6

Attachment Processes and Commitment

to Romantic Relationships
Hillary J. Morgan and Phillip R. Shaver

INTRODUCTION

The premise of this chapter is that it is impossible to understand commitment to romantic re-
lationships unless one considers how the attachment system affects the processes of falling in
love and choosing a mate. Most theorists account for commitment to romantic relationships
with economic models predicting commitment from the costs and benefits of the relationship
(e.g., Rusbult, 1980, 1983). Although this approach has made a large contribution to our un-
derstanding of commitment, in the present chapter we consider the possibility that commit-
ment to romantic relationships cannot be fully captured by cognitive models, because romantic
commitments are due in part to what Bowlby (1969/1982) called the attachment system: a non-
rational, inborn behavioral system designed by evolution to maintain important social ties. In
Western culture, a romantic commitment is almost always made only after a couple has begun
to feel attachment-related feelings toward each other. Yet because the attachment system is
nonrational, commitment to romantic relationships cannot be equated with highly cognitive
commitments, such as the decision to select a particular stock portfolio. In fact, the decision to
love someone is hardly a "decision" at all. Most people experience falling in love as an auto-
matic and largely uncontrollable process. Loving leads to the experience of "heartfelt" com-
mitment: an uncontrollable devotion to a particular person that might persist even against one's
will or better judgment. While people can choose, for rational reasons, not to enter a relation-
ship with someone they love, they may still feel drawn to that person and struggle emotionally
against their decision for years. Similarly, people may decide rationally that a particular person
would be a good mate, yet this alone will not lead them to experience feelings of love
(Baumeister & Wotman, 1992).
Once people fall in love, their feelings and experiences will reflect several nonrational
attachment-related mechanisms. These mechanisms can lead romantic partners to experience

Hillary J. Morgan and Phillip R. Shaver' Department of Psychology, University of California, Davis, Davis,
California 95816.
Handbook ofInterpersonal Commitment and Relationship Stability, edited by Jeffrey M. Adams and Warren H. Jones.
Kluwer AcademiclPlenum Publishers, New York, 1999.

109
110 HILLARY J. MORGAN AND PHILLIP R. SHAVER

puzzling feelings and to act in apparently self-defeating ways. While rational theories appear
consistent with well-functioning relationships in which partners are in love and satisfied, and
can voice the many ways they value their partners, these theories have more difficulty ac-
counting for why people report love for their spouses when their relationships are costly,
painful, or even life threatening.
This chapter is organized as follows: First, we provide a brief overview of attachment the-
ory and explain how it can account for the experience of commitment in romantic relation-
ships. Next, we discuss the important, systematic, individual differences in attachment style
that have been identified in children and adults, and show how these differences are related in
predictable ways to forms of commitment. Finally, we analyze the perplexing issue of why
people in abusive relationships maintain feelings of love and commitment. We argue that at-
tachment theory, compared with other models of commitment, is particularly useful in ac-
counting for this puzzling phenomenon.

General Attachment Processes

Attachment theory was originally formulated by Bowlby (196911982, 1973, 1980) to ex-
plain why young children are distressed by separation from their primary caregivers.
Attachment theory posits an evolutionarily adaptive system, the attachment behavioral system,
that causes human infants to seek proximity to a familiar caregiver (the attachment figure), feel
more psychologically secure in that person's presence, and protest vigorously when proximity
to that person is impeded. In this way, evolution has equipped infants with various mechanisms
to increase their potential for survival in a world requiring the care and protection of what
Bowlby called a "stronger, wiser" other.
The attachment system is characterized by several features, including the following:

Separation Protest
Crying, whining, throwing angry tantrums, and so on, in response to separation from an
attachment figure.

Secure Base
Normal attachment provides a sense of stability that enables the attached individual to ex-
plore the environment more freely and develop interests and skills more readily than if the re-
lationship were troubled or threatened. According to Bowlby, "Human beings of all ages are
happiest and able to deploy their talents to best advantage when they are confident that, stand-
ing behind them, there are one or more trusted persons who will come to their aid should dif-
ficulties arise" (1979, pp. 103-104).

Monotropy
Although most children and adults have more than one attachment figure, they usually
have a preference for one specific person. This primary attachment bond is very difficult to
break. Ainsworth (1991) echoes Bowlby in the following statement: "Let me define an 'affec-
tional bond' as a relatively long-enduring tie in which the partner is important as a unique in-
dividual, interchangeable with none other" (p. 38). Weiss (1991) adds: "Attempts to substitute
other figures fail, no matter how solicitous or caring the other figures may be" (p. 66).
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 111

Endurance of the Bond

Once in place, the attachment bond is not easily broken or reconsidered. According to
Weiss (1991): "Attachment seems to persist in absence of reinforcement" (p. 67). Children do
not habituate to their attachment figures and do not begin to prefer someone else on the basis
of novelty or kindness. Unlike nonattachment bonds, which dwindle when a person is mis-
treated or becomes less interested in his or her partner, once attachment bonds are in place,
they are relatively insensitive to the behavior of the attachment figure.

Lifelong Relevance
Attachment behavior "is a characteristic of human nature throughout our lives-from the
cradle to the grave" (Bowlby, 1988, p. 82). The attachment system remains intact in adults and,
when activated, results in behaviors functionally similar to those exhibited by infants and chil-
dren (Hazan & Shaver, 1987). However, one difference emerges as people develop:
Attachment to caregivers is gradually overshadowed by attachments to romantic partners. In
adulthood, the attachment system functions to maintain romantic partnerships much as, earlier
in life, it maintained ties to caretakers. Just like infants, adults are primed to select one special
figure, and to develop enduring emotional ties to that person. Adults who have fallen in love
and entered a romantic relationship will not easily leave their partners, even if their relation-
ships are less than ideal.
The applications of attachment theory to the study of infant and child development have
been reviewed in several anthologies (e.g., Bretherton & Waters, 1985; Cassidy & Shaver, 1999;
Greenberg, Cicchetti, & Cummings, 1990; Parkes, Stevenson-Hinde, & Marris, 1991) and will
not be discussed in detail here. In recent years, the theory has been expanded to apply to adoles-
cent and adult romantic relationships (see review by Shaver & Clark, 1994; also Feeney, 1999).
Viewed from an attachment perspective, romantic love is a complex process involving at least
three of the behavioral systems discussed by Bowlby: attachment, caregiving, and sex (Shaver,
Hazan, & Bradshaw, 1988; Shaver, Morgan, & Wu, 1996). Romantic partners often serve as
each other's caregivers and primary attachment figures, exchanging these complementary roles
when conditions require it. They also serve as each other's sexual partners, and in fact are likely
to have become attached while acting on sexual attraction. Theoretically, this arrangement makes
use of the attachment behavioral system in a second evolutionarily adaptive way: increasing the
likelihood that parents will stay together long enough to enhance the viability of their off-
spring-the products of their sexual attraction (Hazan & Zeifman, 1994, 1999; Lykken, 1995).
Many researchers studying romantic relationships have found that predictions based on at-
tachment theory, several of them previously tested only with respect to infant-caregiver rela-
tionships, are borne out in studies of adolescents and adults. Because most productive research
ideas about adult romantic attachment have emerged from careful consideration of attachment
dynamics in infancy, it is worthwhile to consider how we might think about commitment in the
infant-caregiver relationship before we consider commitment in adult romantic relationships.

"COMMITMENT" IN INFANCY

Infants in attachment relationships are highly committed to their attachment figures in a

nonrational, heartfelt sense. Almost every infant will become attached to at least one person,
usually the most familiar caregiver or regular social interaction partner. Once "committed,"
children will prefer their attachment figures to other caregivers, even if alternative caregivers
are objectively more sensitive and reliable.
112 HILLARY J. MORGAN AND PHILLIP R. SHAVER

Attached or "committed" children cannot easily detach from their primary attachment
figures. The process of breaking an attachment in childhood can occur only if the child is
separated from the caregiver, and may take months or even years to complete. The breaking
of an attachment occurs in a three-stage process described by Bowlby (1980): (1) Protest-
The child attempts to regain proximity by crying and calling for the attachment figure and
angrily berating him or her for threatening abandonment; (2) Despair-when prolonged
protest fails to effect a reunion with the attachment figure, the child will experience de-
spondency and depression; (3) Detachment-if the separation is long enough, the child will
eventually break the attachment. The detached child appears to have returned to normal, re-
gaining interest in the environment and no longer appearing to be grieving. However, when
detached children are reunited with their caregivers, they show active disregard. The attach-
ment, in essence, has been defensively broken. Only if the attachment figure makes a pro-
longed effort to exhibit care and availability will a child open up to "reattachment." If
children begin to reattach, they are likely to be vigilant, anxious, and clingy for an extended
period of time.

ATTACHMENT PROCESSES IN ADULTHOOD

Attachment in adulthood exhibits many of the same features as attachment in infancy.

One important parallel is that attachment~related feelings are not under rational control. In the
case of loss, even if one knows--cognitively, rationally-that an attachment figure has died
and is no longer available, one can still long intensely for that person, imagine hearing or see-
ing him or her, and imagine having a conversation with the person between earth and heaven
(Hansson, Berry, & Berry, Chapter 16, this volume; Weiss, 1991). The need for the lost person
is unlike the need for a lost umbrella, a wish that an investment had worked out better, or the
incorrect assumption that one's latest research project would succeed. Attachments are in some
sense deeper, more primal, and less open to argument and rational thought.
Just as infants will attach to their caregivers regardless of their caregivers' competence,
adults generally cannot weigh the pros and cons of potential partners and fall in love accord-
ingly. People may love someone whom they know is not good for them, and may be unable to
fall in love with someone they think is a good match. Baumeister and Wotman (1992) present
compelling examples of people who experience distress because they cannot fall in love with
a willing partner whom they believe, "rationally," they should love. This is one area in which
Zajonc (1980) was certainly correct when he said that "preferences need no inferences"; that
is, preferences do not seem to be due to a totaling up of separate cognitive evaluations or a cool
comparison of pros and cons.
Whom we love is oddly unpredictable, depending, it seems, on chance encounters, mo-
mentary excitement, small acts of interest or kindness, even haphazard propinquity. As in in-
fancy, proximity plays a crucial role in the selection of friends and attachment figures. People
who live or work near each other and share common spaces such as lounges, parks, and restau-
rants are more likely to become friends and lovers than people who live or work slightly far-
ther apart (Kerckhoff, 1974; Nahemow & Lawton, 1975). Lykken (1995) describes how
identical twins, despite their astounding physical and psychological similarities, tend not to fall
in love with highly similar people. Moreover, people who fall in love with one twin tend not
to fall in love with the other. Lykken further describes how his colleague TeUegen "pointed out
that human romantic infatuation thus resembles the phenomenon of imprinting in geese and
other precocial birds and ... speculates that we evolved this susceptibility to infatuation be-
cause it tended to sustain the bond long enough for the more enduring bonds of companionate
affection to mature" (p. 56).
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 113

INDIVIDUAL DIFFERENCES IN PATTERNS OF ATTACHMENT

Ainsworth, Blehar, Waters, and Wall (1978) identified three major patterns of infant at-
tachment: secure, avoidant, and anxious. (The names of these patterns have been shortened
here for convenience.) The three patterns have been traced to systematic differences in parent-
ing, and, to some extent, to differences in infant temperament. Secure infants explore novel en-
vironments in the presence of their attachment figures, become moderately distressed when
their attachment figures leave them briefly, but return to a positive affective state when they re-
turn. Home observations indicate that this pattern occurs in relationships in which the attach-
ment figure shows sensitive and responsive caregiving. Avoidant infants explore novel
environments, though perhaps not as creatively as secure infants, and outwardly do not appear
upset when their attachment figures leave. Upon reunion, these infants show little joyful re-
sponse and may actively turn away from their attachment figures. Although avoidant infants
appear unruftled in the Strange Situation, heart-rate monitors indicate that they are physiolog-
ically aroused-they are simply not showing any overt protest (Sroufe & Waters, 1977). Home
observations indicate that the primary caregivers of avoidant infants are uncomfortable with
close bodily contact and unlikely to respond warmly if the infant exhibits neediness or distress.
Anxious infants are vigilant toward their attachment figures, exceedingly distressed upon sep-
aration, and are both clingy and angry during reunions. At home they receive inconsistent, in-
trusive, or poorly timed care from their primary attachment figures.
Attachment theorists generally focus on what they call the quality of attachment relation-
ships rather than the strength of the bonds. They argue that attachment is an all-or-nothing
process; infants are not "more" or "less" attached to someone. Rather, some infants are more
secure, and some are more avoidant or anxious in their attachments than others. Therefore, in-
dividual differences in attachment have generally been measured in terms of types (styles) or
dimensions (anxiety, avoidance), rather than strength of attachment. When the dimensions un-
derlying the styles are assessed (see Ainsworth et al., 1978, p. 102), the styles are viewed as re-
gions in a two-dimensional space, and it becomes clear that a person can move from one
"type" to another by changing positions, sometimes only slightly, on one of the dimensions.
Nevertheless, the three styles are relatively stable over time, even when measured cate-
gorically. Stability is maintained, up to a point, by self-perpetuating internal working models,
mental structures reflecting people's expectations about attachment figures that are "tolera-
bly accurate reflections of the experiences those individuals have actually had" (Bowlby,
1973, p. 235). Because working models are designed to reflect actual experience, if circum-
stances change dramatically, the discrepancy between an established working model and cur-
rent relationship conditions might be great enough to lead to gradual changes in the model.
As the model changes, the individual may exhibit or experience a change in his or her at-
tachment style.

Attachment Styles and Commitment to Romantic Relationships

Adult parallels of the major infant attachment patterns have been identified and shown to
affect the functioning and outcomes of romantic relationships (see reviews by Feeney, 1999;
Feeney & Noller, 1996; Shaver & Clark, 1994). Here, we will forego a detailed exposition of
this literature and focus instead on studies that reveal the relationship between attachment style
and the experience of commitment. Simpson (1990) examined the relationship between at-
tachment style, commitment, and other relationship variables in a longitudinal study of col-
lege-aged dating couples. He found that security was associated with greater commitment to
the partner and greater interdependence, trust, and satisfaction. Avoidance showed the oppo-
114 HILLARY J. MORGAN AND PHILLIP R. SHAVER

site pattern, with high avoidance predicting less commitment, interdependence, trust, and sat-
isfaction. Anxiety was not as strong a predictor of relationship variables, yet was associated
with less commitment to the current partner and less trust. Shaver and Brennan (1992) found
that attachment variables predicted commitment to romantic relationships better than the
NEO-PI (Costa & McCrae, 1985), a multifactor personality inventory (hence "PI") that as-
sesses several traits, including Neuroticism, Extraversion, and Openness to Experience. When
attachment dimensions were considered, avoidance appeared to be the best predictor of com-
mitment to romantic relationships, with greater avoidance predicting less commitment.
Keelan, Dion, and Dion (1994) examined changes in relationship characteristics over 4
months as a function of attachment style. Insecure subjects reported decreases in commitment,
satisfaction, and trust, as well as increases in costs over time, and they were more likely than
secure subjects to have ended their relationships during the 4-month study. Secure subjects
maintained high levels of commitment, satisfaction, and trust, and low levels of cost over time.
Pistole, Clark, and Tubbs (1995) examined associations between attachment styles and
Rusbult's (1983) investment model of commitment. Secure attachment was associated with
greater commitment and satisfaction and fewer costs. Highest costs were reported by anxious
subjects, while avoidant adults reported the lowest investments. In general, these studies indi-
cate that security is associated with greater feelings of commitment to current romantic rela-
tionships, while insecurity, particularly avoidance, is associated with less commitment.

Anxiety, Avoidance, and Thresholds for Commitment

Although anxiety and avoidance are both associated with less commitment to current re-
lationships, avoidant and anxious people differ substantially in their orientations to commit-
ment. Anxious people have a low threshold for creating romantic attachments and making
commitments. In a study conducted at a marriage license bureau, anxious men acquired mar-
riage licenses after a shorter courtship (19 months) than secure (49 months) or avoidant men
(46 months; Senchak & Leonard, 1992). Anxious people appear to form emotional bonds very
quickly, sometimes even before their relationships are fully established. Anxious adults are
more likely than secure or avoidant adults to experience love at first sight and to agree with the
statement, "It is relatively easy to fall in love; I find myself falling in love often" (Hazan &
Shaver, 1987). Furthermore, anxious people report higher levels of protest, anger, and despair
following unrequited love than secure or avoidant adults (Fraley, Davis, & Shaver, 1996).
Because anxious people tend to fall in love quickly, often before they know their partners well,
they are more vulnerable to leaping into relationships with people who are destined to disap-
point or hurt them. Thus, as Hazan and Shaver (1987) found, they tend to agree with the state-
ment, "Few people are as willing and able as I am to commit themselves to a long-term
relationship." They want committed relationships, but often select partners who do not share
this wish. They cannot have a committed relationship without the consent of their partner, so
if their partners are not willing to commit, their relationships will not be committed.
Unlike anxious people, avoidant adults have a high threshold for falling in love and making
commitments. Avoidant adults are more likely than their nonavoidant counterparts to report never
having been in love (23.7% of avoidant adults vs. 8.9% of anxious and 12.7% of secure adults)
and currently not being in love (49.4% vs. 25.5% and 29.2%, respectively; Feeney & Noller,
1990). They are more likely than secure or anxious people to believe that "it is rare to find some-
one you can really fall in love with" (Hazan & Shaver, 1987). Even when they are involved in
romantic relationships, they tend to maintain a distance. Kirkpatrick and Davis (1994) found that
avoidant men in romantic relationships reported less commitment, satisfaction, viability, caring,
and intimacy than secure men, and less passion and commitment than anxious men.
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 115

Even though avoidant adults report less commitment to their relationships, they are not
indifferent when their relationships end. Birnbaum, Orr, Mikulincer, and Florian (1994) com-
pared subjects undergoing divorce proceedings with those from intact marriages and found
that divorcing avoidant subjects exhibited more distress than divorcing secure subjects.
However, much like avoidant infants, avoidant adults do not show their distress to their attach-
ment figures. Fraley et al. (1996, Study 1) found that following the breakup ofan exclusive ro-
mantic relationship, avoidant adults engaged in little proximity-seeking or protest behavior
despite high levels of despair.

Attachment Style Pairings in Committed Romantic Relationships

Secure, anxious, and avoidant people have very different approaches to romantic rela-
tionships. Nevertheless, people with different attachment styles will meet, date, and sometimes
form lasting relationships. Kirkpatrick and Davis (1994) and Senchak and Leonard (1992)
found that most couples consist of two secure partners, yet there are a minority of couples in
which one or both partners are insecure. Although most people in committed relationships
have a secure partner, insecure adults are more likely than secure adults to have an insecure
partner. These insecure-insecure pairings almost always take the same form: one partner anx-
ious, the other avoidant. Kirkpatrick and Davis believe that anxious people pair with avoidant
individuals because they meet each others' expectations about relationships (e.g., anxious peo-
ple expect to be more invested in their relationships than their partners, avoidant people expect
to be less committed than their partners). Although these anxious-avoidant couples reported
little satisfaction, when the anxious partner was female and the avoidant partner was male,
they were as stable over a 4-year period as secure-secure pairings. Gender stereotypes may
playa role in preserving these "anxious woman, avoidant man" relationships. Anxious women
display many traditionally feminine qualities (e.g., concern about relationships, "mothering"
their partners, etc.), while avoidant men possess stereotypically masculine traits, such as lim-
ited emotional expressivity and a "tough" exterior. Even though these couples will not experi-
ence as many positive exchanges as secure-secure couples, they may understand and tolerate
their partners' behavior as "gender-appropriate." This could also account for why avoidant
women and anxious men have short-lived relationships: their behavior is counter to gender
stereotypes and may appear especially anomalous.
Considering the differences between anxious and avoidant individuals, it is easy to see that
an anxious-avoidant pair might exhibit some of the problems discussed in the research literature
on marital distress. Christensen (1988), for example, has discussed a destructive "demand-
withdrawal" cycle in which one partner repeatedly complains to the other about his or her cool-
ness and distance, and the distant partner then becomes even more distant in order to avoid the
painful tension caused by confrontation. In these couples, the demanding partner is more in-
vested in the relationship than the withdrawing partner, and is engaging in protest behavior in
an effort to bring the withdrawing partner closer. Just like infants, partners in romantic rela-
tionships will protest if they feel they are being neglected or abandoned. One partner's protest
behavior serves to inform the other partner that he or she needs to attend to the relationship. If
both partners are equally invested in the relationship, the other partner will attempt to reassure
the protester of his or her continuing presence and commitment. But if the other partner is not
invested in changing or repairing the relationship, such protest behavior will be unwelcome and
annoying. The unwilling targets of protest behavior may become exasperated with the protest-
ing person. They may feel guilty or frustrated because they cannot meet the other person's needs
without compromising themselves, and they may begin to feel contempt for the protester
(Baumeister & Wotman, 1992, p. 100). Gottman (1994) has empirically identified contempt as
116 HILLARY J. MORGAN AND PHILLIP R. SHAVER

one of the four most dangerous markers of troubled and deteriorating relationships. From an at-
tachment perspective, contempt is an understandable outcome when an avoidant person is con-
fronted with an anxious partner's protests, and it may help account for the low satisfaction of
anxious-avoidant pairs.

Attachment Styles and Relationship Stability

Attachment styles are associated with different orientations to commitment, but do the
styles predict relationship stability? Hazan and Shaver (1987) found that secure adults had re-
lationships oflonger duration (10.0 years) than anxious (4.9 years) or avoidant adults (6.0
years), and were less likely to report having been divorced (6% of secure vs. 10% of anxious
and 12% of avoidant subjects). Other cross-sectional studies also suggest that secure individ-
uals are less likely than insecure individuals to experience divorce (Birnbaum et al., 1994;
Feeney & Noller, 1990; Hill, Young, & Nord, 1994; but see Mikulincer & Erev, 1991, for an ex-
ception), while anxious adults are more likely than secure adults to experience breakups
(Hazan & Shaver, 1987; Kirkpatrick & Hazan, 1994). Yet when Kirkpatrick and Hazan (1994)
examined the stability of adults' relationships over 4 years, they found that anxious subjects'
breakups were often low points in otherwise long-standing, volatile relationships. Anxious
adults were more likely than secure adults to break up with their original partners during the 4-
year study, yet in one-third of the cases, they soon resumed the relationship. This trend was so
strong that at follow-up, anxious subjects were as likely as secure subjects to report still being
in a relationship with their original partner.

Commitment and Attachment Security: A Reciprocal Relationship?

Many studies indicate that subjects in ongoing committed relationships are more likely to
be secure than those without such a relationship (Birnbaum et al., 1994; Kirkpatrick & Davis,
1994; Kobak & Hazan, 1991; Senchak & Leonard, 1992). Although this observation might
suggest that secure people find it easier than insecure people to establish and maintain rela-
tionships, it may also indicate that being in a committed relationship promotes security. The
proportion of secure subjects in committed relationships is quite large and surprisingly stable
across studies (cited earlier) even though these studies were conducted in very different loca-
tions. On average, between 75% and 82% of subjects in committed relationships are secure,
whereas fewer than 55% of subjects not currently in a relationship call themselves secure.
Kirkpatrick and Hazan (1994) compared the stability of subjects' self-reported attach-
ment styles over a period of 4 years with changes in their relationship status. Among people
who reported being secure at both Time 1 and Time 2 (4 years apart), 81% were married at
Time 2, whereas only 27% of Time 1 secure subjects who became insecure were married at
Time 2. Of the people classified as insecure at Time 1 and secure at Time 2, 63% were mar-
ried at Time 2, whereas among those who were insecure at both times, 39% were married at
Time 2. Thus, being married at Time 2 appears to contribute to security regardless of Time 1
attachment style. The divorce figures tell a similar story, with Time 2 security associated with
intact marriages. Among Time 1 secure subjects who remained secure, only 10% had di-
vorced by Time 2, whereas 27% of subjects who changed from secure to insecure had di-
vorced. Of the insecure subjects who became secure, only 16% had divorced by Time 2,
whereas 29% of the subjects who remained insecure had divorced. In summary, these find-
ings suggest that the relationship between attachment style and relationship stability is reci-
procal rather than unidirectional.
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 117

Other indications that relationship experiences can affect attachment security come from
Baumeister and Wotman's (1992) study of unrequited love. Women who loved without recip-
rocation were especially likely to suffer, making many of them hesitant to love again, at least
for a while (p. 184). These women questioned whether they were lovable and worried that they
had degraded themselves in the process of trying to win love. One said that since her experi-
ence as an unrequited lover, "I've never allowed myself to like a guy that much, even the ones
I've dated" (p. 185). Some men expressed similar sentiments: for example, "I have noticed
more women interested in me but I do not want to get involved, because I am being cautious .
. . . I don't want to get burned again" (p. 184); "This taught me a lesson that getting 'hung up'
over someone is a preoccupation that generally increases stress and decreases well-being"
(p. 185). These comments suggest that some people who might have been secure before being
hurt were somewhat insecure afterwards, at least for a while.
Given that attachment styles seem to be partly determined by relationship conditions,
how stable are they? In studies of infants, secure attachment is generally more stable than in-
secure attachment (Egeland & Farber, 1984). The same pattern seems true of adults. Keelan et
al. (1994) looked at the attachment-style stability of subjects over a 4-month period, finding
that 85% of secure subjects at Time 1 were also secure at Time 2. For the two insecure groups,
the corresponding figures were 77% for the stability of avoidant subjects and 50% for anx-
ious subjects. Kirkpatrick and Hazan (1994) reported similar figures over a period of 4 years:
secure, 83 % stability; avoidant, 61 % stability; and anxious, 50% stability. It seems that the
anxious category is especially unstable, perhaps because it is most affected by relationship sta-
tus and quality.

Anxiety, Avoidance, and Relationship Perceptions

Anxious subjects often report that their partners are not as committed to them as they
would like (e.g., Collins & Read, 1990; Hazan & Shaver, 1987; Kirkpatrick & Davis, 1994;
Mikulincer & Erev, 1991; Simpson, 1990). It is important to consider whether the partners of
anxious people are truly distant and undevoted, or whether anxious attachment leads people
to believe that their relationships are more dismal than the facts warrant. The evidence on this
matter is interesting and fairly clear: Anxious people's perceptions of their partners and their
relationships are quite accurate. Mikulincer and Erev (1991) found that the partners of anxious
subjects reported feeling less intimacy and commitment than the partners of secure and
avoidant subjects. In the Kirkpatrick and Davis (1994) study, men in relationships with anx-
ious women reported less commitment, satisfaction, viability, and intimacy, and more con-
flict-ambivalence than men paired with secure women. Collins and Read (1990) reported that
both men and women were less interested in marrying partners scoring high on the anxiety di-
mension, and men were especially disenchanted with anxious partners. Men with anxious part-
ners reported less satisfaction, more conflict, less closeness, poorer communication, less
self-disclosure, less willingness to help their partner disclose, less faith in their partner, less
liking for their partner, and more feelings that their partner was not dependable. Simpson
(1990) found that female partners of anxious males reported less commitment, love, depen-
dency, and investment. And the effect was even stronger for the male partners of anxious
women. The higher a woman's anxiety, the less commitment, love, dependency, self-disclosure,
investment, and satisfaction her partner reported, and the more her partner saw her as unpre-
dictable. Therefore, anxious individuals, especially anxious women, seem justified in their
concern that their partners do not love them enough and are not committed to them.
Some researchers have suspected that anxious individuals have idealistic and unrealistic
expectations for romantic relationships, leading them to feel frustrated and disappointed by the
118 HILLARY J. MORGAN AND PHILLIP R. SHAVER

normal ups and downs of coupled life (Hazan & Shaver, 1987). Yet this does not appear to be
the case. Mikulincer and Erev (1991, Study 3) asked subjects to complete Sternberg's (1986)
three-dimensional measure of passion, intimacy, and commitment four times, once for their
own feelings toward their partner, again for their partner's feelings toward them, once to de-
scribe how they would feel in an ideal relationship, and finally to describe how they would
want their partner, ideally, to feel about them. When asked to describe how they would feel in
ideal love, avoidant people described feeling less commitment, intimacy, and passion than ei-
ther secure or anxious people. There were no differences between secure and anxious subjects
in their reports of the feelings they would like to have for their partner. When subjects were
asked about the feelings they would like their partner to have for them, secure subjects wanted
greater intimacy than insecure subjects, and, interestingly, anxious subjects reported wanting
less commitment than either secure or avoidant subjects. In summary, anxious individuals did
not have extreme ideals or needs. The level of intimacy they sought was no higher than the
level that secure subjects felt they were actually receiving, and the level of commitment sought
by anxious individuals was even lower than what secure subjects actually received.
What about the perceptions of avoidant subjects? Unlike anxious subjects who appeared
"sad but wise," avoidant subjects tended to misperceive their partners, believing that their part-
ners felt less intimacy and commitment than the partners actually reported (Mikulincer & Erev,
1991). Therefore, these low scores reflected the avoidant subjects' own feelings of intimacy
and commitment rather than the partners' actual feelings. In general, then, it appears that if any
group is misperceiving their relationships, it is the avoidant rather than the anxious group.

THE MAINTENANCE OF ABUSIVE RELATIONSHIPS

Rusbult (1980, 1983) formulated her investment model of commitment with the under-
standing that commitment cannot be defined by satisfaction alone. Although satisfaction sig-
nificantly predicts commitment, the prediction is even more accurate when two other variables
are added to the equation: investment in the relationship and the availability of alternatives.
Thus Rusbult's model can account for "nonvoluntary dependence": cases in which people
would prefer to leave a relationship but are not willing to do so because they would be losing
investments, or because their flawed relationship is better than the available alternatives.
Rusbult and Martz (1995) assert that women in abusive relationships are experiencing non-
voluntary dependence. Yet social and economic barriers cannot fully account for a woman's de-
cision to stay with an abusive partner. There are certainly cases in which women feel trapped
in loveless relationships, but Rusbult's model does little to illuminate the feelings oflove and
longing reported by many abused women. Follingstad, Rutledge, Polek, and McNeill-Hawkins
(1988) found that women in ongoing violent relationships reported more commitment and love
for their partners than women who had experienced only one violent incident. Similarly,
Dutton and Painter (1993) found that the more abusive the husband, the more likely the wife
was to report strong feelings of attachment. Both of these findings are difficult to explain with
any theory focusing largely on the weighing of rewards and costs.

Love and Commitment in Abusive Relationships

Not only do many battered women love their partners, but love is often a reason why they
choose to stay with them. Strube and Barbour (1983) interviewed battered women seeking
counseling services while living with their abusive partners. At the onset of therapy, counselors
asked the women an open-ended question about why they were involved with their partners.
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 119

They found that women who mentioned love or economic dependence were least likely to end
their relationships during the course of therapy; both factors significantly and independently
predicted whether women would leave their relationships. Only 18% of the women who men-
tioned economic dependence had left their partners upon termination of therapy, compared to
71 % of the women who did not mention economic dependence. Similarly, only 35% of women
who mentioned love chose to leave their partners by the end of therapy, whereas 71 % of the
women who did not mention love left their partners.
In a sample of battered women at a shelter, Rusbu1t and Martz (1995) found that contin-
ued "positive feelings for the partner" predicted feelings of commitment but not whether the
women actually returned to their partners within 1 year. Many of the women who returned to
their partners were committed to leaving them when they entered the shelter, but either a
change of heart or pressing circumstances led them to return. Snyder and Fruchtman (1981)
also found that many battered women returned to their abusers despite their stated commit-
ment to leave. In their sample of battered women at a shelter, only 13% of the women, at in-
take, reported planning to return to their abusers, yet when they were contacted 6---10 weeks
after leaving the shelter, a full 60% of them were living with their abusive partners. Dutton and
Golant (1995) write, "The bonds that bind abuse victims to their tormentors are legendary.
They are like giant bungee cords. As the woman dives out of the relationship, the cord stretches
to the breaking point. But the further she goes, the greater the tension to snap back" (p. 57).

Traumatic Bonding

Why is it so difficult to permanently leave an abusive partner? Dutton and Painter (1993;
Painter and Dutton, 1985) outlined a theory of traumatic bonding that portrayed bond forma-
tion in abusive relationships as a function of reinforcement principles and power imbalance.
They argued that people in abusive relationships experience a chaotic and sometimes reward-
ing bond. People are therefore hooked into the relationship through intermittent reinforcement,
the reinforcement schedule most likely to prevent extinction. Much like gamblers who con-
tinue to lose as they throw coins into a slot machine, people in abusive relationships cannot
break away because their next move might be rewarded. Although reinforcement principles
may help to explain why people persist in frequently unrewarding relationships, they do not ex-
plain why love or passion would emerge, nor do they provide a framework for understanding
why certain people might be more vulnerable than others to remaining in abusive relationships.
Dutton and Golant (1995) argue that "there is no special deficit in a battered woman's person-
ality that makes her susceptible to getting trapped in an abusive relationship. To the contrary,
the features of the relationship itself are sufficient to account for the trapping" (p. 57).

Attachment Theory and Bond Formation

In contrast, attachment theory (Bowlby, 196911982) provides both a mechanism for un-
derstanding the emotional draw of abusive relationships and a basis for understanding indi-
vidual differences in persistence. Ironically, the tendency to desire contact with a raging,
frightening attachment figure is a natural outcome of the attachment system that under normal
circumstances serves a protective function. The attachment system was "designed" by evolu-
tion to become activated under conditions of external threat, often signaled by caretakers who
become distressed and thus indicate impending danger. Among primates,
when a dominant male senses a predator or other danger he commonly threatens
or even attacks a juvenile that unwarily approaches the danger spot. The dominant
120 HILLARY J. MORGAN AND PHILLIP R. SHAVER

male's behavior, by frightening the juvenile, elicits the juvenile's attachment be-
havior. As a result the juvenile seeks the proximity of an adult animal, as often as
not that of the very male that frightened it; and by so doing the juvenile also re-
moves itself from danger. (p. 227, our emphasis)
The attachment system was not "designed" for situations in which the attachment figures
themselves are the source of danger. According to Bowlby, "No system whatever can be so
flexible that it suits all and every environment" (1969/1982, p. 50). "Thus, only when the en-
vironment conforms exactly to the environment of evolutionary adaptedness does each of the
behavioural systems ... produce its effect and the whole organization lead to behavior of sur-
vival value" (p. 76). In the case of abuse, we are observing the dysfunction of a system oper-
ating under unusual circumstances. When attachment figures are repeatedly abusive, their
children or romantic partners will experience heightened feelings of attachment and will be
confronted with the paradoxical dilemma of longing for the comfort of the attachment figure
while also experiencing a desire to flee from him or her.
Victims of abusive relationships grapple with another paradox: They know rationally that
their relationships are destructive, yet they continue to love and long for their abusive partners.
This paradox is perfectly understandable when one considers that attachment bonds are non-
rational, enduring, and largely uncontrollable. "Love is blind" once an attachment is in place;
rational arguments do little to dissuade people from loving whomever they love. The Scottish
psychoanalyst Fairbairn (1952) theorized that people experiencing the discrepancy between
their loving feelings and their knowledge that the people they love are not deserving may ex-
perience "splitting," a dissociative defense whereby tension and confusion are bypassed
through the formation of two separate models of the attachment figure, one positive and one
negative. Once the models have formed, people will be able to draw on an "all good" model
of their attachment figure to justify their current involvement.
People do not form attachments by objectively surveying the field of potential attachment
figures and selecting someone accordingly. Both children and animals regularly form emo-
tional ties to abusive caregivers and will protest separation from them. Celani (1994) recounts
a case in which Fairbairn observed one of his colleagues asking a girl who had recently been
removed from her home if she would like the colleague to find her a nice new mommy. Despite
the abuse her mother had inflicted upon her, "the young girl recoiled in horror and insisted that
she wanted her own ... mother back" (p. 24). Celani adds, "The return to a batterer by the vic-
tim is psychologically identical to Fairbairn's observation that his abandoned children pre-
ferred to be beaten in their homes by their parents rather than live safely in a foundling home"
(pp.151-152).

INDIVIDUAL DIFFERENCES IN VULNERABILITY

TO REMAINING IN ABUSIVE RELATIONSHIPS

Object Relations and the IIBad Object"

Celani (1994) draws on the work of Fairbairn and other object relations theorists when he
asserts that battered women stay with their abusers because they are drawn to "bad objects"-
partners "who hold out the promise of gratification, yet fail time after time to satisfy the needs
of the dependent individual" (p. 137). He believes that only certain people will feel drawn into
a relationship with a bad object: those who were deprived or neglected in childhood.
Specifically, adults who did not receive adequate positive experiences with their caregivers in
childhood may seek to re-create the dynamics of those early experiences by entering relation-
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 121

ships with people like their parents, who promise much but rarely deliver. His theory offers a
wealth of potential research questions in a largely uncharted area, yet he focuses exclusively
on individual differences when accounting for the maintenance of abusive relationships.
Although it is extremely likely that some people are more vulnerable to entering abusive rela-
tionships than others, it is also likely that the dynamics of abusive relationships can lead almost
anyone to feel drawn to his or her abusive partner.

Anxious Attachment and the Maintenance of Commitment

in Abusive Relationships

Henderson, Bartholomew, and Dutton (1995) found that anxious attachment (which they
labeled "preoccupied" attachment) was greatly overrepresented, and secure attachment under-
represented, in their sample of battered women seeking shelter or counseling. The majority of
women in their sample were anxiously attached (53% compared with 20% in normal samples)
while only 7% were secure (compared with the usual 45-55%). Attachment style alone did not
predict continuing involvement with the abusive partner, yet when level of abuse was con-
trolled, anxious women were more likely than nonanxious women to be emotionally and sex-
ually involved with their partners at a 6-month follow-up. Also at follow-up, battered women
with an anxious attachment style were more inclined to agree with the statement, "I still love
my partner and I want to get back together."
Morgan and Pietromonaco (1997) found that people with a more anxious (or "preoccu-
pied") attachment style reported greater commitment and experienced more rewards in highly
conflictual romantic relationships than people with a less anxious attachment style. Although,
in general, conflict tended to reduce commitment and rewards, people high in anxious attach-
ment reported as much commitment and rewards in high-conflict relationships as in low.
Rusbult's model significantly predicted commitment for the sample as a whole, yet her model
was less effective in predicting the commitment of anxious people in highly conflictual rela-
tionships. Her model accounted for 75% of the variance in commitment for nonanxious peo-
ple in low-conflict relationships, yet it accounted for only 49% of the variance in commitment
for anxious people in high-conflict relationships. Thus, people with an anxious attachment
style may be especially vulnerable to following their hearts (i.e., attachment-related needs)
rather than their heads when they are involved with an abusive partner.

CONCLUSION

The leading theories of commitment (e.g., Rusbult, 1980, 1983) derive from economic
models of close relationships. They portray commitment as resulting from the weighing and
averaging of costs, gains, investments, and alternatives, leading to a decision concerning the
value of staying with a particular relationship partner. We suggest, in addition, that romantic
partners are held together in part by the attachment system-a fairly primitive, nonrational,
and powerful binding force. Although commitment to a career, an organization, or a project has
many features in common with commitment to a loved one, the latter seems different in its
depth and dependence on innate mechanisms. This is indicated in part by the security accom-
panying committed romantic relationships. Few objects of commitment provide the funda-
mental sense oflove-worthiness and connectedness experienced by people who are securely
attached to a loving partner.
Once an attachment is in place, the bond is largely impervious to the behavior of the at-
tachment figure, a quality rarely shared with career commitments, commitments to causes, and
122 HILLARY J. MORGAN AND PHILLIP R. SHAVER

so on. Although propagandists want citizens to recite "my country, right or wrong," or ''my po-
litical party, right or wrong," most citizens do not really accept this. The feeling that one's at-
tachment figures (like one's children) deserve allegiance, right or wrong, comes more
naturally, even if it sometimes leads to tragedy. Of course, not all marital or romantic commit-
ments contain an attachment element. Sternberg (1986) described eight fundamental forms of
love based on different combinations of passion, intimacy, and commitment. One of these
forms, empty love, was characterized by high commitment but little passion or intimacy. We
would expect such relationships to show a different pattern from relationships in which part-
ners are clearly attached to each other.
Attachment style is intimately linked to people's approach to commitment, with anxious
people tending to making commitments quickly and easily, and avoidant people being hesitant
to invest fully in relationships. Although commitment and relationship longevity are pre-
dictable from attachment style, the relationship between commitment and attachment appears
to be reciprocal. Longitudinal studies indicate that changes in relationship status can alter at-
tachment style, with people in committed relationships tending to move toward security, and
people experiencing a divorce or breakup becoming less insecure. Anxious people fall in love
quickly, making them susceptible to choosing partners who are inappropriate or even danger-
ous. Battered women seeking shelter are especially likely to be anxiously attached, and there is
some indication that anxiously attached people report more commitment to highly conflictual
relationships than people who are less anxiously attached.
In short, attachment theory helps to explain how commitment in romantic relationships dif-
fers from other, more rational commitments. It shows how a generally functional biological sys-
tem can, under certain circumstances, contribute to dysfunctional, even destructive, relationships.

ACKNOWLEDGMENT: Preparation for this article was supported in part by a grant from the
National Institute of Mental Health, T32MH1893I, to the Postdoctoral Training Program in
Emotion Research (Paul Ekman, Director).

REFERENCES

Ainsworth, M. D. S. (1991). Attachments and other affectional bonds across the life cycle. In C. M. Parkes, J.
Stevenson-Hinde, & P. Marris (Eds.), Attachment across the life cycle (pp. 33-51). New York: Routledge.
Ainsworth, M. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the
strange situation. Hillsdale, NJ: Erlbaum.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model.
Journal ofPersonality and Social Psychology, 61, 226-244.
Baumeister, R. E, & Wotman, S. R. (1992). Breaking hearts: The two sides of unrequited love. New York: Guilford.
Birnbaum, G. E., Orr, I., Mikulincer, M., & Florian, V. (1994). When marriage breaks up: Does attachment style con-
tribute to coping and mental health? Unpublished manuscript, Department of Psychology, Bar-Han University,
Ramat Gan, Israel.
Bowlby, J. (1982). Attachment and loss: Vall. Attachment. New York: Basic Books. (Original published 1969)
Bowlby, 1. (1973). Attachment and loss: Vol 2. Separation: Anxiety and anger. New York: Basic Books.
Bowlby, J. (1979). The making and breaking ofaffoctional bonds. London: Tavistock.
Bowlby, 1. (1980). Attachment and loss: Vol 3. Loss. New York: Basic Books.
Bowlby, 1. (1988). A secure base: Parent-child attachment and healthy human development. New York: Basic Books.
Bretherton, I., & Waters, E. (Eds.). (l985). Growing points of attachment theory and research. Monographs ofthe
Society for Research in Child Development, 50( 1-2, Serial No. 209).
Cassidy, J., & Shaver, P. R. (Eds.). (1999). Handbook ofattachment: Theory. research, and clinical application. New
York: Guilford.
Celani, D. P. (1994). The illusion of love: Why the battered woman returns to her abuser. New York: Columbia
University Press.
Christensen, A. (1988). Dysfunctional interaction patterns in couples. In P. Noller & M. A. Fitzpatrick (Eds.),
Perspectives on marital interaction (pp. 31-52). Clevedon, UK: Multilingual Matters.
ATTACHMENT PROCESSES AND COMMITMENT TO ROMANTIC RELATIONSHIPS 123

Collins, N. L., & Read, S. J. (1990). Adult attachment, working models, and relationship quality in dating couples.
Journal ofPersonality and Social Psychology, 58, 644-{)63.
Costa, P. T., Jr., & McCrae, R R. (1985). The NED Personality Inventory. Odessa, FL: Psychological Assessment
Resources.
Dutton, D., & Golant, S. K. (1995). The batterer: A psychological profile. New York: Basic Books.
Dutton, D., & Painter, S. L. (1981). Traumatic bonding: The development of emotional attachments in battered women
and other relationships of intermittent abuse. Victimology: An International Journal, 6, 139-155.
Dutton, D. G., & Painter, S. L. (1993). Emotional attachments in abusive relationships: A test of traumatic bonding
theory. Violence and Victims, 8, 105-120.
Dutton, D. G., Saunders, K., Starzomski, A., & Bartholomew, K. (1994). Intimacy-anger and insecure attachment as
precursors of abuse in intimate relationships. Journal ofApplied Social Psychology, 24, 1367-1386.
Egeland, B., & Farber, E. (1984). Infant-mother attachment: Factors related to its development and changes over time.
Child Development, 55, 753-771.
Fairbarn, W. R D. (1952). Psychoanalytic studies of the personality. London: Tavistock.
Feeney, J. A. (1999). Adult romantic attachment and couple relationships. In 1. Cassidy & P. R Shaver (Eds.),
Handbook ofattachment: Theory, research, and clinical applications (pp. 355-377). New York: Guilford
Feeney, 1. A., & Noller, P. (1990). Attachment style as a predictor of adult romantic relationships. Journal of
Personality and Social Psychology, 58, 281-291.
Feeney, 1., & Noller, P. (1996). Adult attachment. Thousand Oaks, CA: Sage.
Follingstad, D. R, Rutledge, L. L., Polek, D. S., & McNeill-Hawkins, K. (1988). Factors associated with patterns of
dating violence toward college women. Journal ofFamily Violence, 3, 169-182.
Fraley, R C., Davis, K. E., & Shaver, P. R. (1996). Attachment behavior and romantic relationship dissolution.
Unpublished manuscript, Department of Psychology, University of California, Davis.
Gottman, J. M. (1994). What predicts divorce? Hillsdale, NJ: Erlbaum.
Greenberg, M. T., Cicchetti, D., & Cummings, M. E. (Eds.). (1990). Attachment in the preschool years: Theory. re-
search, and intervention. Chicago: University of Chicago Press.
Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal ofPersonality and
Social Psychology, 52, 51l-524.
Hazan, C., & Zeifman, D. (1994). Sex and the psychological tether. In K. Bartholomew & D. Perlman (Eds.),
Advances in personal relationships (Vol. 5, pp. 151-180). London: Jessica Kingsley.
Hazan, C., & Zeifinan, D. (1999). Pair bonds as attachments: Evaluating the evidence. In J. Cassidy & P. R. Shaver
(Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 336--354). New York:
Guilford.
Henderson, A. J. Z., Bartholomew, K., & Dutton, D. G. (1995). He loves me; he loves me not: Attachment and sepa-
ration resolution of abused women. Unpublished manuscript, Department of Psychology, Simon Fraser
University, Burnaby, BC.
Hill, E. M., Young, 1. P., & Nord, 1. L. (1994). Childhood adversity, attachment security, and adult relationships: A pre-
liminary study. Ethology and SOCiobiology, 15, 323-338.
Keelan,1. P. R, Dion, K. L., & Dion, K. K. (1994). Attachment style and heterosexual relationships among young
adults: A short-term panel study. Journal of Social and Personal Relationships, 11,201-214.
Kerckhoff, A. C. (1974). The social context of interpersonal attraction. In T. L. Huston (Ed.), Foundations ofinter-
personal attraction (pp. 61-78). New York: Academic Press.
Kirkpatrick, L. A., & Davis, K. E. (1994). Attachment style, gender, and relationship stability: A longitudinal analy-
sis. Journal ofPersonality and Social Psychology, 66, 502-512.
Kirkpatrick, L. A., & Hazan, C. (1994). Attachment styles and close relationships: A four-year prospective study.
Personality Relationships, I, 123-142.
Kobak, R R, & Hazan, C. (1991). Attachment in marriage: Effects of security and accuracy of working models.
Journal ofPersonality and Social Psychology, 60, 861-869.
Lykken, D. T. (1995). The antisocial personalities. Hillsdale, NJ: Erlbaum.
Mikulincer, M., & Erev,l. (1991). Attachment style and the structure of romantic love. British Journal of Social
Psychology, 30, 273-291.
Morgan, H. 1. (1993). Emotional extremes and attachment in conflictual romantic relationships. Unpublished doctoral
dissertation, University of Massachusetts, Amherst.
Morgan, H. J., & Pietromonaco, P. (1997). Preoccupied attachment as a risk factor for remaining in conflictual ro-
mantic relationships. University of California, Davis.
Nahemow, L., & Lawton, M. P. (1975). Similarity and propinquity in friendship formation. Journal ofPersonality and
Social Psychology. 32, 205-213.
Painter, S. L., & Dutton, D. (1985). Patterns of emotional bonding in battered women: Traumatic bonding.
International Journal of Women :s, Studies, 8. 363-375.
Parkes, C. M., Stevenson-Hinde, J., & Marris, P. (Eds.). (1991). Attachment across the life cycle. New York: Routledge.
124 HILLARY J. MORGAN AND PHILLIP R. SHAVER

Pistole, M. C., Clark, E. M., & Tubbs, A. L. (1995). Love relationships: Attachment style and the investment model.
Journal ofMental Health Counseling, 17, 199-209.
Rusbult, C. E. (1980). Commitment and satisfaction in romantic associations: A test of the investment model. Journal
ofExperimental Social Psychology, 16, 172-186.
Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and deterioration) of sat is-
faction and commitment in heterosexual involvements. Journal of Personality and Social Psychology, 45,
101-117.
Rusbult, C. E., & Martz, 1. M. (1995). Remaining in an abusive relationship: An investment model analysis of non-
voluntary dependence. Personality and Social Psychology Bul/etin, 21, 558-571.
Senchak, M., & Leonard, K. E. (1992). Attachment styles and marital adjustment among newlywed couples. Journal
ofSocial and Personal Relationships, 9, 51-64.
Shaver, P. R, & Brennan, K. A. (1992). Attachment styles and the "big five" personality traits: Their connection with
each other and with romantic relationship outcomes. Personality and Social Psychology Bul/etin, 18, 536-545.
Shaver, P. R, & Clark, C. L. (1994). The psychodynamics of adult romantic attachment. In 1. M. Masling & R. F.
Bomstein (Eds.), Empirical perspectives on object relations theory (pp. 105-156). Washington, DC: American
Psychological Association Press.
Shaver, P., Hazan, c., & Bradshaw, D. (1988). Love as attachment: The integration of three behavioral systems. In R
Sternberg & M. Barnes (Eds.), The psychology of love (pp. 69-99). New Haven, CT: Yale University Press.
Shaver, P. R, Morgan, H. 1., & Wu, S. (1996). Is love a "basic" emotion? Personal Relationships, 3, 81-96.
Simpson, 1. A. (1990). Influence of attachment styles on romantic relationships. Journal ofPersonality and Social
Psychology, 59, 971-980.
Snyder, D. K., & Fruchtman, L. A. (1981). Differential patterns of wife abuse: A data-based typology. Journal of
Consulting and Clinical Psychology, 49, 878-885.
Sroufe, L. A., & Waters, E. (1977). Heart rate as a convergent measure in clinical and developmental research.
Merrill-Palmer Quarterly, 23, 3-27.
Sternberg, R. 1. (1986). A triangular theory of love. Psychological Review, 93, 119-135.
Strube, M. 1, & Barbour, L. S. (1983). The decision to leave an abusive relationship: Economic dependence and psy-
chological commitment. Journal ojMarriage and the Family, 45, 785-793.
Weiss, R S. (1973). Loneliness: The experience ojemotional and social isolation. Cambridge, MA: MIT Press.
Weiss, R S. (1991). The attachment bond in childhood and adulthood. In C. M. Parkes, 1. Stevenson-Hinde, & P.
Marris (Eds.), Attachment across the life cycle (pp. 66-76). New York: Routledge.
Zajonc, R B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175.