will
tive importance of each of these behavioral sys- survive (a) to provide the adult with reciprocal
tems across relationships, and indeed there can protection and care and (b) to provide the shared
be such differences between partners within re- offspring with protection and care. It is perhaps
lationships. Not only are there variations in the easiest to see how activation of the sexual be-
importance of these behavioral systems, there havioral system, with the creation of offspring
are also variations in the competence of couples as the predictable outcome, increases an indi-
across these behavioral systems (e.g., a couple vidual's reproductive fitness.
may function better in relation to attachment In this article, I discuss how individual dif-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and caregiving than in relation to sexual behav- ferences in adult romantic attachments may
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ior). And within a relationship, partners may emerge, in part, from childhood attachments. I
differ in their relative competence across areas begin by examining the developmental precur-
(e.g., one partner may be more competent in sors of individual differences in two of the
giving care rather than in seeking care, and this behavioral systems prominent in adult romantic
may not be true for the other partner). Further- relationships: the attachment system and the
more, within an adult romantic relationship, the caregiving system. For each of these behavioral
relative importance of and competence in these systems, I discuss theory and then review the
behavioral systems may change over time. empirical literature. In the remainder of the ar-
There is evidence, for example, that the sexual ticle, I first discuss continuity across develop-
behavioral system is most active and intense ment (i.e., factors that may account for the
during the early stages of a relationship (Fisher, influences of early attachments on later roman-
1992) and that attachment, on the other hand, tic relationships). I then discuss discontinuity
develops more gradually over the first 2 years (i.e., factors that may account for change in the
(Hazan & Zeifrnan, 1994; see also Ainsworth, quality of attachments from childhood to adult-
1989). As Hazan and Zeifman (1999) noted, the hood). This article is largely speculative: There
sexual system may serve to keep the couple are no prospective longitudinal studies that be-
together long enough for an attachment to form. gin with infant-child attachment and examine
Although these behavioral systems are con- later attachment and caregiving within romantic
sidered to be largely independent, each with its relationships. Nonetheless, the existing empiri-
own inherent motivational system (Bowlby, cal developmental and adult romantic literatures
1969/1982), functioning in one behavioral sys- offer converging evidence, consistent with at-
tem may influence functioning in another. For tachment theory, that provides a basis for the
example, a woman grateful that her partner re- speculation presented.
sponded to her attachment signals may be
prompted to improve her own reciprocal care- The Adult Attachment and Caregiving
giving; or, a man whose insecure attachment Behavioral Systems: Developmental
contributes to his hostile ambivalence may be-
Precursors
have in ways that interfere with the sexual com-
ponent of his relationship. The Attachment Behavioral System
A defining feature of a behavioral system is
that it increases an individual's ability to trans- The attachment behavioral system, a biolog-
mit his or her genes into future generations ically based system of behaviors most likely to
(i.e., increases his or her reproductive fitness; be activated in times of threat, results in an
Bowlby, 1969/1982). The attachment behavior individual gaining or maintaining proximity to
system serves to keep the individual close to a an attachment figure (Bowlby, 1969/1982; see
protective attachment figure, thereby increasing Cassidy, 1999, for a summary). The attachment
the individual's odds of survival through repro- system is thought to have evolved because in-
ductive age (this is true for adults as well as dividuals who sought an attachment figure
infants; Bowlby, 1969/1982). Activation of the when threatened were more likely to survive to
caregiving system in adult relationships can be pass on the genes for this tendency. Even
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 113

though most sexual pair bonds are likely to be themselves as not lovable and worth little. This
attachments, they are not all secure attachments. notion that people learn about themselves from
In this section, I examine the three styles of seeing how others view them is a perspective
adult romantic attachment that Hazan and central to several theories, beginning with the
Shaver (1987) identified (see Shaver & Hazan, earliest theorizing about the self (e.g., Cooley's,
1993, and Feeney, 1999, for reviews) and dis- 1902, "looking glass self; see also J. Baldwin,
cuss ways in which early childhood attachments 1897; Mead, 1934; and Sullivan, 1953; see
may contribute to these styles. According to Bretherton & Munholland, 1999, and Main,
Bowlby, individual differences in secure and Kaplan, & Cassidy, 1985, for extensive discus-
insecure attachment are best understood by con- sion of representational models; see M. W.
sidering what he called "representational mod- Baldwin, 1992, for a review of similarities be-
els" (he also called these "internal working tween these models and a variety of constructs
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

models") of attachment figures and of the self within the literatures of developmental, social,
(for a review of the working model concept in clinical, and cognitive psychology).
adults, see Pietromonaco & Feldman Barrett, Secure attachment. A well-functioning at-
2000). Representational models of the attach- tachment system in adults is one in which a
ment figure are thought to be based on an indi- person is able to turn to a few appropriately
vidual's experiences with that attachment fig- selected others, effectively, in times of trouble.
ure; Bowlby described these models as "tolera- Adults whose attachment is classified as secure,
bly accurate reflections of the experiences those according to Hazan and Shaver's (1987) classi-
individuals have actually had" (1973, p. 235). fication system, are those who endorse the fol-
These models are similar to cognitive maps lowing description of themselves when asked to
that permit successful navigation of an organ- think about their romantic relationships in gen-
ism's environment; unlike maps, however, eral: "I find it relatively easy to get close to
working models are not static representations others and am comfortable depending on them
but can grow and change. Working models and having them depend on me. I don't worry
guide an individual's expectations, feelings, be- about being abandoned or about someone get-
havior, and cognitive processes. For example, a ting too close to me."
child who is responded to sensitively and What is needed for such a capacity? Accord-
promptly is likely to form a representational ing to Bowlby (1988), two factors are impor-
model of the mother as sensitively responsive; tant. The first is a representational model of
such a model will guide expectations (that the others as available and responsive in times of
mother will respond when needed), feelings trouble. Such a representational model of others
(calmness and security), behavior (seeking the would, according to Bowlby, have emerged in
mother when alarmed and exploring when not large part from childhood experiences of attach-
alarmed), and cognitive processes (positive in- ment figures as available and responsive to at-
terpretation of ambiguous stimuli). Representa- tachment-related signals. A number of studies,
tional models are thought to become so deeply beginning with Ainsworth's pioneering obser-
ingrained that the ways in which they influence vations of mothers and infants in the home,
the individual may become automatic. have reported that secure infants, more than
During childhood, when representational insecure infants, have mothers who are sensi-
models are first developed, there is, according to tively responsive, particularly when the infants
Bowlby (1973), an inextricable intertwining of are upset and signal for comfort. These mothers
the working model of the attachment figure and have been found to be more accepting, cooper-
the working model of the self. Over time, chil- ative, available, comforting, and tender than
dren develop beliefs that their mother will be- other mothers (e.g., Ainsworth, Blehar, Waters,
have in certain predictable ways, and they si- & Wall, 1978; Belsky, Rovine, & Taylor, 1984;
multaneously develop a complementary view of Egeland & Farber, 1984). A recent meta-analy-
themselves. If, for instance, children are loved sis examining 66 studies and more than 4,000
and valued, they will come to view themselves dyads revealed a "moderately strong" associa-
as lovable and valuable. If, however, they are tion between maternal sensitivity and infant at-
rejected or not loved, they will come to view tachment (De Wolff & van IJzendoom, 1997).
114 CASS1DY

Because of this maternal sensitivity, secure chil- children extend these positive representations of
dren are thought to develop representations of their mother to positive representations of oth-
the mother as loving, responsive, and sensitive. ers in general.
There are an increasing number of studies A second factor thought to be important for
that support the proposition that secure attach- secure adult attachment is a representational
ment (and thus, presumably, experiences of ma- model of the self as lovable and worthy of care.
ternal sensitivity) is associated with positive What might be the developmental precursors of
attachment-related models. Main and her col- such a model? Again, there are convergent data
leagues (Kaplan, 1987; Main et al., 1985) re- from several laboratories that secure attachment
ported the first work examining attachment and in childhood is associated with positive models
children's representational models with a longi- of the self. The first evidence of this came from
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tudinal sample followed from infancy to 6 years Sroufe and Egeland's Minnesota Parent-Child
This document is copyrighted by the American Psychological Association or one of its allied publishers.

of age. In that study, attachment to mother was Project; preschoolers who had been securely
assessed in infancy with Ainsworth's strange attached to their mother as infants had higher
situation procedure (Ainsworth et al., 1978); teacher-rated self-esteem (Sroufe, 1983).
children's mental representations related to at- In my own work (Cassidy, 1988), I examined
tachment were tapped at age 6. Children were attachment (assessed at 6 years of age with
asked to discuss photographs of children under- Main and Cassidy's, 1988, reunion procedure)
going increasingly stressful separations from and 6-year-olds' representations of self in sev-
their parents (e.g., the 1st day of school, the eral ways. First, I used a doll story procedure to
parents going out for the evening, or the parents examine "representations of the self within the
going away for a 2-week vacation). At age 6, relationship with the mother." The representa-
the children who had been securely attached as tions of secure children were more positive than
babies had more positive representations. For those of insecure children: Secure children had
example, not only were secure children more representations of the self as valued and worthy
emotionally open, they also were able to de- of care. Second, children's global self-esteem
scribe more active ways of dealing with these was assessed with Harter's (1982) Perceived
separations, a fact that, according to Main and Competence Scale for Children; attachment se-
her colleagues, reflects the experience-based curity scale scores were positively correlated
representations that children securely attached with global self-esteem scores. Third, I tapped
as infants have of their mothers as accessible. children's self-representations by interviewing
A series of studies has followed. Several re- a puppet about the child; the child responded for
searchers have examined attachment represen- the puppet. Securely attached children were
tations in preschool and early school-aged chil- more likely than insecurely attached children to
dren (e.g., Bretherton, Ridgeway, & Cassidy, view themselves in a positive yet realistic light
1990; Cassidy, Kirsh, Scolton, & Parke, 1996; (e.g., "She usually does the right thing but
Shouldice & Stevenson-Hinde, 1992; Slough & sometimes she fights with her sister"). Similar
Greenberg, 1990; Solomon, George, & De findings of a connection between security of
Jong, 1995). The methods of these studies have child-mother attachment representations and a
varied. Some presented children with photo- positive model of the self were obtained in
graphs, and some presented doll stories. In all a study of Belgian 5-year-olds (Verschueren,
cases, the attachment system was activated (the Marcoen, & Schoefs, 1996).
child was hurt and frightened; there was a sep- Representational models of the attachment
aration). In all of the studies, secure children figure as available if needed and of the self as
were more likely than insecure children to re- effective in eliciting care facilitate the infant's
spond in ways reflecting positive representa- ability to use an attachment figure as a "secure
tions of the mother as available, responsive, and base from which to explore," a phenomenon
comforting (see Solomon & George, 1999, for a that Ainsworth (1963) proposed as a key com-
review of these studies; see also Jacobsen, Edel- ponent of secure attachment. The infant's rep-
stein, & Hofmann, 1994, and Jacobsen & Hof- resentation of access to the attachment figure if
mann, 1997). Bowlby (1973) proposed that, trouble arises frees the infant to turn her or his
through the process of generalization, secure attention to exploration. (According to attach-
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 115

merit theory, the intertwining of attachment and relationships—expectations that others are ac-
exploratory behavior systems evolved to ensure cepting of them, imperfections and all—and,
that while the child is protected by maintaining through a variety of self-fulfilling mechanisms,
proximity to attachment figures, he or she none- these positive expectations contribute to the ca-
theless gradually learns about the environment pacity to form later secure attachments. Conver-
through exploration.) In Ainsworth's home ob- gent with these propositions based on studies of
servations during the infant's 1st year, secure secure children are a series of studies examining
infants were characterized by the most flexible the recollections of secure adults: In comparison
balance on an "attachment-exploration contin- with insecure adults, secure adults described
uum" (Ainsworth, Bell, & Stayton, 1971). Ac- their childhood relationships in more positive
cording to Ainsworth, it was their experience- terms (e.g., described their mothers as more
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

based representations of the attachment figure caring, accepting, and dependable; Feeney &
This document is copyrighted by the American Psychological Association or one of its allied publishers.

as available if needed that permitted secure in- Noller, 1990; Hazan & Shaver, 1987; Miku-
fants to follow a simple strategy: seek the at- lincer, Florian, & Tolmacz, 1990; see Rothbard
tachment figure if needed and otherwise ex- & Shaver, 1994, for a review).
plore. Insecure infants, lacking a representation Insecure-avoidant attachment. Adults clas-
of the attachment figure as available, had a less sified as avoidant, according to Hazan and
flexible attachment-exploration balance, in- Shaver (1987), describe themselves as follows:
stead emphasizing excessively one end of the "I am somewhat uncomfortable being close to
continuum or the other. Attachment, then, is others; I find it difficult to trust them com-
viewed not as interfering with exploration but as pletely, difficult to allow myself to depend on
fostering exploration. Thus, it seems likely that them. I am nervous when anyone gets too close,
the capacity in adulthood to balance autonomy and often, love partners want me to be more
and relatedness—the capacity to feel comfort- intimate than I feel comfortable being." A con-
able with closeness while at the same time re- verging body of research indicates that these
specting separateness and enjoying times of adults are not interpersonally oriented, are sus-
solitude—would be rooted in patterns of a picious of the intentions of others, limit their
comfortable balance between attachment and intimacy with others, minimize expressions of
exploration experienced during childhood. distress, and engage in relatively little self-dis-
In sum, there are converging data that se- closure (see Feeney, 1999, and Shaver &
curely attached children have had positive ex- Hazan, 1993, for reviews).1
periences when they have turned to others for What might be the childhood experiences of
care. And it is likely that these experiences such individuals? Ainsworth's home observa-
foster their capacity for secure attachment in
adulthood by making them comfortable and 1
confident in seeking care and by contributing to There is increasing evidence that there are two subtypes
of avoidant adults—avoidant-dismissing and avoidant-
positive mental representations of others as car- fearful—and that these subtypes are associated with differ-
ing and of themselves as worthy of care. Feeling ent patterns of functioning in a variety of areas (e.g., be-
lovable—reflecting a positive representation of havior with romantic partners and information processing;
the self—means that a person is unlikely to Bartholomew & Horowitz, 1991; Fraley, Davis, & Shaver,
1998). Because the majority of adult romantic attachment
engage in any of a number of behaviors destruc- research has been conducted with assessment procedures
tive to a healthy romantic relationship, such as mat identify a single dimension of avoidance, there is as yet
seeking endless reassurance from the partner of relatively little known about these subtypes. It is reasonable
his or her own worth, demanding excessive to predict that these subtypes are associated with different
developmental precursors. Yet, speculation about these pre-
closeness (or refusing any closeness) to ward cursors seems, at this point, not only limited by scarce data
off expectations of being abandoned, refusing but also theoretically problematic. The tripartite adult ro-
closeness for fear of losing the self by becoming mantic research tradition (which contains a single avoidant
intimate with another, allowing himself or her- group) is more closely linked theoretically to the conceptu-
alization of individual differences within the developmental
self to be mistreated, denigrating the partner to literature; that is, both conceive of individual differences
avoid feeling inferior, and insisting on continual along two dimensions that can be viewed as (a) security
proof of love and commitment. Secure individ- versus insecurity and (b) emphasis versus deemphasis on
uals thus bring a set of expectations into new attachment.
116 CASSIDY

tions of mothers and infants during the 1 st year it is in secure babies (Spangler & Grossmann,
identified a group of babies (later classified as 1993; Sroufe & Waters, 1977). Furthermore,
insecure-avoidant) whose bids for comfort whereas the heart rate of babies truly interested
were rejected (Ainsworth et al., 1978). Their in play typically slows down, the heart rate of
mothers were uncomfortable with close bodily avoidant babies does not slow down, suggesting
contact, and they tended to pick up their babies a lack of true focus on the play (Spangler &
in an abrupt and controlling manner. Mothers of Grossmann, 1993). Thus, it seems that these
these infants were also described as having infants are not shifting attention to the toys but,
restricted emotional expressiveness. Evidence rather, away from the mother. Bowlby bor-
from subsequent studies revealed additional rowed the term "diversionary activity" from the
ways in which the caregiving environment of ethologists to describe this behavior. Bowlby
these infants was associated with rejection and (1973; see also Ainsworth et al., 1978) proposed
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

restriction of emotional expression. For in- that sight of the mother might activate the at-
stance, an observational study of toddlers and tachment system (might activate seeking to be
mothers during free play revealed that mothers held or seeking to be comforted, which in the
of avoidant infants joined in play when the past has been met with painful rejection), and so
toddler was content but withdrew when the tod- the baby turns toward play activity, but without
dler expressed negative affect. The pattern was true interest.
reversed among mothers of securely attached Evidence that avoidant children have diffi-
infants, who were more likely to increase inter- culty seeking care also comes from several ad-
action with the toddler (in a facilitative way) ditional studies of young children. In one study
when the toddler expressed negative affect examining infant behavior in the strange situa-
(Grossmann, Grossmann, & Schwan, 1986). In tion, observations of the communication pat-
a study of mothers' emotion socialization prac- terns of mother-infant dyads revealed that
tices, mothers of toddlers who had been inse- avoidant infants communicated directly with
cure-avoidant in infancy reported, as expected, their mothers only when they (the infants) were
the greatest control of their toddlers' emotional feeling well. When distressed, these infants
expressiveness (Berlin & Cassidy, 2000; see tended not to directly signal the mother, and
also Malatesta, Culver, Tesman, & Shepard, they did not seek bodily contact (Grossmann et
1989). Infants in a caregiving environment in al., 1986). In a study of infants during the
which emotional expressiveness is restricted strange situation, avoidant infants did not make
and activation of their attachment system con- visual contact with the mother when they were
sistently leads to painful rejection might under- aroused; they did so only in situations with low
standably develop a strategy in which their at- emotional distress (Spangler & Grossmann,
tachment system would be activated as little as 1993). Findings from this study converge well
possible. Such a strategy would be adaptive to with findings from a study of attachment and
their circumstances (Main, 1990). emotion regulation (Braungart & Stifter, 1991).
There are data suggesting that certain infants Even though insecure-avoidant infants were
appear to actively suppress activation of the just as likely as some secure infants to be dis-
attachment system (i.e., have trouble seeking tressed during separation from their mother dur-
care). Some babies separated from their mother ing the strange situation, they were less likely to
in the laboratory do not seem distressed during seek comfort from (or even interaction with) the
separation and do not seek her for comfort or mother on reunion, engaging instead in self-
even interaction on reunion, as most infants do. soothing behaviors. In another study, avoidant
These infants instead actively avoid the mother children similarly masked negative affect, this
and become focused on toy play (and thus, on time in a social situation—a tower building
the attachment-exploration continuum, an em- game—with an adult stranger (Lutkenhaus,
phasis toward exploration and away from at- Grossmann, & Grossmann, 1985). Children's
tachment; Ainsworth et al., 1978). Yet, physio- affective reactions to winning and losing this
logical data are interesting. Heart-rate data gath- competitive building game were analyzed. The
ered during the strange situation reveal that the avoidant children manifested sadness about los-
attachment system is activated in these babies as ing during the game, but not when the game was
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 117

over and the experimenter was available for need for help; and, when there was a successful
social communication. In fact, during the social resolution, it was brought about entirely by the
communicative exchange, there was a tendency child. In the same study, avoidant children re-
for the avoidant children to replace sadness with sponded to a puppet interview with descriptions
smiling. In contrast, securely attached children of themselves as perfect in every way, even
showed their sadness after the game, when the when repeatedly pressed. This claim of perfec-
adult was more available. Thus, even though tion is viewed as a form of defensive idealiza-
avoidant children felt and displayed sadness, tion of the self by a child who fears rejection
they did so only when there was no child-adult should any imperfection be found (see Cassidy
eye contact, thereby precluding any comfort or & Kobak, 1988, for further discussion of avoid-
reassurance the adult might offer. ance and defensiveness).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

As Bowlby's claim that representations are In sum, there is converging evidence that
This document is copyrighted by the American Psychological Association or one of its allied publishers.

largely experience based would lead one to pre- children who have experienced maternal rejec-
dict, the attachment-related representations of tion and caregiving in which emotional expres-
avoidant children are not as positive as those of siveness was restricted, ignored, or discouraged
secure children. In one study, 3-year-olds were understandably form representational models
asked questions about stories in which a peer wherein others are not particularly comforting
caused something negative to happen to the or helpful. Furthermore, there are data suggest-
child (e.g., the child's toy was broken; Cassidy ing that attachment-related characteristics of
et al., 1996). Children were asked "What did adults classified as avoidant are foreshadowed
your mother say to you after this happened?" by characteristics of children classified as
The responses of children who had been classi- avoidant. Avoidant adults are not comfortable
fied as insecure-avoidant with the mother in depending on others (Hazan & Shaver, 1987);
infancy suggested that they, unlike children avoidant children have representations of others
who had been secure in infancy, were unable to as not particularly comforting or helpful
form representations of the mother as being (Cassidy, 1988; Cassidy et al., 1996) and have
helpful or comforting. Representations related difficulty remembering stories (in a laboratory
to separation and reunion also reflect less opti- task) involving mothers responsive to distress
mal experiences. In Main et al.'s (1985) study in (Kirsh & Cassidy, 1997). Avoidant adults are
which 6-year-olds were asked about photo- less likely to turn to others for care (Fraley &
graphs of separation, children who had been Shaver, 1998; Simpson, Rholes, & Nelligan,
avoidant infants were unable to suggest how 1992), as are avoidant infants (Braungart &
children might cope with separation (see also Stifter, 1991; Grossmann et al., 1986). Avoidant
Shouldice & Stevenson-Hinde, 1992). In an- adults emphasize their autonomy (Hazan &
other study, a similar pattern emerged for Shaver, 1987); avoidant children report less
avoidant 6-year-olds: Fears about separation loneliness than others (Berlin, Cassidy, & Bel-
were not expressed directly, and reunions were sky, 1995). Avoidant adults do not enjoy phys-
hampered by psychological unavailability of ical contact and cuddling within romantic rela-
family members or by delay and distraction tionships (Fraley et al., 1998); mothers of
(Solomon et al., 1995). avoidant infants are observed to be more un-
The representational model of the self of comfortable with close bodily contact than
avoidant children has also been found to be other mothers (Ainsworth et al., 1978); and
problematic. Cassidy (1988) used a doll stories 6-year-olds who had been avoidant infants draw
procedure to assess "children's representation family portraits containing people whose bodies
of the self in relation to attachment." The stories are tense and rigid and who often have no arms
of children classified as insecure-avoidant to (Kaplan & Main, 1985). Avoidant adults at-
mother at 6 years of age reflected representa- tempt (sometimes successfully) to suppress ac-
tions of the self as lacking a supportive relation- tivation of the attachment system (Fraley &
ship with the mother. The doll protagonist was Shaver, 1997; see also Dozier & Kobak, 1992),
isolated or rejected, and the importance of rela- and avoidant infants attempt (relatively unsuc-
tionships was denied; the existence of conflict cessfully) to do the same (Spangler & Gross-
was denied in stressful situations, as was the mann, 1993). Avoidant adolescents are more
118 CASSIDY

hostile than others (Cooper, Shaver, & Collins, tachment system might be rooted in a similar
1998); avoidant toddlers are rated by observers pattern begun in infancy and childhood. In her
as more aggressive than others (Main & Stadt- original home observation study, Ainsworth
man, 1981), and avoidant children are more (Ainsworth et al., 1978) observed that the moth-
likely than others to be bullies (Troy & Sroufe, ers of a small group of infants (later classified as
1987). Avoidant adults report that their mothers insecure-ambivalent) were inconsistent in their
were "cold and rejecting" (Hazan & Shaver, parenting: sometimes loving and responsive,
1987); avoidant infants have been observed to but only when they could manage, not in re-
have mothers who are more rejecting, emotion- sponse to the infant's signals (see Cassidy &
ally restricted, uncomfortable with close bodily Berlin, 1994, for a review of the maternal be-
contact, and insensitive than mothers of secure havior associated with this pattern). What
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

infants (Ainsworth et al., 1978). would be an adaptive strategy for an infant


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Insecure—ambivalent attachment. Adults whose mother was sometimes responsive but


classified as insecure-ambivalent, according to other times too preoccupied, too overwhelmed,
Hazan and Shaver (1987), endorse the follow- or too inept to respond? An adaptive strategy
ing description of themselves: "I find that others would be staying near, clinging, and monitoring
are reluctant to get as close as I would like. I closely the availability of the mother so that, if
often worry that my partner doesn't really love need for her does arise, the infant, who cannot
me or won't want to stay with me. I want to count on the mother to monitor his or her needs,
merge completely with another person, and this will have quick access to her (Cassidy & Berlin,
desire sometimes scares people away." These 1994; Main & Hesse, 1990). Main and Solomon
descriptions suggest that, despite high activa- (1986) stated that "in its heightened display of
tion of the attachment system, these individuals emotionality and dependence upon the attach-
do not experience others as sufficiently respon- ment figure, this infant successfully draws the
sive, and this activation does not lead to comfort attention of the parent" (p. 112). This heighten-
and satisfaction. There is evidence from their ing strategy can be viewed as adaptive if it
discussion of romantic relationships that these serves to gain the attention of an unavailable
adults prefer unqualified closeness, commit- caregiver. However, according to Bowlby
ment, and affection (Feeney & Noller, 1991), (1973), such a strategy can be dysfunctional if
and it seems, from a variety of studies, that they the negative emotionality is so pervasive that it
are particularly distressed by threats to the ro- threatens the existence of the relationship. It can
mantic relationship (e.g., Feeney, 1998; Feeney also be dysfunctional to the extent that it inter-
& Noller, 1992; Simpson, Rholes, & Phillips, feres with other developmental tasks such as
1996). This seeming insatiability for closeness exploration. Bretherton (1985) described this
can be considered a hyperactivation of the at- situation as the infant's having to take on more
tachment system that would be likely to con- than her or his share of the burden of maintain-
tribute to problems within a romantic relation- ing the relationship.
ship, because to expect someone else to fill all The strange situation behavior of infants clas-
of one's needs, to be utterly dependent on an- sified as ambivalent is characterized by extreme
other, and to be threatened by another's natural distress on separation and difficulty in calming
autonomy strivings are likely to contribute to on reunion; these infants display angry, resistant
ambivalence and resentment in both partners. behavior toward the parent (Ainsworth et al.,
Heightened emotionality is also associated with 1978). This heightened negative emotionality
this pattern: Ambivalent adults reported greater can be viewed as a component of the child's
distress and hostility during a laboratory prob- strategy to gain the mother's attention (Cassidy,
lem-centered discussion (Simpson et al., 1996); 1994; Main & Solomon, 1986). The negative
in another study, ambivalent adults described emotionality of the insecure-ambivalent child
themselves as experiencing greater affective in- may be exaggerated and chronic because the
tensity and emotionality in their daily lives than child recognizes that to relax and allow herself
did others (Pietromonaco & Feldman Barrett, or himself to be soothed by the presence of the
1997). attachment figure is to run the risk of then losing
A pattern of adult hyperactivation of the at- contact with the inconsistently available parent.
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 119

A history of exaggerated negative emotionality competence during toy play (e.g., Belsky, Gar-
may explain findings revealing that these chil- duque, & Hrncir, 1984; Matas, Arend, &
dren have trouble maintaining a boundary be- Sroufe, 1978), and greater fear and inhibition
tween another person's distress and their own during peer play (e.g., M. F. Erickson, Sroufe,
(Kestenbaum, Farber, & Sroufe, 1989). An ad- & Egeland, 1985; see Cassidy & Berlin, 1994,
ditional strategy may involve fearfulness in re- for a review). According to attachment theory
sponse to relatively benign stimuli. Through (Ainsworth et al., 1978; Bowlby, 1988), this
exaggerated fearfulness, the infant increases the relatively incompetent exploration would result
likelihood of gaining the attention of a fre- from turning attention away from the environ-
quently unavailable caregiver should true dan- ment and toward the mother, provoked by un-
ger arise (Main & Hesse, 1990). However, the certainty about her availability. Although few
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

critical research needed to demonstrate a height- studies have directly addressed the issue of in-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ening of negative affect—or, indeed, a height- fant attention to mother, several have shown
ened activation of the attachment system—by greater attention to characterize ambivalent ba-
ambivalent infants does uot exist. With the bies; no study has revealed discrepant findings
avoidant pattern, minimization of negative af- (e.g., Ainsworth, 1992; Dickstein, Thompson,
fect does exist; avoidant infants and adults can Estes, Malkin, & Lamb, 1984; see Cassidy &
feel negatively aroused while suppressing ex- Berlin, 1994).
pression of such feelings (albeit with varying In sum, there is converging evidence that
degrees of success; Dozier & Kobak, 1992; children who have experienced an inconsis-
Fraley & Shaver, 1997; Grossmann et al., 1986; tently available attachment figure and who thus
Lutkenhaus et al., 1985; Malatesta et al., 1989). may be unsure of access to her if needed attend
Comparable evidence in relation to the ambiv- to her more, are more clingy, and show less
alent pattern would consist of a demonstration competent exploration of the environment, toys,
that the ambivalent individuals show greater and peers. There is also evidence that some of
negative reactivity than they actually feel. The the characteristics of ambivalent adults charac-
operationalization of "heightening" has not yet terize ambivalent infants and children as well.
been attempted. Ambivalent adults seek emotional closeness but
A child with the experiences typical of am- do not find it satisfying (Hazan & Shaver,
bivalent infants might develop a representation 1987); ambivalent infants seek physical close-
of the mother as inconsistently available and of ness on reunion with the parent in the strange
the self as able to gain care only by sending situation but do not find it satisfying (i.e., they
exaggerated signals of need. Studies examining show resistance to contact; Ainsworth et al.,
the representations of young children have not 1978). Ambivalent adults describe desires to
yielded a consistent picture of the ambivalent merge with a partner (Hazan & Shaver, 1987);
group (which may be due in part to the small ambivalent children have difficulty with dis-
number of children typically classified in this tress-related boundary maintenance (Kesten-
group). One study producing results that can be baum et al., 1989). Ambivalent adults describe
interpreted as reflecting a representation of the themselves in relationships in ways classified as
mother as relatively unavailable was conducted "preoccupied" (Feeney & Noller, 1990); infants
by Solomon et al. (1995). In stories of child- are classified as ambivalent in the strange situ-
parent reunion, the reunions of ambivalent ation on the basis of their (unsatisfactory) pre-
6-year-olds were characterized by "delay and occupation with their mother (Ainsworth et al.,
distraction." 1978). Ambivalent adults report greater loneli-
The balance between attachment and explo- ness than others (Hazan & Shaver, 1987), as do
ration of ambivalent infants and children is not ambivalent children (Berlin et al., 1995). Am-
fluid: These children emphasize attachment at bivalent adults are particularly upset by rela-
the expense of exploration. There is convergent tionship breakups (loss of an attachment figure;
evidence that these children show exploratory Feeney & Noller, 1992); ambivalent infants are
deficits: restricted exploration of the general particularly upset by laboratory separation from
environment (e.g., Cassidy, 1986; Hazen & their mother in the strange situation (Ainsworth
Durrett, 1982), less focused attention and less et al., 1978). (Relatedly, 3-year-olds of mothers
120 CASSIDY

classified as "preoccupied" with the Adult At- fact, studies of caregiving in adult romantic
tachment Interview [George, Kaplan, & Main, relationships have revealed that the type of care-
1985] have been shown to exhibit greater anx- giving that secure adults provide to romantic
iety than other children when their mothers are partners is similar to the caregiving that secure
preparing to leave the room during a laboratory infants have received. For example, Kunce and
procedure [Crowell & Feldman, 1991].) Shaver (1994) found the dimensions of proxim-
ity (vs. distance), sensitivity (vs. insensitivity),
The Caregiving Behavioral System cooperation (vs. control), and compulsive care-
giving to differentiate adult attachment groups.
Giving care means being available to an adult The most compelling evidence for this claim
romantic partner in times of trouble, that is, that early care received contributes to later care
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

serving as a haven of safety in times of alarm. It given would emerge from longitudinal studies
This document is copyrighted by the American Psychological Association or one of its allied publishers.

means being able to recognize when the person examining individual differences in the caregiv-
needs care and doing what it takes to provide it. ing within romantic relationships of adults
Giving care means being loving: being respect- whose infant attachment classifications were
ful of the truth of another, accepting of a range known. There are as yet no prospective longi-
of ways of being and feeling. It involves open- tudinal data, although infants observed in sev-
ness, flexibility, and acceptance. Giving care eral samples are soon reaching the age when
also involves serving as a secure base from they will become involved in long-term roman-
which the partner can explore: being comfort- tic relationships (e.g., studies conducted by
able with and accepting another's natural Egeland and Sroufe, Grossmann and Gross-
wishes for exploration. There is evidence from mann, and Main). Data from these studies will
several studies that secure adults are more likely provide important insights.
to give sensitive care to their partners than are There are already a few studies that provide
insecure adults (Carnelley, Pietromonaco, & suggestions that childhood security may be as-
Jaffe, 1996; Collins & Feeney, in press; Fraley sociated with a later ability to provide care.
& Shaver, 1998; Rholes, Simpson, & Orina, First, a series of studies has shown that securely
1999; Simpson et al., 1992). attached children behave in ways that reflect
How does the ability to give care develop? skills related to positive caregiving. For in-
Bowlby, focusing on adult caregiving to infants, stance, Main (1981) observed toddlers playing
described caregiving as developing partly in a laboratory when a clown entered and cried
through the experience of having been cared in distress. Toddlers securely attached to their
for. This idea is widely held. As Fraiberg (1980) mother showed more "concerned attentiveness"
aptly noted, "We do unto others as we were to the crying clown than did insecurely attached
done to." Similarly, Sroufe and Fleeson (1986) children. Other studies have produced evidence
pointed out that babies learn both sides of im- that secure children are more prosocial, less
portant early dyadic relationships; at the same aggressive, and less hostile (e.g., Lyons-Ruth,
time they learn what it is like to be a baby in a Alpern, & Repacholi, 1993; Renken, Ege-
relationship, they observe how it is to be a land, Marvinney, Mangelsdorf, & Sroufe, 1989;
caregiver. According to attachment theory, Suess, Grossmann, & Sroufe, 1992; see Belsky
adults who, as infants, had experiences with a & Cassidy, 1994, and Weinfield, Sroufe, Ege-
sensitive, responsive caregiver will give such land, & Carlson, 1999, for reviews). Still other
care to their own infants. Conversely, adults data suggest that secure children are more em-
who had mostly negative experiences as infants pathic than insecure children (Kestenbaum et
will be at higher risk for being insensitive to al., 1989). Second, studies of sibling relation-
their own infants, even though the clinical and ships offer insight into attachment security and
empirical literatures are clear that this is not an caregiving. Tn one study, preschool children se-
inevitable pathway (Pearson, Cohn, Cowan, & curely attached to their mother were more likely
Cowan, 1994; Phelps, Belsky, & Crnic, 1998). than others to soothe and give care to a toddler
It is reasonable to expect that some of these sibling who was distressed during parental sep-
predictions about caregiving of children would aration (Teti & Abalard, 1989). In a longitudinal
also apply to caregiving of romantic partners. In study, 6-year-olds who had been secure infants
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 121

were less aggressive to siblings than were chil- cesses and behavior contribute to continuity
dren who had been insecure infants (Veiling & across development.
Belsky, 1992). First, representational models are thought to
guide expectations of the availability and prob-
able responsiveness of others. According to
Continuity and Discontinuity: Moving Bowlby, "the kinds of experiences a person
From Childhood Attachment to Adult has, especially during childhood, greatly affect
Romantic Attachment . . . whether he expects later to find a secure
personal base, or not" (1979, p. 104). Expecta-
Attachment theory contains predictions about tions that others will be responsive if ap-
both continuity and discontinuity of individual proached increase the likelihood that an indi-
differences across development. Bowlby (1979) vidual will feel comfortable with closeness and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

was clear about the extent to which he believed with seeking care, both characteristics of secure
This document is copyrighted by the American Psychological Association or one of its allied publishers.

in a model of continuous development: "There adult attachment.


is a strong causal relationship between an indi-
Second, representational models are thought
vidual's experiences with his parents and his
to guide attributions of the behavior of others in
later capacity to make affectional bonds" (p.
ambiguous situations. In close relationships,
135). More specifically, Bowlby predicted that
there are many times when the cause of a neg-
an individual securely attached in childhood
ative situation is unclear. An example is a per-
would be likely to form secure romantic attach-
son left waiting for a partner who never arrives
ments in adulthood. Yet, Bowlby (1973, 1980)
for a date. There are infinite possible attribu-
was also clear that individual adaptation is al-
tions that the waiting person can make (e.g., he
ways a product of developmental history and
got stuck in an elevator; he was delayed in
current circumstances and that change is always
traffic; he forgot the date; he found something
possible. In this section, Tfirstdiscuss continu-
he preferred doing; he wanted to humiliate me),
ity and describe the mechanisms through which
and which of these is chosen is influenced partly
early attachment experiences are thought to
by the person's representational models of what
contribute to later relationships with romantic
others are like. The interpretation selected, in
partners. I then discuss discontinuity and de-
turn, is likely to influence the behavioral re-
scribe factors from both childhood and adult-
sponse and thus the quality of the relationship
hood that may contribute to a lack of predict-
(see Collins, 1996, for evidence of links be-
ability from early attachment to romantic
tween adult attachment style and attributions
attachment (for additional discussion of conti-
within close relationships).
nuity, see Fraley & Shaver, 2000).
Third, representational models are also
thought to guide attention and memory (Main et
Continuity al., 1985), which in turn structure incoming
information and experience so that they match
For Bowlby (1973), the construct of repre- existing models. The nature of the representa-
sentational models provides the foundation for tional models that secure children develop—of
understanding continuity across development. the self as competent in eliciting care from
As noted earlier, Bowlby believed these models responsive others—means that they can regu-
of others, the self, and relationships to be based late attention relatively flexibly across a variety
largely on repeated daily experiences with par- of environmental stimuli (an attachment figure,
ents. According to Bowlby, these experience- toys, or peers) depending on actual, present
based models play central roles in guiding the conditions. Their confidence that help will be
creation and nature of new relationships. This available if needed allows them to turn attention
idea that existing mental structures are used to to exploration in a calm environment and,
create new ones is compatible with Piaget's should danger or threat arise, to turn attention to
(1955) ideas about the child's "construction of the attachment figure.
reality." Bowlby set forth a series of proposi- Insecure attachment, conversely, can be
tions about how the influences of these repre- viewed as selective inattention. The task of
sentational models on cognitive-affective pro- avoidant children is to maintain sufficient prox-
122 CASSIDY

imity to elicit care if truly needed from an In a memory task in which children were told
attachment figure known to reject attachment stories depicting various maternal responses to a
behavior (Main, 1981). Avoidant children are child's bids for care after a minor injury, these
thought to shift attention away from stimuli that children did not have the bias for remembering
might activate the attachment system, because, responsive child-mother interactions common
in the past, such activation was met with rejec- to other children. In another study, 6-year-olds
tion (Ainsworth et al., 1978). A history of such who had been avoidant infants were found to
repeated shifts in attention away from attach- shift attention away from a photograph of them-
ment-related stimuli might lead an individual to selves and their parents (Main et al., 1985).
have difficulty viewing a partner as a secure Insecure-ambivalent children, conversely,
attachment figure in adulthood. Such a history are thought to shift attention toward attachment-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of attentional shifts might also interfere with the eliciting situations; thus, such children may, for
This document is copyrighted by the American Psychological Association or one of its allied publishers.

caregiving component of a romantic attach- instance, selectively attend to the frightening


ment; sensitive attention to the attachment aspects of the environment, interpreting "an en-
needs of another may not be possible (e.g., vironment known at some level to be quiescent
awareness of a wish for comfort in another as threatening" (Main, 1990, p. 61). A history of
would necessitate an attentional shift toward shifting attention toward attachment-related
attachment that might awaken the pain of one's stimuli might lead to far-reaching negative ef-
own childhood rejection; e.g., Fraiberg, 1980). fects (e.g., leaving the person so focused on
Fraley, Davis, and Shaver (1998) have delin- attachment that he or she not only incessantly
eated an elegant model that addresses the pro- demands care but also cannot focus on another
cess through which an avoidant infant might person's needs for care or on other aspects of
become an avoidant-dismissing adult. This the environment that require considerable atten-
model explains continuity in the context of de- tion). Furthermore, this long-practiced tendency
velopmental change. As the authors pointed out, may lead a person to be uncomfortable with the
there is evidence that infant avoidance is a de- autonomy (of both the self and the partner)
fensive attempt, albeit a partly unsuccessful required for true intimacy (E. Erickson, 1950;
one, to prevent activation of the attachment Reiss & Patrick, 1996). It may be that, through
system. In adults, however, there is evidence regulation of cognitive and affective processes,
that avoidant-dismissing individuals are more the neural networks that ambivalent individuals
successful in dampening activation of the at- develop are "supersaturated11 with synaptic con-
tachment system. The authors proposed a devel- nections with the attachment system so that a
opmental progression wherein relatively unsuc- relatively large array of stimuli may activate
cessful attempts in infancy become increasingly connections with this system.
effective. They proposed that through atten-
It is widely held that—within childhood—
tional shifts and a variety of cognitive pro-
another substantial force for continuity is sta-
cesses, avoidant individuals develop neural net-
bility of the environment (e.g., Lamb, 1984).
works that have relatively few synaptic connec-
For instance, the sensitive parent of an infant is
tions with the attachment system (because of
thought to remain the sensitive parent of a tod-
motivation to keep it, in essence, separate in the
dler, supportive of later developmental chal-
brain). Fraley and Shaver (1997) found empir-
lenges. If the environment changes whereby a
ical support for their proposition: In two exper-
previously sensitive caregiver becomes less
imental procedures, avoidant-dismissing adults
sensitive, then attachment quality is expected to
were, in fact, able to use defensive strategies to
change, and in fact there is evidence that this is
successfully disengage certain components of
the case (M. F. Erickson et al., 1985; Vaughn,
their attachment system. Kirsh and Cassidy
Egeland, Waters, & Sroufe, 1979). Just as sta-
(1997) reported evidence of ways in which both
bility of the environment is thought to contrib-
attention and memory processes in avoidant
ute to stability of attachment within childhood,
children may foster this developmental path-
so might stability of the environment contribute
way. When shown attachment-related stimuli,
to stability of attachment from childhood to
3-year-olds who had been avoidant infants
adulthood. Although children do not select their
turned attention away more than other children.
parents, by adulthood, individuals are very
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 123

much active contributors to creating their own Figure 1 presents a summary model of some
attachment environments (i.e., to selecting their of the processes involved in continuity from
own romantic partner), and these selections may childhood to adulthood attachments. Represen-
contribute to the stability of the environment. tational models, based on experiences with
The following are important questions that need childhood attachment figures (Path a), guide
further exploration: To what extent do early cognitive-affective processes (attention, mem-
attachments to parents influence selection of ory, expectations, and attributions; Path b),
partners? and To what extent do early attach- which in turn guide behavior (Path c). Behavior
ments to parents influence the quality of rela- is thought to guide treatment received from
tionships with those partners? Insight into these others (Path d). This treatment, in turn, contrib-
questions emerged from a study conducted by utes to representational models (Path e), usually
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Collins and Read (1990, Study 3). Adults' de- in a reinforcing way. This model is useful for
This document is copyrighted by the American Psychological Association or one of its allied publishers.

scriptions of their childhood relationship with understanding how the behavioral responses to
opposite-sex (but not same-sex) parent were many typically occurring relationship situations
related to the way their partner described him- are likely to be (more or less) influenced by
self or herself in attachment-related terms. attachment-related representations, based in
Women who described their fathers as having part on childhood experiences.
been warm and responsive were more likely to For example, an individual's experiences of a
be dating men who felt they could depend on sensitively responsive mother will contribute to
others and (marginally) men who felt comfort- the formation of a representation of her as sen-
able getting close. Men who described their sitively responsive and, through a process of
mother as cold or inconsistent were more likely generalization, to such a representation of oth-
to be dating women who described themselves ers more broadly (Path a). Having such a rep-
as anxious about being abandoned. resentation within an adult romantic relation-

Cognitive/affective processes
(attention, memory, expectations, attributions)

Experiences ^.Representational Models Behavior Towards Others

Treatments
Received From
Others
Figure 1. Processes involved in continuity from childhood to adulthood attachments.
124 CASSIDY

ship means that an individual is likely to have not the child's only attachment figure. During
positive expectations of receiving care (Path b), the 1st year, most children become attached to
and these expectations guide behavior (Path c; one or a few additional people (typically the
e.g., comfortable seeking of care when needed; father or a familiar caregiver; Goosen & van
Fraley & Shaver, 1998); behavior contributes to Uzendoorn, 1990), and these attachments will
treatment received from the partner (Path d; be influential. Are these additional attachments
e.g., the partner's providing comfort; Kobak & as influential as the attachment to the mother?
Hazan, 1991); and treatment received contrib- Do different attachments influence the child in
utes to the representational model (Path e; e.g., different ways? What happens when one attach-
validation of the existing representational ment figure suggests that the child is lovable
model: "My partner provides care when I need and valuable and another suggests that the child
is of little value? Do these conflicted views
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

it"). Similarly, during a negative ambiguous


This document is copyrighted by the American Psychological Association or one of its allied publishers.

situation (e.g., the partner failed to do some- become integrated into one representational
thing she or he promised), models of a secure model of the self and others? Tf so, through what
individual might contribute to an attribution that process, and when does this occur? Can secure
would lead to positive rather than destructive attachment to one parent serve as a buffer in
behavior; this positive behavior would contrib- light of an insecure attachment to the other
ute to being well treated by the partner and parent? Are cross-sex relationships (i.e, father-
confirmation of the positive representational daughter and mother-son) particularly impor-
model. In a conflict situation, the representa- tant for influencing (heterosexual) romantic
tions of a secure individual might mean that he relationships?
or she views the conflict within a context of a Moreover, it is important to remember that
committed relationship wherein the partner is the attachment component of the parent-child
loving and the self is lovable; this view may relationship is not the only component of that
influence behavioT within the conflict, and there relationship likely to influence children's ro-
is evidence that secure individuals feel better mantic relationships. The connection between
after a laboratory conflict task than do insecure attachment and nonattachment components of
individuals (Simpson et al., 1996). child-parent relationships is important to con-
sider. The attachment component of the rela-
Discontinuity tionship deals specifically with the parent's
serving as a secure base for the child. But par-
In this section, I examine factors related to ents are not only attachment figures for their
discontinuity between early attachment experi- children. They are also playmates, teachers, dis-
ences and later adult romantic relationships. ciplinarians, and care providers, and they influ-
The theoretical and clinical empirical literatures ence their children in these roles. A parent's
concur that children with insecure attachment ability in one of these roles is at least somewhat
histories and even with histories of abuse can independent of his or her abilities in the other
have successful adult romantic attachment rela- roles. What are the implications of these differ-
tionships. In fact, because of the large number ent types of parenting? Which of these compo-
of possible influences on adult romantic devel- nents of parenting would have greater influence
opment and the complexities of personality de- on children's later relationships? In sum, we
velopment, one would expect only modest as- need to learn more about how much the child's
sociations between early and later attachments attachment-related experiences, as compared
in any case. I examine two sets of factors that with the child's other experiences with the par-
could account for discontinuity between early ent, influence romantic relationships.
attachment and adult romantic attachment: Relationships with other family members
(a) factors in childhood and (b) factors in also may be important. In some families, a
adulthood. grandparent or an aunt may serve as the child's
Childhood. What sorts of childhood factors principal attachment figure. Or the child may
could account for discontinuity between child have important nonattachment relationships
attachment to mother and adult romantic attach- with other relatives within which the child
ment? To begin with, of course, the mother is learns skills relevant to romantic relationships.
SPECIAL ISSUE: ADULT ROMANTIC ATTACHMENTS 125

Relationships with siblings may be particularly and former romantic partners may contribute to
important because, like romantic relationships, representations of attachment figures and may,
sibling relationships are more egalitarian in in fact, override the influence of representations
structure than are child-parent relationships generalized from representations of the parent.
(Dunn, 1985). Observation of the parents' mar- Thus, if a partner truly has hostile intent, if a
riage is also likely to be important. This is partner truly is likely to be hurtful or likely to
typically the first romantic relationship the child abandon the relationship, even an individual
experiences. Just as children learn what it who had been securely attached is likely to
means to be a parent by participating in the develop negative representations (see Figure 1,
child-parent relationship (Sroufe & Fleeson, Path e) that will influence both the current and
1986), it is likely that they also learn, in part, subsequent romantic relationships. It may also
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

what it means to be a romantic partner by be that insecure individuals with negative rep-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

closely observing the parents' marriage. This is resentations develop more positive representa-
a case in which components of parental behav- tions if they are well treated by a partner. For
ior unrelated to the child directly influence the instance, a person may have negative working
child's later relationships; in other words, the models based on negative childhood experi-
child's observation of an attachment relation- ences and may become involved in a relation-
ship of which he or she is not a member may ship with a person who is loving and responsive
influence his or her own later attachment (rather than, for example, controlling and reject-
relationships. ing, as the mother had been), and these experi-
Children's wider social network of relation- ences may change the working model of others
ships outside the family is also likely to influ- and of the self. This scenario seems less likely,
ence their representational models. Experiences however, given that individuals with negative
with peers are likely to be particularly important representations may behave in ways that reduce
because, along with sibling relationships, these the likelihood that they will become well
are the first (relatively) egalitarian relationships treated. The therapeutic relationship may also
(as opposed to the more hierarchical parent- bring about changes in working models, in part,
child relationship), and children learn much according to Bowlby (1988), because the ther-
about compromise, negotiation, cooperation, apist serves as a new attachment figure who
victory, and defeat from a different perspective directly and indirectly helps the individual re-
(Collins & Sroufe, 1999; Furman, 1998). It will work old, problematic models.
be interesting to explore whether and how chil- Given how greatly working models of the self
dren's nonattachment relationships influence can influence the quality of an adult romantic
their later attachment relationships. relationship ("I am lovable" or "I am competent
Adulthood. In the remainder of this article, in eliciting care from others" vs. "I am unlov-
I address one set of factors contributing to dis- able" or "People will leave me"), any change to
continuity: experiences in adulthood that mod- working models of the self may contribute to
ify representational models of the self and oth- change in the nature of the romantic attachment
ers. When models developed within early at- relationships in which a person engages. Not
tachment relationships change, it may be that only can new relationships change working
current rather than earlier models are more in- models of the self, but situations and events
fluential on behavior in new relationships. Fur- outside of relationships can contribute to
thermore, change can occur not only between change as well. A person's newly discovered
childhood and adulthood but within adulthood talents and successes, for instance, may contrib-
as well (see Davila, Burge, & Hammen, 1997, ute to a representation of the self that enhances
and Hazan & Hutt, 1993). A number of factors adult attachments. An aspect of the working
in adulthood can contribute to change in these model of the self that is particularly likely to
models. influence romantic attachment, and is likely to
Experiences of being in new relationships, of emerge largely after childhood, is the view of
forming new attachments, can bring about one's desirability as a romantic partner ("I am
changes in working models of the self and oth- attractive to potential romantic partners" or "I
ers. For instance, experiences with both current can be a good romantic partner"). Emergence of
126 CASSIDY

this model may relate to many factors, including starting at birth (e.g., Bell & Harper, 1977;
experiences of being pursued as a romantic part- Swann, Stein-Seroussi, & Giesler, 1992), con-
ner, experiences within a relationship, and self- tribute to creating their own environments; spe-
perceptions of personal attributes (physical cifically, there is evidence that individuals' ear-
attractiveness, power, money, and emotional lier attachment history influences the sort of
health). treatment they elicit from others (Elicker, En-
A set of findings within the research literature glund, & Sroufe, 1992; Sroufe & Fleeson,
based on the Adult Attachment Interview 1986). Within a romantic relationship, an indi-
(George, Kaplan, & Main, 1985) is relevant vidual may be influenced by early attachment to
here. There are some individuals whose "cur- behave in ways that elicit, for instance, rejection
rent state of mind with respect to attachment" is or withdrawal from the current partner. The
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

classified as secure, yet whose childhood expe- attachment environment within which the indi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

riences are considered by interview coders to vidual then finds herself or himself is a product
have been largely negative. These individuals not only of the current treatment she or he
are able to provide their children with a secure receives but also of early experiences that con-
base, and their children are securely attached to tribute to eliciting that treatment. Further adding
them (Pearson et al., 1994; Phelps et al., 1998). to the complexity of understanding continuity
(This is not true for individuals with negative and discontinuity is the likelihood that factors
childhood experiences whose current state of contributing to each are likely to be different for
mind with respect to attachment is classified as different people. That is, some people are more
insecure.) Thus, there are cases in which the likely to be continually influenced by early ex-
commonly found connection between a parent's periences, and some are more likely to be open
negative childhood experiences and his or her to change. These variations contribute to the
child's insecure attachment does not hold. It is rich diversity of developmental trajectories.
interesting to speculate whether a secure "cur-
rent state of mind" would also help alter a
connection between insecure childhood attach-
ment and insecure adult romantic attachment. References
As Berlin and Cassidy (1999) discussed, there is Ainsworth, M. D. (1963). The development of infant-
much to be learned from examining ways in mother interaction among the Ganda. In B. M.
which relations among relationships can be Foss (Ed.), Determinants of infant behavior II (pp.
changed. 67-112). London: Methuen.
Ainsworth, M, D. (1989). Attachments beyond in-
fancy. American Psychologist, 44, 709-716.
Summary
Ainsworth, M. D. (1992). A consideration of social
referencing in the context of attachment theory and
The nature of the interplay of factors contrib-
research. In S. Feinman (Ed.), Social referencing
uting to continuity and those contributing to and the social construction of reality (pp. 349-
discontinuity is inherently complex. In this sec- 367). New York: Plenum.
tion, I have described ways in which adult ro- Ainsworth, M. D., Bell, S., & Stayton, D. (1971).
mantic attachments may be related to both pre- Individual differences in strange situation behavior
vious experiences and current circumstances, of one-year-olds. In H. R. Schaffer (Ed.), The
treating these two sets of factors as independent origins of human social relations (pp. 17-52).
when, in fact, they are not. As Sroufe and his New York: Academic Press.
colleagues (Sroufe, Egeland, & Kreutzer, 1990) Ainsworth, M. D., Blehar, M., Waters, E., & Wall, S.
have pointed out, "While in a researcher's mind, (1978). Patterns of attachment. Hillsdale, NJ: Erl-
past experience and current environment may baum.
be independent sources of variance, whose rel- Baldwin, J. (1897). Social and ethical interpretations
ative power may be assessed in a linear analy- in mental development. New York: Macmillan.
Baldwin, M. W. (1992). Relational schemas and the
sis, in reality children play a large role in cre- processing of social information. Psychological
ating their own environments, in accord with Bulletin, 112, 461-484.
past experiences" (p. 1370). There is evidence Bartholomew, K., & Horowitz, L. (1991). Attach-
of the extensive ways in which individuals, ment styles among young adults: A test of a four-

You might also like