Attachment as an Organizational Framework for Research on Close Relationships Author(s): Cindy Hazan and Phillip R.

Shaver Reviewed work(s): Source: Psychological Inquiry, Vol. 5, No. 1 (1994), pp. 1-22 Published by: Taylor & Francis, Ltd. Stable URL: http://www.jstor.org/stable/1449075 . Accessed: 16/03/2012 10:28
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to Psychological Inquiry.

http://www.jstor.org

Psychological Inquiry 1994, Vol. 5, No. 1, 1-22

Copyright1994 by LawrenceErlbaum Associates, Inc.

TARGET ARTICLE

Attachment as an Organizational Framework for Research on Close Relationships

Phillip R. Shaver
Universityof California,Davis Recent years have witnessed a proliferation of research on close relationships and the emergenceof a new relationshipsubdisciplinewithinthe social sciences. To date, the new science of relationshipshas been dominatedby data. Thisarticle is based on the convictionthatprogress now hinges on the development theoryto organize and of interpretextantfindings and to guide future investigations.Througha selective but extensive review of the major bodies of empiricalliterature,we attemptto show that attachmenttheory can incorporatea broad range of findings on adult relationships. In addition, attachmenttheory addresses an impressivearray of research questions concerningthefunctions,emotionaldynamics,evolutionaryorigins, anddevelopmental pathways of humanaffectional bonds. We conclude that a comprehensivetheory of close relationshipsis both desirable and, with the integrationof existing theories and concepts, currentlyachievable. In 1958, HarryHarlowwrote:
Ourassigned mission as psychologists is to analyzeall facets of human and animal behavior into their component variables. So far as love or affection is concerned,psychologists have failed in theirmission. The little we know about love does not transcendsimple observation and the little we write about it has been written better by poets and novelists. (Harlow, 1958, p. 673)

Today, more than 30 years later, there is an international network of researcherswhose investigative efforts are devoted entirely to the study of personal relationships. Have we made any progress in understanding love and affection? Most definitely, we have. Many have arguedthatwhat is needed in the field of personal relationships is what every new science requires, a descriptivebase of knowledge from which to derive principles and construct theory (e.g., Hinde, 1979; Rubin, 1984). This articleis basedon the conviction that, after more than three decades of researchon close relationships, there is ample data to justify a search for meaningful patternsand organizingprinciples. In what follows, we draw on attachmenttheory

(Bowlby, 1973, 1980, 1969/1982, 1988) to propose a for generativetheoreticalframework organizingextant data and for making predictionsbeyond what can be currentlydocumented.Whatwe seek is a comprehensive theoryof close relationships,and what we propose here is the outline of such a theory. In searchingfor a theoryof relationships,we are led to ask what, ideally, one would want from such a theory. What questions should it answer?We believe that, at a minimum, it should address the following: Whatmakesa potentialrelationship partner appealing? How is a relationship formed,andhow does it develop? How are relationships maintained, and what makes themsatisfyingor enduring? Why andhow arerelationdissolved? Whatare the reactionsto relationship ships breakup? Beyond these basics, a comprehensive theory of relationships should offer models of both normative and individual-difference phenomena and account for the role of relationships in a person's overall adaptationand functioning throughoutlife. It should be able to explain the universal human tendency to form close relationships and the similarities and differences in affectional bonds from infancy through adulthood. Perhapsmost important,a theory of close

HAZAN & SHAVER

relationshipsmust be consistent and compatiblewith existing empirical findings. And, as with any theory that earns its keep, it must be parsimonious,testable, and generative. The challenge awaiting a comprehensive theory of close relationships is formidable. Considerthe diversity of findings and phenomena to be incorporated: Frequencyof mutualgazing is a good indicatorof the amount of love that partners feel for each other (Rubin, 1973). We tend to fall in love with people who seem especially responsive to our needs (A. P. Aron, Dutton, E. N. Aron, & Iverson, 1989; Berscheid, 1984). Inadequatecare during infancy is predictive of later troubled relations with peers (Arend, Gove, & Sroufe, 1979). Children who are anxiety prone are more inclined to develop romantic crushes (Hatfield & Rapson, 1987). Within the context of laboratory marital interaction, physiological arousal predicts the eventual demise of the marriage (Levenson & Gottman, 1985). Fear of intimacy is associated with workaholism (Hazan & Shaver, 1990). Adolescents who idealize their parents are judged as hostile by their peers (Kobak & Sceery, 1988). The typical initial reaction to the ending of a close relationship is anxiety (Weiss, 1988). Obstacles tend to enhance romantic passion (Driscoll, K. W. Davis, & Lipetz, 1972). Parentaldivorce during childhood is associated with chronic loneliness in adulthood (Shaver & Rubenstein, 1980). People tend to overlook the faults and limitations of a new partner (Tennov, 1979). Relationship satisfaction typically declines in the early years (Spanier, Lewis, & Cole, 1975). And so on. Is it possible to detect laws of relationshipstructure and process that lend coherence to these and other relationshipfacts? We believe thata parsimoniousexplanationof much of the existing datais both desirable and possible, andwe thinkattachment theoryis a good to start. We are not ready to claim that, in its place currentform, attachmenttheory tells us all we would ever want or need to know in orderto understand close relationships.Some shortcomingsof attachment theory can be handled nicely by interdependence theory (Kelley et al., 1983; Thibaut & Kelley, 1959) and theoriesthatstress the mutualconstructionof relationship narratives(e.g., Duck, 1991). At the end of this article, we say more about how we see the alternative theoriesfitting together. Attachment Theory In 1950, John Bowlby was invited by the World Health Organization(WHO) to reporton the mental healthof London's many homeless children.The invi2

tation followed his publication(Bowlby, 1944) of an article entitled "Forty-FourJuvenile Thieves: Their Charactersand Home Life," which reporteda strong association between early maternal separations and subsequentdelinquencyamongboys. The WHOreport (Bowlby, 1951) assertedthatmaternaldeprivation,especially duringthe first 3 years of life, puts childrenat increasedriskforphysicalandmentalillness. Although influentialand well received, the reportwas deficient in one important respect:It failed to explainwhy or how has earlymaternal deprivation such deleteriouseffects. Bowlby had been trained in the psychoanalytic traditionbut, almost as soon as he began to practice child therapy, he found himself troubled by what he perceived to be inadequacies in psychoanalytic theory. Most objectionable was its exclusive focus on fantasy and the internallife; a child's real-life experiences were of little interest relative to intrapsychic events. Bowlby was also troubledby inconsistencies between psychoanalytic theory and his own observations. That institutionalized children suffered extreme distress and even sometimes failed to thrive despite being fed and cared for by staff did not follow from psychoanalytic notions that children love their mother simply because they associate her with the satisfaction of a hunger drive. Bowlby's growing dissatisfaction with psychoanalytic theory helped launch a search for answers to questions raised in his WHO report. The searchtookhim throughtheliteratures several of but it was in ethology that he found what disciplines, he believed to be an importantpart of the answer. Researchon the bonding behaviorof birds and mammals was particularly influentialin his thinking,gradually leadingto the insightthatmaternaldeprivationis developmentallyharmfulbecause it thwartsthe satisfaction of an inbornneed. Further, work of etholothe gists like Lorenz and Tinbergen provided dramatic examples of how even inborn, instinctualtendencies can become distortedor fail to develop in non-optimal environments.Almost 20 years later, Bowlby published the first of three majorvolumes on attachment theory.For obvious reasons,we cannot review all the supportingevidence on which the theory is based; we will settle insteadfor an explicationof the theoryitself. (For a more extensive review of the evidence, see Bowlby, 1973, 1980, 1969/1982.)

The Attachment System A basic assumption of attachmenttheory is that, because of their extreme immaturityat birth, human infantscan surviveonly if an adultis willing to provide protectionand care. As a resultof selection pressures,

ATIACHMENT AND CLOSERELATIONSHIPS

infantsevolve behaviorsthatfunctionto maintain proxAdult caregivingis regimity to a protector/caregiver. ulatedby a complementarybehavioralsystem. Babies smile, and parentsfind the smiles rewarding.Babies cry, and parentsare motivatedto soothe them. Parents move away, and babies follow visually or physically. These two systems conspire to create the kind of relationshipthatfosters the infant's survival. A behavioral system consists of a set of behaviors that serve the same function, although they may be morphologically dissimilar (e.g., crying, smiling, following). The attachmentsystem is similar in some respects to the physiological systems that regulate body temperature,blood pressure, and the like. Any real or perceived obstacle to proximity maintenance results in anxiety, which in turn triggers attachment behaviors designed to reestablish proximity. Such behaviors persist until the "set goal" for proximity has been achieved. The degree of proximity required to keep anxiety at bay is related to a variety of endogenous and exogenous factors, including the child's age, emotional and physical state, and perceived environmental threat. The establishment and maintenanceof proximity engender feelings of security and love, whereas disruptions in the relationship typically beget anxiety and sometimes anger or sadness (depending on particular appraisals). Hence, Bowlby argued that an attachment is an emotional bond. Attachmentis one of severaldistinctbutinterlocking behavioralsystems, includingexploration,caregiving,

affiliation, and sexual mating. Each system serves uniquefunctionsandrespondsto differentenvironmental cues. At least duringinfancyand childhood,attachment is the preeminentsystem, and its full activation precludesactivationof other systems. As long as the child experiences "felt security" (Sroufe & Waters, 1977), the attachmentsystem is quiescent, and other behavioral systems become available for activation, althoughperiodicchecking of the attachmentfigure's availabilitycontinues to occur. In evolutionaryterms, it has been adaptive for human young to feel safe enough to engage in play and explorationonly as long as a familiarprotectoris availableto respondif needed. The safest reactionto any threator strong uncertainty is to devote all energy and attentionto reestablishing proximity.(See Figure1 for a model of the attachment system.)

Attachment Formation Childrencould conceivably direct their attachment behaviorsto any availableperson.In reality,however, by the sixth or seventhmonthof life, all normalinfants selectively direct these behaviors to one person, with whom they also seek proximityand from whom they object to being separated (Schaffer& Emerson,1964). (See also Mizukami,Kobayashi,Ishii, & Iwata, 1990, for evidence of selective attachmentby 2 to 4 months of age.) How is this person "selected?" Of definite significance to infantsis who usually respondsto their

behavioralsystem.In the diamondis the testquestion(G.A. Miller, Galanter,& Pribram, Figure 1. Theattachment 1960). The circles represent emotions triggered by the appraisals that answer the test question, and the boxes representbehaviors thatfollow from the appraisals and emotions.
3

HAZAN & SHAVER

signals of distress.Also of significance is the qualityof the response.Thus, familiarityand responsivenessdictate preferences and influence the selection of an attachmentfigure. The functions of attachment are apparent in the infant'sbehavior.Proximityto the attachment figure is is fearful especially likely to be soughtwhen the infant or distressedfor any reason.The caregiverserves as a haven of safety to which the infant can retreat for comfortandreassurance duringsuch times. In addition, the caregiverserves as a base from which to engage in nonattachment behaviors,such as exploration.According to Bowlby,proximilymaintenance(includingproximity seeking and separationprotest),safe haven, and secure base are the three defining features of attachment and the functions of an attachmentrelationship (see Figure2). The process of attachmentformationtakes an average of 2 or 3 years (Bowlby, 1979, 1969/1982). The endpoint of the process is referredto as a goal-corrected partnership, in which the goal of proximity maintenance is adjusted for the child's ability to delay gratification and to mentally represent the caregiver's availability. At this point, caregiver and child can begin to negotiate the terms of their rela-

tionship and verbally communicateabout and coordinate theirrespectivegoals.

Attachment Disruption and Dissolution Bowlby's investigations of attachment stemmed from his interestin the effects of maternaldeprivation. He observed infants and young children who were being housed in residentialnurseries,separatedfrom theirfamiliarcaregiversfor extendedperiodsof timein some cases, forever. Bowlby was struck by two aspects of the children's responses. First, there was a remarkable degree of similarityacross children in the way they respondedto the separations,gradually revealing a predictableand invariantsequence of emotional reactions.The first was protest, which involves crying, active searching, and resistance to others' soothing efforts. This is followed by despair, characterizedby passivityandobvious sadness.The thirdand final phase is emotionaldetachment. The second striking aspect of the children's responses was thateven short-termseparationsseemed to have prolonged effects. Children who were reunited with their caregivers while still in the protest

Attachment

Figure 2. The definingfeatures of attachment.

AND ATIACHMENT CLOSE RELATIONSHIPS

phase exhibited heightenedanxiety over abandonment and an excessive need for physicalcontactandreassurance. This insecurity continued, in some cases, for months after the separation had ended. Those reunited after passing through the phase of despair tended initially to avoid contact with their mothers, as if they had coped with the separation by emotionally detaching. However, in time they resumed seeking contact and comfort. According to Bowlby, reactionsof anxiety and protest, even detachment,arehighly adaptiveresponsesto separationfrom one's primaryprotector.A child expresses distress because it usually brings the caregiver around. If, however, there appears to be no hope of reestablishingproximity,continuedexpressionsof distress not only risk attractingthe attentionof predators (a very realthreatin earlierperiodsof humanevolution) but also physicallyexhaustthe child. The characteristic inactivityof the despairphase keeps the child quietand Detachmentmakesposstill, allowing for recuperation. sible the resumptionof normalactivity, possibly even the search for a new attachmentfigure. Just as with routineproximitymaintenance,reactionsto prolonged separation reflect the functioning of the attachment system. Internal Working Models The attachmentsystem is an organism-levelsystem that is organizedand regulatedby social input,specifically by primarycaregiver responsiveness to distress signals. On the basis of repeatedinteractionswith the caregiver,infantslearnwhat to expect, and they adjust their behavior accordingly. These expectations form the basis of mentalrepresentations to use Bowlby's (or, term, internal working models) that can be used to forecast caregiver availabilityand responsivenessand thatinclude interrelated models of self and attachment figure.
Confidence that an attachmentfigure is, apart from being accessible, likely to be responsivecan be seen to turn on at least two variables: (a) whether or not the attachment figure is judgedto be thesortof personwho in generalrespondsto calls for supportandprotection; [and](b) whetheror not the self is judged to be the sort of person towards whom anyone, and the attachment figure in particular,is likely to respond in a helpful way. Logically these variables are independent. In practicethey are apt to be confounded.As a result,the model of the attachmentfigure and the model of the self are likely to develop so as to be complementary and mutuallyconfirming.(Bowlby, 1973, p. 238)

are determined in part by the responsiveness of the to caregivingenvironment individualneeds for comfort and security (Cassidy, 1988). According to Bowlby, these models guide thoughts,feelings, andbehaviorin subsequentclose relationships.

Individual Differences Theoretically and logically speaking, there is no limit to the amount and kind of variability that could exist in models of the caregiving environment. In reality, however, infants parse the flow of information about caregiver behaviors into a limited number of categories corresponding to responses to the following question: "Can I count on my attachment figure to be available and responsive when needed?" There are three possible answers to this question: yes, no, and maybe. That is, as concerns the internal working model, a caregiver is consistently responsive, consistently unresponsive, or inconsistent. In fact, these three types of caregiver responsiveness have been empirically linked to three major patterns of infant-caregiver attachment (Ainsworth, Blehar, Waters, & Wall, 1978). The procedure that Ainsworth developed for assessing attachmentquality-the Strange Situationwas designed to activate an infant's attachment system through repeated separations from the caregiver in an unfamiliar environment. It was also meant to activate the exploration system throughthe availability of attractive toys. Ainsworth was especially interested in whether and when infants sought proximity and contact, to what degree they accepted and were comforted by such contact, and whether their exploratory behavior was facilitated by the caregiver's presence. In other words, she was interested in observing proximity-maintenance, safehaven, and secure-base behaviors. Infantbehavior in the laboratorysetting was assumed to reflect expectations (internal working models) based on the caregiver's past responsiveness to the infant's bids for contact and comfort. The models themselves were assumed to be founded on a history of actual interactions. Home observations confirmed the link between daily caregiver responsiveness and infant laboratory behavior. The following paragraphs, based on passages in Ainsworth et al. (1978), describe the three major patterns. Secure. The behavior of securely attached infants matched Bowlby's conception of nature's prototype, in terms of proximity maintenance, comfort seeking, and the ability to use
5

Attachmenttheorythusimplies thatbeliefs andfeelings aboutthe self, especially social andglobal self-esteem,

HAZAN SHAVER &

the caregiveras a secure base for exploration.In the laboratory,the typical securely attachedinfant was distressed when the mother left the room, was comfortedby her return,and engaged in active explorationas long as she was present. During home observationsmade before the laboratoryvisits, caregiverswere judged to be consistently available and responsive. This is the most commonly observed pattern, averaging about 60% in Americansamples (Campos,Barrett,Lamb, Goldsmith,& Stenberg,1983). Anxious/Ambivalent. typicalcaregiverof The an anxious/ambivalently attached infant, observed in the home, exhibited inconsistent responsiveness to the infant's signals, being sometimes unavailable or unresponsive and at other times intrusive. In the laboratory,anxattachedinfantsappeared both ious/ambivalently anxious and angry and were preoccupiedwith theircaregiversto such a degree thatit precluded exploration.This is the most uncommonpattern, averaging about 15% in American samples (Camposet al., 1983). Anxious/Avoidant.At home, caregivers of avoidantlyattachedinfantsconsistentlyrebuffed or deflected theirinfants' bids for comfort,especially for close bodily contact. In the laboratory setting, these infants appeared not to be distressedby separations, avoidedcontactwith their

caregivers, and kept their attentiondirected towardthe toys (althoughwith less apparent interest and enthusiasm than the securely attached infants). On average, about 25% of American infants are classified as avoidantly attached (Camposet al., 1983). Note that the effects of psychological availability of the caregiver are remarkablysimilar to the effects observed for physical availability. Inconsistent responsiveness is functionally equivalent to short-term separations, and the corresponding attachment pattern-anxious/ambivalence-is characterizedby the same protest behaviors. Consistent caregiver unresponsiveness, like long-term absence, results in avoidance and apparent emotional detachment. Again, the behaviorsare traceableto the organization and functioning of the attachmentbehavioral system (see Figure 3). In recent years, researchers have identified a fourth pattern, disorganized/disoriented attachment (Main & Solomon, 1990), that is distinguishable by the absence of a coherent strategy for managing anxiety and that is manifested in a mixture of avoidant andambivalentbehaviors. Researchsuggests that this patternarises in infancy, when the infant's primary caregiver is depressed, disturbed,or abusive in some way (e.g., Crittenden, 1988; Main & Hesse, 1990).
SecureType

as Figure 3. Threemajorpatternsof attachment theycorrespondto variousaspects of normativeattachment-system dynamics.

ATTACHMENTAND CLOSERELATIONSHIPS

Source and Stability of Individual Differences Perhapsthe two most controversialclaims of attachment theory are that caregiver responsivenesslargely determines the quality of the attachmentrelationship and that working models of attachment tend to be stable. Many writershave criticizedattachment theory for its apparentfailureto acknowledgethe importance of infantcharacteristics, especially temperament (e.g., Campos et al., 1983; Lamb, Thompson, Gardner, Chamov, & Estes, 1984), and the theory's seemingly deterministicview of development,which some see as allowing little room for growthandchange.The debate over temperamentis by no means settled, but there is mounting evidence that both temperamentand caregiver responsiveness are importantinfluences on attachment quality. For example, one temperamental characteristic-distress proneness-has been linked to anxious/ambivalent attachment (Goldsmith & Alansky, 1987). On the otherhand,responsivenesstraining for the caregiversof distress-proneinfantsappears to overridethe risks for later insecureattachment (van den Boom, 1990). At least one study (Crockenberg, between tem1981) has suggested thatthe relationship peramentandattachment may be mediatedby maternal social support. To date, no investigation has shown to temperament be betterthancaregiverresponsiveness at predictingattachment classification,andno reported findings would lead to the conclusion that the consistency and quality of caregiver responsivenessare not of importantdeterminants infantattachment behavior. (See Colin, 1991, for a comprehensivereview.) In some ways, the debate over whether individual differencesare stable is even murkier.Researchsubsequentto Ainsworth'sidentification the threepatterns of (e.g., K. E. Grossmann& K. Grossmann,1991; Main, Kaplan,& Cassidy, 1985; Sroufe, 1983; Waters,1978) has indicatedthatthe patternsaregenerallystableover the first several years of life if family conditions are stable but can change if a child's social circumstances change (Egeland & Farber, 1984; Vaughn, Egeland, Sroufe, & Waters, 1979). Attachmenttheory does not dictate absolute stability of individual differences induced duringinfancy.Nevertheless,as with any cognitive construction,internalworkingmodels areresistant to change, in partbecause they tend to be overlearned and operateout of awareness, and in partbecause the defaultstrategyfor processingincominginformation is to assimilate it to existing schemes ratherthanmodify the schemes to accommodatethe information (Fiske & Taylor, 1991; Piaget, 1952). Whendefensive emotionalprocessesareintertwined with overlearnedcognitive and behavioralpatterns,as is hypothesizedto be the case with insecure-attachment

models, changecan be difficult. It seems moresensible to ask not whetherworking models are stable or unstable but underwhat conditions they are most and least likely to change. Bowlby has suggested several possible routesto change, including(a) the capacityto think aboutand reflect uponone's own workingmodels and (b) "corrective"relationshipexperience. Both come into play in good therapeuticrelationships(Bowlby, 1988). In light of what is knownaboutthe dynamicsof the attachment system andthe inbornneed for security, change would seem more likely to be in the direction of secure than insecure attachment.The attachment system's "primarystrategy"(Main, 1990) should always be to seek security if security is perceived as have possible. Consistentwith this, secure attachments been foundto be morestable(Egeland& Farber,1984). Further, moving froman avoidantto a secure model of relationshipswould involve acknowledging long-repressed insecuritiesand, as such, may necessitate an intermediate phaseof anxietyandambivalence.In fact, Heinicke and Westheimer(1966) observed that children reunitedwith caregivers after prolonged separations initiallyexhibitedavoidancebehavior,which was followed by a period of clinginess and preoccupation before normal(presepara(i.e., anxious/ambivalence) tion) behaviorwas eventuallyresumed. Attachment patterns set in infancy need not be fixed for life. As with cognitive structuresin general, internal working models of attachmentface the stability-plasticity dilemma (Grossberg, 1980). Mental models strive for stability but have to remain plastic if they are to continue to be adapative and useful. To date, longitudinal studies have indicated significant but not perfect continuity of attachmentpatternsover the first several months and years of life (Cassidy, 1988; Owen, Easterbrooks, Chase-Lansdale, & Goldberg, 1984; Waters, 1978). Ultimately, the degree of continuity from infancy through adulthood and the circumstances under which change is facilitated must be determinedempirically.

Attachment Beyond Infancy Attachmentis an integral part of human behavior "from the cradle to the grave" (Bowlby, 1979). The functions and dynamics of the attachmentbehavioral system arehypothesizedto be virtuallythe same across the life span. Presumably,this is because the neural foundationof the attachmentsystem remains largely As unchanged. Konner(1982) put it, "Theevolutionof the brainwould have to be consideredunparsimonious if it were not able to drawupon the same basic capacities of emotionandactionin the varioussettingswhere is strongattachment called for"(p. 298).
7

HAZAN & SHAVER

Despite some basic similarities, adult attachment differs from infant attachment in important ways (Weiss, 1982). First, childhood attachmentsare typically complementary.An attachmentfigure provides but does not receive care; an infant or child seeks but does not normally provide security. In contrast,adult attachmentrelationshipsare typically reciprocal,with each partnerbeing both a providerand a recipientof care. In addition, the attachmentrelationshipmoves from the level of external, observable interactionsto internallyrepresentedbeliefs and expectations(Main et al., 1985). Whereasinfantsand young childrenmay requirephysical contact with an attachmentfigure to feel completely secure, older children and adults are often able to derive comfort from the mere knowledge thattheirattachment figures can be contactedif needed (althoughthe need for physical-contactcomfort probably never disappearsentirely). What mattersis "felt security" (Sroufe & Waters, 1977), and adults have more options for achieving it thando infants. Another difference is that a child's primaryattachment figure is usually a parent, whereas an adult's primaryattachmentfigure is most commonly a peer, usually a sexual partner. Prototypicaladultattachment thus involve the integrationof three berelationships havioralsystems-attachment, caregiving, and sexual mating (Shaver, Hazan, & Bradshaw, 1988; Weiss, 1982). (See Figure4.)

Related to this are differences in what motivates proximity seeking. Anxiety and distress appearto be primarymotivators in people of all ages. However, adultproximity-seeking also resultfrom a desireto can or offer comfort (caregiving) or to engage in protect sexual activity (sexual mating). Attachmentsare hypothesizedto formin the contextof physicalcloseness, but the forces promotingsuch closeness may change with development. If adult peers begin to serve similar functions and satisfy the same needs for emotional support and security for which parents are primarily responsible during infancy and childhood, then at some point attachmentwill be transferredfrom parents to peers. The timing of this transferand the processes involved are not specified within the theory. However, from both the theory and the empirical literature on parent-child and peer relations (e.g., Furman & Buhmester, 1985; Gottman, 1983; Hartup, 1983; G. Levinger & A. C. Levinger, 1986; Rubin, 1980; Steinberg & Silverberg, 1986), a process model can be derived. The model (Hazan, Hutt, Sturgeon, & Bricker, 1991) is based on the assumptionthat, rather than being shifted in concert, all attachment functions are gradually transferred,one by one. According to this model, attachmentsbegin with proximity seeking (motivated by security needs when directed toward parents and by exploratory and affiliative

Prototypical Pair Bond

Figure 4. The componentsof a prototypicalpair bond. 8

RELATIONSHIPS AND ATTACHMENT CLOSE

PHASE DEVELOPMENTAL Infancy

Parents maintenance proximity safe haven securebase

Peers

EarlyChildhood safe haven securebase LateChildhood/ EarlyAdolescence securebase Adulthood

maintenance proximity

maintenance proximity safe haven maintenance proximity safe haven securebase

parents are gradually Figure 5. A model of attachmenttransferprocesses. In the course of normal development, figures as attachmentbehaviorsare, one by one, redirectedto peers. relinquishedas primaryattachment An Attachment Perspective on Relationship Data Having outlined attachmenttheory, we now briefly review some of the major bodies of data on close relationshipsin order to show how each could be organized and explainedby the theory.We have tried to preserve the fundamentaltheoretical constructs and processes,but we have not hesitatedto extend or mold In the theorywhereverwarranted. our opinion, the real strengthof attachmenttheory lies in its ability to explainfindingsderivedfromotherapproachesandother theoriesin termsof the same limited numberof proximal processes and evolved tendencies as it uses to explain infant-caregiverbonds. The degree to which attachmenttheory and we succeed in this endeavor reflects on the generative power of the theory. It is conceivable that each new finding in the field of personal relationshipswould necessitate the additionof a new theoretical construct. However, we attempt to a show thatattachment theoryis able to incorporate vast of such findingswithoutsignificantalterationor range addition.And, when additionsarerequired, they follow established for relationnaturallyfrom the principles ship functioningduringinfancy. of Ourfocus hereis on close relationships the attachmentvariety-which adultsprototypicallyformwith a romanticor sexual partner-but much of what we say is also applicableto other types of close relationships. is Ourreview of the literature necessarilyselective, but it is extensive enough to illustrateboth the diversityof available findings and the integrativecapacity of at-

needs when directed toward peers). In late childhood and early adolescence, close proximity provides the context that eventually fosters support-seeking(i.e., safe-haven) behavior. Repeated interactionsin which comfort is sought and providedor distressis expressed and alleviated may lead to relianceon the responderas a base of security. Parentsare never completely relinquished as attachment figures, but their place in a hierarchyof attachmentfigures, relative to thatof the place of peers, naturallychanges by adulthood.The model, still being developed and tested, is presentedin Figure5.

Summary of Attachment Theory As a result of selection pressuresover the course of evolution, humans are naturallypredisposed to form close relationships. Further, certain basic needs, of are which the need for securityis the most fundamental, best satisfied within social relationships. Behavioral systems have developed to promotethe satisfactionof all needs that have been importantfor survival and reproduction.The dynamics of close relationshipstheirformation,maintenance,and dissolution-can be understoodin termsof the functioningof thesesystems. Individualdifferencesin thoughts,feelings, andbehaviors in close relationshipsarelargelydetermined the by social environmentsto which individualshave had to adapt. Finally, these individual differences are maintainedby mentalmodels constructedout of actualrelationshipexperience.

HAZAN & SHAVER

tachmenttheory.The review is organizedaroundwhat traditionallyhave been considered fundamentalquestions in the study of close relationships.

What Makes a Potential Relationship Partner Appealing? From an attachmentperspective, humans possess basic needs that are naturallysatisfied by social relationships,such as the needs for emotionalsupport,care, and sexual gratification. Theoretically, each need is regulatedby a distinct behavioralsystem designed to respondto specific social cues. We should, therefore, be attractedto people who display these cues. Given that the most basic need (for felt security) is regulated by the attachmentsystem, and given thatthis system is assumed to function similarly across the life span, characteristics a potential of among the most important partnershould be the very characteristics shown to be centrally importantin the selection of an attachment figure during infancy and childhood-namely, familiarityand responsiveness. attractionis not only the first stage of Interpersonal many relationships,it is also the topic with which the field of close relationshipsbegan, in the pioneering work of Berscheidand Hatfield(Berscheid& E. Walster, 1974) and Rubin (1973). The list of featuresand characteristics have been shown to be determinants that of attraction long andheterogeneous reviews,see is (for Aronson, 1988; Berscheid, 1984). Forexample,we are attractedto people whose values, attitudes,opinions, and even physical features are similar to our own (Hinsz, 1989; Rubin,1973). Unless we dislikesomeone to begin with, increased exposure is associated with increased liking (Zajonc, 1968). Socially responsive children are more attractive to their peers (Rubin, 1980). Among all the people with whom we could socialize, we typically choose those who live or work nearby (Festinger, Schachter, & Back, 1950; Newcomb, 1961). We have a predilectionfor people who smile frequently, who are good-humored, and who make us laugh (Folkes & Sears, 1977). We are espeto cially attracted people we know find us appealing(A. P. Aron et al., 1989). Anxiety tends to enhanceattraction (Dutton & A. P. Aron, 1974; Hatfield & Rapson, 1987) and the desire to affiliate (Schachter,1959). These data indicatethatattraction anotherperson to tends to increasewith familiarityand the likelihoodof a positive response.Anyone who is like us (in appearance, attitude,etc.) or who likes us, is smiling, or makes us smile is more likely thanan unfamiliar unresponor sive personto be viewed as safe, approachable, thus and attractive. intensifiesthe need to be Anxiety apparently near anotherperson or makes that person's security10

promotingpotential more salient. Within attachment theory,anxiety is a signal to get closer. is However,attachment not the only behavioralsystem motivatinginterpersonal attraction.As stated earlier, adultlove can be conceptualizedas a joint function of the attachment,caregiving, and sexual mating systems (Shaveret al., 1988), so, theoretically,attraction can result from one person's seeing the possibility of another person's meeting attachment,caregiving, or sexual needs. Another's attactiveness is thus determinedby the type of relationshipthatis sought and the kinds of needs or desires thatare likely to be satisfied. Fromanattachment perspective,the multiformfeatures of attractivenessare reducible to a small number of conceptualcategoriesthatcorrespondto the behavioral systems relevantto close relationships,each of which is triggeredby differentand distinctcues. The caregiving system, for example, responds to babyishfeatures,distress,andvulnerability-which in the case of adultsmay include self-disclosures of fear or weakness,or the vulnerabilityinherentin the letting down of defenses. A personwho wants to providecare should logically be attracted someone who seems to to needsuch care.Cues associatedwith targetsof caregiving constituteone interpersonal-attraction categorythat has been understudied,perhaps due to the lack of a theorythatwould encouragesuch investigations.Also, attraction involve adolesmanystudiesof interpersonal cents andyoung adults,for whom caregivingmay be a relativelyundevelopedor less salient need. There are, in fact, age-relatedchanges in the relevance of such cues. One study(Fullard& Reiling, 1976) documented a developmentalshift in preferencefor infant relative to adultfaces. Before puberty,both males and females of preferred photographs adultfaces;afterpuberty,they spent more time looking at infantfaces. A person who seeks gratificationof sexual needs should be attractedto someone who displays cues of sexual availabilityand value. In sexual relationships, an attractivephysical appearancecan be remarkably important(e.g., E. Walster, Aronson, Abrahams, & Rottmann,1966). Also, there appearsto be considerable cross-cultural cross-ageconsistencyin what is and viewed as physically attractive,including characteristics associatedwith youthand/orhealth-smooth skin, clear eyes, white teeth, a lively gait. Also desirablein a potentialsexual partner-especially in a male being evaluatedby a female-is evidence of social statusand materialwealth,physical resources,such as popularity, strength,intelligence,and wit (Buss, 1985). Because romanticlove includes elements of attachment,caregiving,andsexuality,the self-attribution that one is "in love" may result from noticing that another personis especially responsive,needs to be takencare of, or is a sexual "turn-on." This could easily result in

ATTACHMENT CLOSE AND RELATIONSHIPS misunderstandings and confusion: Responsiveness need not mean sexual interest,sexual interestneed not imply more generalresponsiveness,and the wish to be taken care of carries no guaranteeof reciprocalcare. People can mistakenlybelieve that their entire prototype of romanticlove (Fehr, 1988) is aboutto be realized when in fact only one component is actually present. To say thatwe are attracted someone is to say that to we would like to be physically and/orpsychologically close to thatperson.Withinan attachmentframework, what motivates such proximity seeking depends on which social-behavioralsystem is activated.Conceptualizing interpersonalattractionin this way makes it possible to reduce the many interpersonal-attraction factors to a few conceptually meaningful categories. This in itself advances our understanding attraction of beyond the mere documentionof factors.In our view, however, the real strengthof attachmenttheoryin this domain is its ability to go beyond classification to an explanationof when andwhy particular characteristics of a person should be attractive. Even though the motivation to seek closeness is hypothesized to be somewhat different and more complex for adults than for infants, the first phase in what may eventually become an attachmentrelationship is remarkablysimilar (Shaver et al., 1988). Like an adult in love, the infant is preoccupied with and notably vigilant for signs of the target person's responsiveness. Further,for infants as well as adults, one's emotional state hinges on the target person's behavior, with responsiveness generally leading to feelings of security and joy and unresponsiveness evoking anxiety and distress (Ainsworth et al., 1978; Tennov, 1979). In both kinds of relationships-between infant and caregiver and between adult lovers-strong forces attract the individuals to each other and, in some cases, hold them together long enough for an emotional bond to develop. In both, the formation of such a bond is facilitated by close physical contact. Only between parents and children or between adult lovers is prolonged bodily contact considered normal. It is noteworthy that laypersons judge these two types of relationshipsto be the "closest" (Berscheid & Graziano, 1979). For good or ill, the intensity of the need for close contact eventually diminishes (Fisher, 1992; Traupmann & Hatfield, 1981)-an importantfact that needs to be explained. Logically, the safe-havencomponentof attachment would be expected to develop within the context of closeness. Consistentwith this, researchers have found that mutual attractionand sexual passion are most importantearly in a relationship, but the degree to which a partnerprovides comfort and emotional supover time (Reedy, portbecomes increasinglyimportant Birren, & Schaie, 1981). In attachmentterms, what eventuallycomes to mattermost is whetherthe partner serves as a reliable haven of safety. Mutualattraction and sexual interest can get couples together, but, if partnersfail to satisfy each other's needs for comfort and security, dissatisfactionwill likely result. Kotler (1985) found that sensitive and responsive care, not sexual attraction,was the most accuratepredictorof maritalstrength. Whendoes a relationship become a base of security? Itis relativelysafe to assumethatparentsarecommitted to their offspring for life. This commitment is not between typicallyquestionedor broken.Commitments peers,however,tendto be less robustandmoresusceptible to both internalandexternalinfluences(G. Levinger, 1976). It is likely that,only afteranextendedperiod of time and/orafter an explicit commitmenthas been made, a peer relationshipcan serve as a secure base with a degreeof certaintyapproaching of the base that providedby parents.Marriage,for example, is usually accompaniedby a legally binding, public promise to care for the partner until death. 11

How Is a Relationship Formed, and How Does It Develop? Some attractions develop into relationships, and some do not. Also, it is generally assumed and well documentedthatrelationships, afterbeing formed,tend to change over time (e.g., Berscheid & E. Walster, 1974; Huston, Surra, Fitzgerald, & Cate, 1981; G. Levinger, 1983; G. Levinger& Snoek, 1972; Lewis & Spanier, 1979; Taylor & Altman, 1987; E. Walster& G. W. Walster, 1978). A nascent relationshipis obviously different from an established relationship,and discoveringjust what changes andwhy is an important taskfor relationshipresearchers has been the focus and of much study (Altman & Taylor, 1973; Kerckhoff& K. E. Davis, 1962; G. Levinger, 1983; Lewis, 1973; Murstein, 1976; Reiss, 1960). In general, efforts to identify a uniform sequence of stages for all close relationships have met with great difficulty (e.g., Hustonet al., 1981). From an attachmentperspective,the formationof a close relationshipbetween two individualsof any age typicallyformsin the contextof close physicalproximity. Inotherwords,for an attachment formtheremust to be a strong force promoting closeness. In infancy, proximityis regulatedby theattachment systemandthe infant's need for security. In adult romanticrelationships, the sexual mating system (sexual attraction)is hypothesized to be anotherprimaryinstigatorfor the proximity seeking that is the first step towardattachmentformation.

HAZAN SHAVER & In terms of the three defining attachmentfeatures described earlier (see Figure 2), the process of attachment formation, at any age, is hypothesized to involve the same sequence: proximity seeking followed by safe-haven behavior followed by the establishment of a secure base. In some cases, of course, the process will not be completed. As suggested earlier, the major difference between infantcaregiver and adult-pair bonds is in the motivation for seeking closeness in the first place. The hypothesized process is siummarized Figure 6. in An attachment perspectiveaddsto ourunderstanding of how close relationshipsdevelop and change over time and helps explain some of the phenomena that have been repeatedlydocumentedand described.The attachment view of relationshipdevelopmenthelps account for some importantcharacteristicsof the time course of close relationships. It also makes implicit predictionsconcerningboth the natureand timing of milestones and transitionsin developing reimportant lationships.The threecomponentsof a prototypical pair to bond,each corresponding a separatebehavioralsystem-attachment, caregiving, and sexual matinggradually become integrated. The hypothesized which is simdevelopmentalcourseof this integration, ilarto Sternberg's(1986) proposalconcerningthe temporal developmentof various components of love, is in represented Figure7.

What Makes Relationships Satisfying and/or Enduring? As the wording of this question implies and as the data confirm, an enduringrelationshipmay not be a satisfyingone. Therefore,it is essential thata theoryof

PHASE RELATIONSHIP Initialattraction Established relationship Goal-corrected partnership, or long-term committed relationship

ATTACHMENT BEHAVIORS
*proximity maintenance proximity maintenance safe haven proximity maintenance safe haven secure base

Figure 6. A model of attachment formationprocesses. Attachment formationbegins withproximityseeking, which may be motivatedby a desire for security, sexual gratification,or theprovision of care. In subsequentphases, the partner may gradually become a safe haven and then a secure base. *Motivations proximityseeking include for security (attachment),nurturance(caregiving), and sexual attraction(sexual mating).
7attachment caregiving
5-

Importance, intensity

4321A ,-??-.....5....... -

K i

sexual mating

4./

i I

I I

I I

II

Time in Years

Figure 7. The developmentalcourse of a prototypical adult attachmentrelationshipin terms of three behavioral systems. 12

ATTACHMENTAND CLOSERELATIONSHIPS

close relationshipsspecify and explain the factors that predicteach. According to attachment theory, a relationship is satisfying to the extent thatit meets basic needs. At any age, attachment qualityturnsin largeparton theanswer to the question, "CanI trustmy partnerto be available and responsive to my needs?"Trustpromotesself-disclosure and the development of intimacy (Reis & Shaver, 1988). Trustis also associatedwith open communicationabout and the "voicing"of needs (Holmes & Rempel, 1989; Rusbult& Zembrodt,1983). Satisfying relationshipsare not conflict free, but they involve the kind of trustthatallows couples to argueconstructively (Rands, G. Levinger, & Mellinger, 1981) and to engage in effective problem-solvingbehaviors(Kobak & Hazan, 1991). The "hiddenagendas"that interfere with successful conflict resolution are often about unmetneeds (Gottman,Notarius,Gonso, & Markman, 1976). Attachmenttheory tells us what the needs are likely to be andexplainswhy trustin a partner's responsiveness to these needs is critical. Many unsatisfying relationshipsendure, and these are among the most challenging for researchers.What makesa partner decide to leave? The belief thatimportant needs could be bettermet in anotherrelationshipwhat has been called comparison level for alternatives-is influentialin decisions to continue or end a relationship(Kelley, 1983; Rusbult,1980,1983). In defining relationshipcommitment, it is helpful to distinguish between the desire to continue a relationship because it is satisfying and the tendency to stay simply because the constraintsagainst breakupseem too great to overcome (Stanley, 1986). Externalconstraintsto breakup includejoint ownershipof property, poverty,children,and a relativeabsenceof alternatives (G. Levinger, 1976). Attachment theory suggests an additional factor that may contribute to the maintenance of an unsatisfying relationship-the emotional bond of attachment. Recall that an attachment is typically formed in the context of close proximity. Weiss (1982) argued that proximity alone can maintain the bond. Often couples are unawareof the bond between them until it is disrupted or threatened in some way (Berscheid, 1983). Even a burdensome and unsatisfying relationship can contribute to one's sense of security. Perhaps the best evidence for the securitypromoting function of an unhappyrelationship is the intense anxiety that typically accompanies separation (Weiss, 1975). Bowlby (1973) theorized that separation from an attachmentfigure is one of many naturalcues to potential danger and, as such, triggers a fear response and, in turn, attachmentbehaviors. To summarize,from an attachmentperspective,relationshipsatisfactiondepends largely on the satisfac-

tion of basic needs for comfort, care, and sexual gratification. Trustin a partner'swillingness and ability to meet needs is determinedin partby the partner'sactual behaviorand in partby the expectationsor comparison levels (Kelley, 1983; Rusbult,1983) each personbrings to the relationship.A history of close relationships lacking in trust, for example, might be expected to result in the kind of minimal expectations that could lead one to stay in an unsatisfyingrelationship.Thus, relationshiplongevity may be influenced by relationship history.Afteran emotionalbond has developed, it can act as a psychological tether that provides some security and holds two people together regardless of whetherthey still enjoy being together.Anxiety resulting fromcontemplationof or attemptedseparationcan activateattachment behaviorsthatlead one back to the relationship,unless there is an available and willing alternative.

What Are the Precursors and Reactions to Relationship Dissolution? Justas researchhas uncoveredmultiplecorrelatesof relationshipsatisfaction, scores of studies have been directed at revealing the causes of breakup (e.g., Felmlee, Sprecher,& Bassin, 1990; Gottman, Markman,& Notarius,1977; G. Levinger,1966; Lund,1985; Simpson, 1990). At the risk of sounding glib, one can conclude from this research that failure to engage in behaviors that enhance relationship satisfaction and longevity rendersdissolution more likely. If partners lack trustin each other and, as a result, do not openly and clearly communicatetheir thoughts and feelings, do not engage in effective strategiesof conflict resolution, or are not committed to remainingtogether, the that will probability the relationship endureis lessened. Structural factors of the sort mentionedearlier-children,limited financialresources,and religious or societalprohibitions-also figureprominently a couple's in decision to end or maintaintheirrelationship.As long as each memberof a pairhas confidence (i.e., trust)in the other's ability and willingness to supply essential relationalprovisions, each will be motivatedto maintain the relationship.In G. Levinger's (1976) terms, internalattractions will be so strong that externalbarriersandalternative attractions be of little practical will importance. The desire to leave a relationshipnecessarilycomes after the relationshiphas been formed. The question, then,is whatchangesbetweenthe time a persondecides to entera relationshipand the time a decision is made to leave. The answer, we believe, lies in a process model of attachmentformation.Relationshipsatisfactionalways reducesto whetherneeds arebeing satisfied 13

HAZAN & SHAVER

or not (Shaver& Hazan,1984). The problemis thatthe of relativeimportance variousneedschangesover time. In fact, what gets two people into a relationship may be what mattersleast in the long run. If sexual passion is indeed the initial motivating force in the formationof many adult pair bonds, a decline in satisfaction is inevitable unless the relationship meets other needs afterthey have become important. peoUnfortunately, ple in the throesof romanticpassionmaygive relatively little thought to whether the people to whom they are attracted makereliablelong-termprovidersof care will and support-which in time will come to dominate their feelings about the relationship.According to the hypothesizedtime course presentedin Figure7, dissatisfaction would be expected to peak aroundthe time comhas thatintense attraction faded and the partner's petence as a haven of safety and secure base assumes relativelygreaterimportance.It should be at this point in a relationship that expectations, alternatives,and constraintscome into play in a majorway. Attachmenttheory has much to say abouthow people respond when a relationship ends. Response to separationand loss, afterall, was the topic with which Bowlby began his inquiries.It is difficult to discuss the withoutconsideringdisruptions functionof attachment and separations,because the system that regulatesattachment feelings and behaviors includes "built-in" responses to disruption.Even though there is tremendous cultural variation in associated rituals and customs, the human response to the breaking of an attachmentbond hardly varies (Gorer, 1973; Marris, 1958; S. I. Miller & Schoenfeld, 1973; Palgi, 1973). Moreover, the way in which adults respondto attachis mentdisruption notessentiallydifferentfromtheway infantsand childrenrespond(Bowlby, 1980; Hazan& Shaver, 1992; Heinicke & Westheimer,1966; Parkes & Weiss, 1983). The first reactionto the disruptionof an attachment relationship,whetherdue to deathor voluntaryseparabehavior.Individuals tion, is intenseseparation-protest and preoccupiedwith reportfeeling agitated,anxious, thoughtsof the lost partner, coupledwith a compulsion to search for him or her, as though trying to undo the loss even if it is consciously known to be irreversible. Eventually,with the realizationthatthe loss cannotbe recovered,therecomes a periodof deepsadness,during which intense activity and ruminationgive way to depression and despair. Many individualsexperience an unusualandmarkedlack of concernaboutor interest in life and other people. Gradually,the sadness subsides, and most people achieve an adaptivedegree of emotional detachmentfrom the lost partner return and to ordinaryliving. In adults especially, constructinga causal accountof the loss helps bringaboutacceptance and detachment(Harvey, Orbuch,& Weber, 1990). 14

When the attachmentsystem is activated-for example, by the sudden unavailability of the primary attachment-the naturalresponse is to seek proximity to the attachmentfigure, and this seems to happen whether or not establishing proximity is possible or even rationally desirable. That an estranged partner may still be attainable can foster not only hope but also protractedprotest (Weiss, 1975, 1988). In some cases, chronic activation of the attachmentsystem in the absence of the former partnermay result in premature attachment to another person. Likely candidates would include individuals providing support and care during the difficult postseparation period. By the time a couple decides to separate, all former fondness and affection may have eroded. For this reason, many newly separatedindividuals, especially the ones who initiate separation, are surprisedwhen they begin to experience a compulsion to be near the former partner.Weiss (1975) argued that this very common feeling is due to the persistence of attachment and suggested that an attachmentbond can be brokenonly by an extended period of separation.The same sequence of responses would not be expected unless an attachmenthad been fully formed, which may take several years. Weiss (1988) noted that responses to divorce are distinctly different in individuals marriedfor fewer than 2 years. Most of what attachmenttheory has to say about separationand loss applies equally well to loneliness (Peplau & Perlman, 1982), which seems to take two major forms-emotional and social isolation (Weiss, 1973). Emotional isolation is the kind of loneliness associated with the lack of an intimate companion, whereasfeelings of social isolationresultfromthe lack of a social networkor sense of community. The two forms of loneliness correspondwell to Bowlby's distinctionbetween the attachment affiliationbehavand ioral systems, which are thought to have different functions.In addition,the two types of loneliness have differentsymptoms,causes, and cures (Rubenstein& Shaver, 1982). Weiss (1973) viewed loneliness as an adaptive if uncomfortableemotional state because it serves as a reminder important that social needs arenot being met, and it tends to continue until corrective action is taken.Again, reactionsto the loss or absence of an emotionalbond are hypothesizedto be the direct resultof the functioningof the attachmentbehavioral system.

What Is the Role of Relationships in Overall Functioning? Close relationships important the lives of most are in people andareamongthegreatestsourcesof subjective

ATTACHMENTAND CLOSERELATIONSHIPS

well-being (Freedman, 1978; Veroff, Douvan, & Kukla, 1981). As furtherevidence of the significance of close relationshipsfor healthy functioning,the disruption or loss of a relationship, especially through divorce, makes one more susceptible to everything from automobile accidents to alcohol abuse to admission into a psychiatricfacility (Bloom, Asher,& White, 1978). In addition,the grieving and lonely are vulnerable to disease, as studies of their heartsand immune systems have demonstrated (e.g., Kiecolt-Glaser,Garner et al., 1984; Kiecolt-Glaser, Ricker et al., 1984; Lynch, 1977), and are at greater risk of death from cancer (Goodwin, Hurt, Key, & Sarret, 1987). Social deficiencies and losses jeopardizenot only health and and happinessbutjob performance achievementas well (Baruch, Barnett, & Rivers, 1983; Lee & Kanungo, 1984; Vaillant, 1977). The diverse and numerous links between relationships and overall functioning are well documented but poorly understood. Why should the absence of socioemotional connections have such profoundpsychological, physical, and even occupational consequences? We believe that Bowlby's insights supply the missing link: Social deprivation is harmful because it thwarts the satisfaction of inborn needs. Powerful emotional cues signal when these crucial needs are not being met, and the result is a subjective state of anxiety and discomfort. Stated differently, social deprivation is stressful and, as with any prolonged stressor, can be both psychologically and physically harmful. Because of the interrelations among behavioral systems, malfunctioning in one system can cause dysfunction in another. Recall that, in studies with infants, attachment and exploratory behavior are closely intertwined. Smooth functioning of the exploratory system requires a quiescent (satisfied) attachment system, which itself depends on the availability and responsiveness of an attachmentfigure. In adulthood, the balance between relationships and work, between emotional connectedness and independent activity, is in importantrespects similar to the attachment-explorationbalance markinghealthy functioning in early life (Hazan & Shaver, 1990). In Bowlby's (1979) words,
Human beings of all ages are happiest and able to deploy their talents to best advantagewhen they are confidentthat,standingbehind them, thereare one or more trustedpersonswho will come to theiraidshould difficulties arise. (pp. 103-104)

How and Why Do Individuals Differ in the Way They Think, Feel, and Behave in Relationships? Up to this point, we have emphasized normative aspects of close relationships. However, one cannot ignore the immense variability in the ways people relate to one another. Some people fear intimacy, whereas others embrace it (Hatfield, 1984). Some self-disclose to an excessive degree, whereas others disclose little or not at all (Altman & Taylor, 1973). Attitudes about romantic involvements range from game playing to pragmatism (C. Hendrick & S. Hendrick, 1986). Relationships can involve commitment without passion or passion without intimacy (Sternberg, 1986). Partners may attribute each other's problematic behavior to character flaws or explain it in terms of situational factors (Fincham, Beach, & Nelson, 1987). In response to conflict, partners may withdraw or accommodate (Rusbult, Zembrodt, & Gunn, 1982). Communication may be open and coherent or defensive and disorganized (Bretherton, 1990). Relationships appear to be as diverse as the individuals involved. Ideally, the multitude of differences could be reduced to a more manageable number of conceptual categories.If we accept thatconfidence and trustin the responsiveness of others are central issues in close thena few broadcategoriesof individual relationships, differencesnaturally follow. As Mainet al. (1985) saw it, the social environmentcan be perceived as consistentlyresponsive,inconsistentlyresponsive,or consistently unresponsive to an individual's attempts to establishsecurity-promoting closeness. Given thatthis issue appearsto be the same in infancy and important adulthood,we assume that the importantindividualdifferencecategorieswill also be essentially the same. One would not expect to see the exact same behaviors but rather same basic strategiesfor maintainingfelt the security. In emphasizing normative attachment,we have actually been describing security or the secure attachmenttype. In this section on individual differences, we focus insteadon the otherend of the dimension-insecurity. The strategy associated with inconsistent responsiveness-anxious/ambivalent (preoccupied) attachment-is characterized a lack of confidence in the by reliableresponsivenessof others.The proximalgoal of all attachmentbehavior is to achieve a state of felt security.In the case of anxious/ambivalent attachment, thisis attempted at timesaccomplishedby devoting and immensementalenergy and behavioraleffort to keeping others close by and engaged. It is manifested in intensifiedexpressionsof distressandangeranddiminished exploratoryactivity. 15

Thus, attachmenttheory explains why and how close relationshipsplay a centraland critical role in overall feelings aboutand adjustmentto life.

HAZAN & SHAVER

In studies of adult attachment,anxious/ambivalent attachmentis associatedwith obsessive preoccupation with a romantic partner's responsiveness; falling in love easily; being extremely jealous; being subject to fear, anxiety, and loneliness (even when involved in a relationship); having low self-esteem (Collins & Read, 1990; Feeney & Noller, 1990); and experiencing a higher rate of relationship dissolution (Hazan & attachedalso Shaver,1987). The anxious/ambivalently tend to view partnersas reluctantto commit and as inadequate or insufficiently attentive caregivers (Kunce & Shaver, 1991). They engage in indiscriminant and overly intimateself-disclosure (Mikulincer& Nachshon, 1991) and assert their own feelings and needswithoutadequateregardfor thepartner's feelings and needs (Daniels & Shaver, 1991). In laboratory anxious/amproblem-solvingtaskswith theirpartners, bivalentsubjectstendedto expressdysfunctionalanger (Kobak & Hazan, 1991). They also reported more physical and psychological symptoms (Fiala, 1991; Hazan& Shaver, 1990) andhadgreaterdifficultymaking friends in a new setting (Hazan& Hutt, 1991b). At work, anxious/ambivalenceis associatedwith distraction, procrastination, and suboptimal performance (Hazan & Shaver, 1990) and, in discussions of attachmenthistory,with overly effusive andpoorlyorganized discourse (Main et al., 1985). (For a review of additionalcorrelatesof anxious/ambivalent see attachment, Shaver & Hazan, 1993.) In contrast,avoidantattachment believed to result is from consistent unresponsiveness.The avoidantstrategy for maintainingfelt securityinvolves avoidanceof intimate social contact, especially in stressful or distressingcircumstances,andcompensatory engagement in nonsocial activities. According to researchon adult avoidanceis manifestedin fearof intimacy attachment, anda tendencyto maintaindistancein "close"relationships, with pessimistic views of relationships and a relatively high rate of relationshipdissolution (Hazan & Shaver, 1987). When answeringinterviewquestions concerning childhood relationships with parents, avoidantlyattachedadults(a) use idealized descriptors but are unable to provide supportingexamples (Main et al., 1985) and (b) show spikes in skin conductance when probed for such examples (Dozier & Kobak, in press). They avoid self-disclosure and experience discomfortwith relationship who do self-disclose partners (Mikulincer & Nachshon, 1991). They are more sus& ceptible to suddenreligious conversion(Kirkpatrick Shaver, 1990). They are judged by their peers to be hostile (Kobak& Sceery, 1988). They tendto use work to avoid social interaction(Hazan& Shaver,1990) and are prone to engaging in uncommittedsexual relations and using alcohol and other substancesto reduce tension (Brennan,Shaver,& Tobey, 1991). (Foradditional 16

see findingson avoidantattachment, Shaver & Hazan, 1993.) of Theoretically,these two majorpatterns insecurity arebasedon internal workingmodels constructedfrom actual attachment experience,beginningwith parents. The anxious/ambivalentstrategy is "logical" in the sense that it reflects a history of inconsistent responsiveness. Expecting close relationshippartnersto be somewhatunreliablemay lead to heightenedvigilance and fears of abandonment and neglect, both of which can interferewith nonattachment activities.The avoidant strategy is equally logical in light of a history of frequentrejectionor inhibitionson physical affection and intimateemotional expression. Such experiences can lead to an avoidanceof closeness, extremeself-reliance, and a habit of regulating anxiety by keeping oneself distracted.In line with social psychological researchshowing thatactionsoften follow frombeliefs and interpersonalschemas in a way that encourages repeatedconfirmation(e.g., Snyder & Swann, 1978), internalworking models may have self-fulfilling effects on social behavior and social-informationprocessing. For example, there is evidence that mates are selected for theirability to confirmattachment-related expectations, even if the expectations are negative & (Kirkpatrick K. E. Davis, in press;Swann, Hixon, & De La Ronde, 1992). We believe thatindividualdifferences in attachment,mediated by internal working differmodels, may underliemanyof the interpersonal ences thathave been discoveredby researchersworking from other theoretical bases. Further, insecure attachment might be at the root of many dysfunctional behaviors contributingto relationshipdissatisfaction anddissolution. Despite forces favoring the stability of individual differences in attachment,change is always possible. Forexample,the experienceof just one important relationship that disconfirmsinsecureexpectationsof unreliability or rejection increases the likelihood of forming a secure attachmentin adulthood (Hazan & Hutt, 1991a). In most cases, these disconfirmingrelationship experiences were formed with nonparental adults (e.g., teachers, relatives) during childhood or with romanticpartners duringlate adolescence or early adulthood. Consistentwith studiesof change in infancy and childhood, secure attachmentis the most stable pattern. The distribution adultsacross attachmentcategoof ries-55% secure, 25% avoidant, and 20% anxious/ambivalent-has been replicatedin many studies in several differentcountries (e.g., Feeney & Noller, 1990; Hazan & Shaver, 1987; Mikulincer,Florian, & Tolmacz, 1990). The proportionsare similar to those found in infantstudies using Ainsworth's StrangeSituationprocedure(Camposet al., 1983). To date, there

ATTACHMENTAND CLOSERELATIONSHIPS

have been no reliablegenderdifferencesreportedin the distributionof subjects across the threecategories.

Gender Differences No treatmentof close relationshipswould be complete without some discussion of gender differences. Males and females differ significantlyin theirstyles of communication (Tannen, 1990), in their tendency to affairs (Skolnick, 1978), engage in extra-relationship andin theirskill at readingnonverbalcues (Hall, 1978). They differ as well in their likelihood of initiating breakups (Hill, Rubin, & Peplau, 1976) and in their subsequent adjustment(Wallerstein & Kelly, 1980). They also differ in the degree to which they value physical attractiveness (Berscheid,Dion, E. Walster,& G. W. Walster, 1971) and materialresources(Buss & Barnes, 1986) in a potentialmate. The measurewe developed for assessing individual differences in adult attachment (Hazan & Shaver, 1987) might have been expected to show genderdifferences. The anxious/ambivalentpattern sounds very much like the clingy, dependentaspects of the female stereotype,and the avoidantpatternstronglyresembles the stereotypical intimacy-evadingmale. That males and females do not fall disproportionately into either lends support to Bowlby's claim that all category humanbeings have an inbornneed for felt security. Informalexaminationof the data suggests thatgender differences lie primarilyin the domainsof caregiving and sexuality, ratherthan attachment.In general, females are more oriented toward caregiving, and males aremore orientedtowardsex. Some of the major differencesbetween lesbianandgay male relationships seem to support this view (Peplau & Gordon, 1983). Buss (1985) developed a sociobiological theory of human mate selection that is also consistent with this conceptualization of gender differences. However, within an attachment framework, genderdifferencesdo not entail a biological explanation. Because the caregiving andsexual matingsystems develop laterthanthe attachment system, it is likely thatthey aremoresubject to sex-role socialization pressures.The extent of biological versus social causationis still a legitimatematter for dispute and furtherresearch. Concluding Remarks Our review of the literaturehas necessarily been selective, and there are many additionalfacts that we have not attemptedto integrate,although many seem amenable to explanationwithin an attachmentframework. Even so, we do not believe that the framework

we have proposedwill be able to explain all important relationshipphenomena. Each theory has its boundaries,andattachment theoryis no exception. In fairness to Bowlby, he was not attempting to explain every His aspector type of close relationship. aim was simply to explain the structureand functions of attachment, and it took him threevolumes to do it. Even in beginwe ning to extend his theoryto adultattachments, have takengreatliberties. At the outset, we said that we see attachmentand interdependenceas largely complementary theories. We asked what a good theory of close relationships should be able to do. A reasonableresponse is that it should offer a precise and operational definition of close and should emphasize relationshipsmore than individualrelationshippartners.Interdependence theis superiorto attachmenttheory in both regards, ory and, in its models of interactionand types of transformations that occur over the course of an interaction, interdependencetheory is unparalleled.Attachment theory, in contrast,emphasizeswhat of psychological value individualswill try to accomplish in their interactions.In otherwords, it suppliesmuchof the psychological substance of what Kelley (1983) called the is given matrix.Attachment a motivatedmodel, andthe motives on which it focuses are based in biology. We are active, biological organisms that build internal structures.As with other behavioristicmodels in psychology, the strengthof interdependence theory rests with its operationaldefinitions and measurementof observables.However,it would be a mistaketo believe that all importantrelationshipphenomenacan be reduced to observableones. The theories that emphasize socially constructed, shared meanings and narrativesalso add something importantto attachmenttheory, which, because of its roots in ethology, does not stress the uniqueproperties of humanverbalcommunication.Nevertheless,attachment theory may provide some of the themes of and constraints on interpersonal story construction, whereas communication-oriented theories add to attachment theory's conceptualizationof the ways in which internalworking models of self, relationships, and relationshippartnersget constructed.A complete integrationof attachment,interdependence,and communication theories would require more space and expertisethanwe have at present,butwe firmlybelieve thatsuch a union would be fruitful. It is common to hearattachment theorydescribedas a theory about three (or maybe four) types of babies. Such emphasis on individualdifferences does not reflect the mainthrustof attachment theory,which is first and foremosta normativetheory.An addedstrengthis that it can also account for the nature and form of individualdifferences. Ainsworth's creation of an in17

HAZAN & SHAVER

novative paradigmfor assessing attachment quality in infancyfocused attentionon individualdifferences.By creating a simple self-reportmeasureof adult attachment styles, we inadvertentlyhelped extend the individual-differences approach to the study of adult relationships. The normative implications of attachment theory have rarelybeen spelled out or tested. The prevailing problem with the overemphasison individual differences has been a lack of consensus concerning how to measure these differences beyond infancy. Four or five different methods are currently used to assess attachment in toddlers and children (Greenberg,Cicchetti,& Cummings,1990); for adults, thereare two highly similar interviews(Bartholomew, measuresthat 1990; Main, 1991) andseveralself-report are derivatives of our 1987 measure (e.g., Collins & Read, 1990). Currentdebate includes questions about whethermeasuresof adult attachment should concentrateon parent-childdyadsor adult-pair bonds,how the various methods(e.g., interview,self-report)compare, andhow to conceptualizethe differences(i.e., in terms of dimensions or types). There is even ongoing debate about how many major types of attachmentexist in adulthood(Bartholomew& Horowitz,1991). We think it is importantto point out thatattachment theorydoes not standor fall with the success of any individual-difference measure.Regardlessof which measureis used, the resultsinvariablyshow thatsecureattachment the is norm, so we would do well to devote additionalresearch effort to investigating normative(i.e., secure) attachment phenomena. We imaginethatHarlowwould be pleasedto see that researchersare finally grapplingwith love and affection and are analyzingthese important facets of human behavior into their component variables. Such a bottom-upapproachand the constructionof a descriptive base constitutea useful first phase in the development of ourscience, butit is not the ultimategoal. Eventually, the accumulation of facts in the absence of theory becomes inefficient. Without theory as a guide, research is difficult to plan, and findings are difficult to Our interpret. goal herehas been to persuadethe reader that a theoretical integrationof researchfindings on close relationshipsis neitherpremature impossible nor and that attachmenttheory can provide the core constructs of such an integrativeframework.If we have failed to be persuasive,we hope we have at least been provocative enough to inspire the kind of integrative thoughtand conceptual debate that will help advance the science of close relationships. Notes of Preparation this targetarticlewas facilitatedby a Special Projects Grant from Cornell University to 18

Cindy Hazan and by National Science Foundation GrantBSN-8808736 to Cindy Hazan and Phillip R. Shaver. We thank Rick Canfield, RobertTurgeon, and Joanne Sturgeonfor their helpful comments on draftsof this targetarticle. Cindy Hazan, Departmentof HumanDevelopment andFamilyStudies,Van RensselaerHall, CornellUniversity, Ithaca,NY 14853-4401. References
Ainsworth,M. D. S., Blehar,M. C., Waters,E., & Wall, S. (1978). Patterns of attachment:A psychological study of the Strange Situation.Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. Altman,I., & Taylor,D. A. (1973). Socialpenetration:Thedevelopmentof interpersonalrelationships.New York:Holt, Rinehart & Winston. Arend,R., Gove, F., & Sroufe,L.A. (1979). Continuityof individual from infancy to kindergarten: predictivestudy of A adaptation ego resiliencyandcuriosityin preschoolers.ChildDevelopment, 50, 950-959. Aron, A. P., Dutton, D. G., Aron, E. N., & Iverson, A. (1989). Experiencesof falling in love. Journal of Social and Personal Relationships,6, 243-257. Aronson,E. (1988). Thesocial animal(5th ed.). New York:Freeman. Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective.Journal of Social and Personal Relationships, 7, 147-178. Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults:A test of a four-categorymodel. Journal of Personalityand Social Psychology, 61, 226-244. Baruch,G., Barnett, & Rivers,C. (1983).Lifeprints:Newpatterns R., of love and workfortoday's women.New York:McGraw-Hill. Berscheid, E. (1983). Emotion. In H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E. McClintock, L. A. Peplau, & D. R. Peterson (Eds.), Close relationships(pp. 110-168). New York:Freeman. attraction.In G. Lindzey & E. Berscheid, E. (1984). Interpersonal Aronson (Eds.), Handbookof social psychology (3rd ed., pp. 413-484). Reading,MA: Addison-Wesley. Berscheid, E., Dion, K., Walster, E., & Walster, G. W. (1971). Physical attractivenessand datingchoice: A test of the matching hypothesis. Journal of ExperimentalSocial Psychology, 7, 173-189. Berscheid,E., & Graziano,W. (1979). The initiationof social relaIn tionshipsandsocial attraction. R. L. Burgess & T. L. Huston (Eds.),Social exchangein developingrelationships.New York: Academic. Berscheid,E., & Walster,E. (1974). A little bit about love. In T. L. Huston(Ed.),Foundationsof interpersonal attraction(pp. 355381). New York:Academic. Bloom, B. L., Asher,S. J., & White,S. W. (1978). Maritaldisruption as a stressor:A review andanalysis.Psychological Bulletin,85, 867-894. and Bowlby, J. (1944). Forty-four juvenile thieves: Theircharacters home life. InternationalJournal Psycho-Analysis,25, 19-52, of 107-127. Bowlby, J. (1951). Maternalcare and mentalhealth.Geneva:World HealthOrganization. and Bowlby, J. (1973). Attachment loss: Vol. 2. Separation:Anxiety and anger. New York:Basic. Bowlby, J. (1979). The makingand breakingof affectional bonds. London:Tavistock.

ATTACHMENTAND CLOSERELATIONSHIPS Bowlby, J. (1980). Attachmentand loss: VoL3. Loss: Sadness and depression.New York: Basic. Bowlby, J. (1982).Attachmentand loss: Vol. 1.Attachment(2nd ed.). New York:Basic. (Originalwork published1969) Bowlby, J. (1988). A secure base: Parent-child attachment and healthyhumandevelopment.New York: Basic. Brennan,K., Shaver,P. R., &Tobey, A. E. (1991). Attachment styles, gender, and parentalproblem drinking.Journal of Social and Personal Relationships,8, 451-466. Bretherton,I. (1990). Open communicationand internal working models: Their role in the development of attachmentrelationon ships. In R. A. Thompson(Ed.), NebraskaSymposium Motivation: VoL 36. Socioemotional development (pp. 57-113). Lincoln:Universityof NebraskaPress. Buss, D. M. (1985). Humanmate selection. American Scientist, 73, 47-54. Buss, D. M., & Barnes, M. (1986). Preferences in human mate selection. Journal of Personality and Social Psychology, 50, 559-570. Campos, J. J., Barrett, K., Lamb, M. E., Goldsmith, H. H., & Stenberg, C. (1983). Socioemotional development. In P. H. Mussen (Ed.), Handbookof child psychology: Vol. 2. Infancy and developmentalpsychobiology (pp. 783-915). New York: Wiley. and Cassidy,J. (1988). Child-motherattachment the self in six-yearolds. Child Development,59, 121-134. Colin, V. L. (1991). Humanattachment:Whatwe knownow. Chevy Chase, MD: U.S. Departmentof Healthand HumanServices. Collins, N. L., & Read, S. J. (1990). Adult attachment,working models, and relationshipquality in dating couples. Journal of Personality and Social Psychology, 58, 644-663. Crittenden,P. M. (1988). Relationshipsat risk. In J. Belsky & T. Nezworski(Eds.), Clinical implicationsof attachment (pp. 136174). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. S. motherresponsiveness, Crockenberg, B. (1981). Infantirritability, and social supportinfluences on the security of infant-mother attachment.Child Development,52, 857-869. Daniels, T., & Shaver, P. R. (1991). Attachmentstyles and power strategies in romantic relationships. Unpublished manuscript, State University of New York at Buffalo, Department of Psychology. Dozier, M., & Kobak, R. R. (in press). Psychophysiologyand adolescentattachment interviews:Convergingevidenceforrepressing strategies.ChildDevelopment. Driscoll, R., Davis, K. W., & Lipetz, M. E. (1972). Parentalinterference and romantic love. Journal of Personality and Social Psychology, 24, 1-10. Duck, S. W. (1991, May). New lamps for old: A new theory of relationships and a fresh look at some old research. Paper Network on Perpresentedat the meeting of the International sonal Relationships,Normal,IL. Dutton, D. G., & Aron, A. P. (1974). Some evidence for heightened sexual attractionunder conditions of high anxiety.Journal of Personality and Social Psychology, 30, 510-517. Egeland, B., & Farber, E. A. (1984). Infant-motherattachment: Factorsrelatedto its developmentand changes over time. Child Development,55, 753-771. Feeney, J., & Noller, P. (1990). Attachmentstyle as a predictorof adultromanticrelationships. Journal of Personalityand Social Psychology, 58, 281-291. Fehr, B. (1988). Prototype analysis of the concepts of love and commitment.Journal ofPersonality andSocialPsychology, 55, 557-579. Felmlee, D., Sprecher,S., & Bassin, E. (1990). The dissolution of intimate relationships: A hazard model. Social Psychology Quarterly,52, 13-30. Festinger,L., Schachter,S., & Back, K. W. (1950). Socialpressures in informalgroups:A study of humanfactors in housing. New York:Harper. Fiala, K. B. (1991). Attachmentand psychosocial functioning of depressed, remitteddepressed, and nondepressedwomen and theirpartners.Unpublisheddoctoraldissertation,Universityof Amherst. Massachusetts, Fincham,F. D., Beach,S., & Nelson, G. (1987). Attribution processes in distressed non-distressed and attribution couples:Self-partner forspousebehavior. and CognitiveTherapy Research,11, 71-86. Fisher,H. E. (1992). Anatomyof love. New York:Norton. Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: RandomHouse. Folkes, V. S., & Sears, D. 0. (1977). Does everybody like a liker? Journal of Experimental Social Psychology, 13, 505-519. Freedman,J. (1978). Happypeople: Whathappiness is, who has it, and why. New York:HarcourtBraceJovanovich. Fullard,W., & Reiling, A. M. (1976). An investigationof Lorenz's Child Development,47, 1191-1193. "babyness." Furman,W., & Buhmester,D. (1985). Children'sperceptionsof the personalrelationshipsin their social networks.Developmental Psychology,21, 1016-1024. Goldsmith, H. H., & Alansky, J. A. (1987). Maternaland infant temperamental predictorsof attachment:A meta-analyticreview. Journalof Consultingand Clinical Psychology, 55, 805816. Goodwin,J. S., Hurt,W. C., Key, C. R., & Sarret,J. M. (1987). The effect of marital statuson stage, treatment survivalof cancer and patients.Journal of the American Medical Association, 258, 3125-3130. Gorer, G. (1973). Death, grief and mourning in Britain. In E. J. Anthony & C. Koupernik(Eds.), The child in his family: The impactof disease and death. New York:Wiley. Gottman,J. M. (1983). How childrenbecome friends.Monographs of the Societyfor Research in Child Development,48(3, Serial No. 201). of Gottman, Markman, & Notarius,C. (1977). The topography J., H., maritalconflict: A sequentialanalysis of verbal and nonverbal behavior.Journal of Marriageand the Family, 39, 461-477. Gottman,J., Notarius, C., Gonso, J., & Markman,H. J. (1976). A couple's guide to communication.Champaign,IL: Research Press. Greenberg,M., Cicchetti, D., & Cummings,M. (Eds.). (1990). Attachmentin the preschool years: Theory,research, and intervention.Chicago:Universityof Chicago Press. Grossberg,S. (1980). How does a brain build a cognitive code? PsychologicalReview,87, 1-51. Grossmann,K. E., & Grossmann,K. (1991). Attachmentquality as an organizerof emotionaland behavioralresponsesin a longitudinalperspective.In C. M. Parkes,J. Stevenson-Hinde,& P. Marris(Eds.), Attachmentacross the life cycle (pp. 93-114). London:Tavistock/Routledge. Hall, J. A. (1978). Gendereffects in decoding nonverbalcues. Psychological Bulletin,85, 845-857. Harlow,H. (1958). The natureof love. AmericanPsychologist, 13, 673-685. Hartup,W. (1983). Peer relations.In E. M. Hetherington(Vol. Ed.) & P. H. Mussen (Series Ed.), Handbookof child psychology: VoL4. Socialization,personality, and social development(pp. 103-196). New York:Wiley. Harvey, J. H., Orbuch, T. L., & Weber, A. L. (1990). A social In psychological model of account-making: response to severe stress.Journalof LanguageandSocial Psychology,9, 191-207. Hatfield,E. (1984). The dangersof intimacy.In V. J. Derlega (Ed.), Communication,intimacy and close relationships (pp. 207220). New York:Academic. Hatfield,E., & Rapson,R. L. (1987). Passionatelove: New directions in research.In W. H. Jones & D. Perlman(Eds.), Advances in personal relationships(Vol. 1, pp. 109-140). Greenwich,CT: JAI.

19

HAZAN & SHAVER Hazan, C., & Hutt, M. J. (1991a). Continuityand change in internal working models of attachment.Unpublishedmanuscript,Cornell University, Departmentof HumanDevelopmentand Family Studies. Hazan,C., & Hutt,M. J. (1991b). Fromparentsto peers: Transitions in attachment. Unpublished manuscript,Cornell University, Departmentof HumanDevelopmentand Family Studies. Hazan,C., Hutt,M. J., Sturgeon,J., & Bricker,T. (1991, April).The process of relinquishingparents as attachmentfigures. Paper presented at the meeting of the Society for Researchin Child Development, Seattle. Hazan,C., & Shaver,P. R. (1987). Romanticlove conceptualizedas an attachmentprocess. Journal of Personality and Social Psychology, 52, 511-524. Hazan, C., & Shaver, P. R. (1990). Love and work: An attachmenttheoreticalperspective.Journal of Personality and Social Psychology, 59, 270-280. Hazan, C., & Shaver, P. R. (1992). Broken attachments.In T. L. Orbuch(Ed.), Close relationshiploss: Theoreticalapproaches Associates, Inc. (pp.90-108). Hillsdale,NJ:LawrenceErlbaum Heinicke,C., & Westheimer,I. (1966). Briefseparations.New York: International UniversitiesPress. Hendrick,C., & Hendrick,S. (1986). A theory and methodof love. Journal of Personality and Social Psychology, 50, 392-402. Hill, C. T., Rubin, Z., & Peplau, L. A. (1976). Breakupsbefore marriage:The end of 103 affairs.Journal of Social Issues, 32, 147-168. Hinde, R. (1979). Towards understandingrelationships. London: Academic. Hinsz, V. B. (1989). Facial resemblance in engaged and married couples. Journal of Social and Personal Relationships, 6, 223-229. Holmes, J. G., & Rempel, J. K. (1989). Trustin close relationships. In C. Hendrick(Ed.), Close relationships(pp. 187-220). NewburyPark,CA: Sage. Huston,T. L., Surra,C. A., Fitzgerald,N. M., & Cate, R. M. (1981). Fromcourtshipto marriage: Mate selection as an interpersonal process. In S. Duck & R. Gilmour (Eds.), Personal relationships: VoL2. Developing personal relationships (pp. 53-88). New York:Academic. Kelley, H. H. (1983). Love and commitment. In H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger,E. McClintock,L. A. Peplau,& D. R. Peterson(Eds.), Close relationships(pp. 265-314). New York:Freeman. Kelley, H. H., Berscheid,E., Christensen,A., Harvey,J. H., Huston, T. L., Levinger, G., McClintock,E., Peplau,L. A., & Peterson, D. R. (Eds.). (1983). Close relationships.New York:Freeman. Kerckhoff,A. C., & Davis, K. E. (1962). Value consensus and need complementarityin mate selection. AmericanSociological Review, 27, 295-303. Kiecolt-Glaser,J. K., Garner,W., Speicher, C., Penn, G. M., Holliday, J., & Glaser, R. (1984). Psychological modifiers of immunocompetence in medical students. Psychosomatic Medicine, 46, 7-14. Kiecolt-Glaser,J. K., Ricker, D., George, J., Messick, G., Speicher, G. E., Garner,W., & Glaser,R. (1984). Urinarycortisol levels, cellularimmunocompetence, loneliness in psychiatricinpaand tients. PsychosomaticMedicine, 46, 15-23. Kirkpatrick,L. A., & Davis, K. E. (in press). Attachment style, gender,andrelationshipstability:A longitudinalanalysis.Journal of Personality and Social Psychology. Kirkpatrick,L., & Shaver, P. R. (1990). Attachment theory and religion: Childhoodattachments,religious beliefs, and conversion. Journalfor the Scientific Studyof Religion, 29, 315-334. Kobak,R. R., & Hazan,C. (1991). Attachmentin marriage:Effects of securityandaccuracyof workingmodels.Journalof Personality and Social Psychology, 60, 861-869. to Kobak,R. R., & Sceery, A. (1988). The transition college: Working affect regulation,andperceptionsof self models of attachment, and others.ChildDevelopment,88, 135-146. Konner,M. (1982). The tangled wing: Biological constraintson the humanspirit. New York:Holt, Rinehart& Winston. Kotler,T. (1985). Security and autonomywithin marriage.Human Relations,38, 299-321. Kunce, L. J., & Shaver, P. R. (1991). An attachment-theoretical approachto caregivingin romanticrelationships.Unpublished manuscript,State Universityof New York at Buffalo, Department of Psychology. Lamb,M. E., Thompson,R. A., Gardner,W. P., Charnov,E. L., & as Estes, D. (1984). Securityof infantileattachment assessed in the Strange Situation:Its study and biological interpretation. Behavioraland BrainSciences, 7, 127-171. Lee, M. D., & Kanungo,R. N. (Eds.). (1984). Managementof work and personal life: Problems and opportunities. New York: Praeger. Levenson, R. W., & Gottman,J. M. (1985). Six physiological and affective predictors change in relationshipsatisfaction.Jourof nal of Personalityand Social Psychology, 49, 85-94. Levinger, G. (1966). Systematic distortion in spouses' reports of and preferred actualsexual behavior.Sociometry,29, 291-299. Levinger, G. (1976). A social psychological perspectiveon marital dissolution.Journal of Social Issues, 32(1), 21-47. Levinger,G. (1983). Developmentand change. In H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger,E. McClintock,L A. Peplau,& D. R. Peterson(Eds.), Close relationships(pp. 315-359). New York:Freeman. Levinger,G., & Levinger,A. C. (1986). The temporalcourseof close relationships: Some thoughts about the development of children's ties. In W. W. Hartup& Z. Rubin (Eds.), Relationships and development(pp. 111-134). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Levinger,G., & Snoek,J. D. (1972).Attractionin relationship:Anew look at interpersonal attraction. Morristown, NJ: General LearningPress. Lewis, R. A. (1973). A longitudinaltest of a developmentalframework for premarital Journal of Marriage and dyadic formation. the Family, 35, 16-25. Lewis, R. A., & Spanier,G. B. (1979). Theorizing about the quality and stability of marriage.In W. R. Burr,R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary theories about the family (Vol. 1). New York: Free Press. Lund, M. (1985). The developmentof investmentand commitment scales for predictingcontinuityof personalrelationships. Journal of Social and Personal Relationships,2, 3-23. Lynch,J. J. (1977). The brokenheart: Themedical consequencesof loneliness.New York:Basic. Main, M. (1990). Parental aversion to infant-initiatedcontact is with theparent'sown rejectionduringchildhood:The correlated effects of experience on signals of security with respect to attachment. K. E. Barnard T. B. Brazelton(Eds.), Touch: In & The foundation of experience (pp. 461-495). Madison, CT: International UniversitiesPress. Main,M. (1991). Metacognitiveknowledge,metacognitivemonitoring, and singular(coherent)vs. multiple(incoherent)model of attachment: Findingsanddirectionsfor futureresearch.In C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment across the life cycle (pp. 127-159). London: Tavistock/ Routledge. Main,M., & Hesse, E. (1990). Parents'unresolvedtraumatic experiences are related to infant disorganized status: Is frightened and/orfrightening parentalbehaviorthe linking mechanism?In M. T. Greenberg,D. Cicchetti, & E. M. Cummings (Eds.), Attachmentin the preschool years (pp. 161-184). Chicago: Universityof ChicagoPress. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, A childhood,andadulthood: move to the level of representation.

20

ATACHMENT AND CLOSERELATIONSHIPS Monographsof theSociety forResearch in ChildDevelopment, 50(1-2, Serial No. 209), 66-104. Main, M., & Solomon, J. (1990). Proceduresfor identifyinginfants as disoganized/disoriented duringthe AinsworthStrangeSituation. In M. T. Greenberg,D. Cicchetti, & E. M. OCmmings (Eds.),Attachmentin the preschool years (pp. 121-160). Chicago: Universityof Chicago Press. Marris,P. (1958). Widowsand theirfamilies. London:Routledge & Kegan Paul. Mikulincer, M., Florian, V., & Tolmacz, R. (1990). Attachment styles and fear of personal death: A case study of affect regulation.Journal of Personality and Social Psychology, 58, 273-280. Mikulincer, M., & Nachshon, 0. (1991). Attachment styles and patternsof self-disclosure. Journal of Personality and Social Psychology, 61, 321-331. Miller, G. A., Galanter,E., & Pribram,K. H. (1960). Plans and the structureof behavior. New York:Holt, Rinehart& Winston. Miller, S. I., & Schoenfeld, L. (1973). Grief in the Navajo: Psychodynamics and culture.InternationalJournal of Psychiatry,19, 187-191. Mizukami, K., Kobayashi, N., Ishii, T., & Iwata, H. (1990). First selective attachment begins in early infancy: A study using telethermography. Infant Behavior and Development, 13, 257-271. Murstein,B. I. (1976). Whowill marrywhom?New York:Springer. Newcomb, T. M. (1961). Theacquaintanceprocess.New York:Holt, Rinehart& Winston. M. Owen, M. T., Easterbrooks, A., Chase-Lansdale, & Goldberg, L., W. A. (1984). The relationbetweenmaternal employmentstatus and the stability of attachmentsto mother and father. Child Development,55, 1894-1901. Palgi, P. (1973). The socio-culturalexpressions and implicationsof death, mourningand bereavementarising out of the war situation in Israel.IsraelAnnals of Psychiatry,11, 301-329. Parkes, C. M., & Weiss, R. S. (1983). Recoveryfrom bereavement. New York:Basic. Peplau, L. A., & Gordon,S. L. (1983). The intimaterelationshipsof lesbians and gay men. In E. R. Allgeier & N. B. McCormick (Eds.), The changing boundaries: Gender roles and sexual behavior(pp. 226-244). Palo Alto, CA: Mayfield. Peplau, L. A., & Perlman, D. (Eds.). (1982). Loneliness: A sourcebookof currenttheory,researchand therapy.New York: Wiley-Interscience. Piaget, J. (1952). The origins of intelligence in children. New York: International UniversitiesPress. Rands, M., Levinger, G., & Mellinger, G. D. (1981). Patternsof conflict resolution and maritalsatisfaction.Journal of Family Issues, 2, 297-321. Reedy, M. N., Birren,J. E., & Schaie, K. W. (1981). Age and sex differences in satisfying love relationshipsacross the adultlife span. HumanDevelopment,24, 52-66. Reis, H. T., & Shaver, P. R. (1988). Intimacy as an interpersonal process. In S. Duck (Ed.), Handbookof research in personal relationships(pp. 367-389). London:Wiley. Reiss, I. L. (1960). Toward a sociology of the heterosexual love relationship.Marriage and Family Living, 22, 139-145. Rubenstein,C., & Shaver, P. R. (1982). In search of intimacy.New York:Delacorte. Rubin, Z. (1973). Liking and loving. New York: Holt, Rinehart& Winston. HarvardUniRubin, Z. (1980). Children'sfriendships. Cambridge: versity Press. Rubin, Z. (1984). Toward a science of relationships.Contemporary Psychology, 29, 856-858. Rusbult, C. E. (1980). Commitment and satisfaction in romantic associations:A test of the investmentmodel. Journal of ExperimentalSocial Psychology, 16, 172-186. Rusbult,C. E. (1983). A longitudinaltest of the investmentmodel: The development (and deterioration)of satisfaction and commitmentin heterosexualinvolvements.Journal of Personality and Social Psychology, 45, 101-117. Rusbult,C. E., & Zembrodt,I. M. (1983). Responses to dissatisfaction in romanticinvolvement:A multidimensional scaling analSocial Psychology, 19, 274-293. ysis. Journal of Experimental I. Rusbult,C. E., Zembrodt, M., & Gunn, L. K. (1982). Exit, voice, loyalty, and neglect: Responses to dissatisfactionin romantic involvements.Journal of Personality and Social Psychology, 43, 1230-1242. Schachter,S. (1959). The psychology of affiliation: Experimental studiesof thesources of gregariousness.Stanford,CA: Stanford UniversityPress. Schaffer,H. R., & Emerson,P. E. (1964). The developmentof social in attachments infancy.Monographsof theSocietyforResearch in ChildDevelopment,29(3, Serial No. 94). Shaver,P. R., & Hazan, C. (1984). Incompatibility,loneliness, and "limerence."In W. Ickes (Ed.), Compatibleand incompatible relationships(pp. 163-184). New York:Springer-Verlag. Shaver, P. R., & Hazan, C. (1993). Adult romantic attachment: Theory and evidence. In D. Perlman & W. Jones (Eds.), Advances in personal relationships (Vol. 4, pp. 29-70). Greenwich, CT: JAI. D. Shaver,P. R., Hazan,C., & Bradshaw, (1988). Love as attachment: The integrationof threebehavioralsystems. In R. J. Sternberg & M. L. Barnes(Eds.), Thepsychology of love (pp. 68-99). New Haven, CT: Yale UniversityPress. Shaver,P. R., & Rubenstein,C. (1980). Childhoodattachment experience and adult loneliness. In L. Wheeler (Ed.), Review of personalityand socialpsychology (Vol. 1, pp. 42-73). Beverly Hills, CA: Sage. Simpson,J. A. (1990). Theinfluenceof attachment styles on romantic Journalof Personalityand SocialPsychology, 59, relationships. 971-980. Skolnick,A. (1978). Theintimateenvironment: Exploringmarriage and thefamily (2nd ed.). Boston: Little, Brown. Snyder,M., & Swann, W. B., Jr.(1978). Behavioralconfirmationin social interaction: Fromsocial perceptionto social reality.Journal of Experimental Social Psychology, 14, 148-162. Spanier,G. B., Lewis, R. A., & Cole, C. L. (1975). Maritaladjustment over the family life cycle: The issue of curvilinearity. Journal of Marriageand the Family, 37, 263-275. Sroufe, L. A. (1983). Infant-caregiverattachmentand patternsof in and adaptation preschool:The rootsof maladaptation competence. In M. Perlmutter (Ed.), MinnesotaSymposiumon Child Psychology (Vol. 16, pp. 41-83). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Sroufe,L. A., & Waters,E. (1977). Attachmentas an organizational construct.ChildDevelopment,48, 1184-1199. and the maintenance and enStanley, S. M. (1986). Commitment hancementof relationships.Unpublisheddoctoraldissertation, Universityof Denver. Steinberg,L., & Silverberg,S. B. (1986). The vicissitudes of autonomy in early adolescence. Child Development,57, 841-851. Sternberg,R. J. (1986). A triangulartheory of love. Psychological Review,93,119-135. Swann,W. B., Jr.,Hixon, J. G., & De La Ronde, C. (1992). EmbracNegative self-concepts and maritalcoming the bitter"truth": mitment.Psychological Science, 3, 118-121. Tannen,D. (1990). Youjust don't understand:Womenand men in conversation.New York:Morrow. in Taylor,D. A., & Altman,I. (1987). Communication interpersonal relationships:Social penetrationprocesses. In M. Roloff & G. Miller (Eds.), Interpersonal processes: New directions in communicationresearch (pp. 257-277). Newbury Park,CA: Sage. Tennov, D. (1979). Love and limerence: The experience of being in love. New York:Stein & Day.

21

HAZAN & SHAVER Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York:Wiley. J., Traupinann, & Hatfield,E. (1981). Love: Its effects on mentaland physical health.In J. March,S. Kiesler,R. Fogel, E. Hatfield,& E. Shana(Eds.),Aging: Stabilityand change in thefamily. New York:Academic. Vaillant, G. E. (1977). Adaptation to life: How the best and the brightestcame of age. Boston: Little, Brown. van den Boom, D. (1990). Preventiveinterventionand the qualityof mother-infant interaction and infant exploration in irritable infants.In W. Koops, H. J. G. Soppe, J. L. van der Linden,P. C. M. Molenaar, & J. J. F. Schroots (Eds.), Developmental psychology behind the dikes: An outline of developmentalpsychology research in The Netherlands. The Netherlands: Uitgeverij Eburon. Vaughn, B., Egeland, B., Sroufe, L. A., & Waters,E. (1979). Individual differences in infant-mother attachmentat 12 and 18 months: Stability and change in families under stress. Child Development,50, 971-975. R. Veroff, J., Douvan, E., & Kukila, A. (1981). TheinnerAmerican: A self-portraitfrom 1957 to 1976. New York:Basic. J. Wallerstein, B., & Kelly, J. B. (1980). Survivingthe breakup:How childrenandparents cope with divorce. New York: Basic. D., Walster,E., Aronson,E., Abrahams, & Rottmann,L. (1966). The in importanceof physical attractiveness datingbehavior.Journal of Personalityand Social Psychology, 4, 508-516. Walster,E., & Walster,G. W. (1978). A new look at love. Reading, MA: Addison-Wesley. Waters,E. (1978). The reliabilityand stability of individualdifferences in infant-mother attachment.Child Development, 49, 483-494. Weiss, R. S. (1973). Loneliness: The experience of emotional and MA: MIT Press. social isolation. Cambridge, Weiss, R. S. (1975). Maritalseparation.New York: Basic. Weiss, R. S. (1982). Attachmentin adults. In C. M. Parkes & J. Stevenson-Hinde(Eds.), The place of attachment in human behavior(pp. 171-184). New York:Basic. Weiss, R. S. (1988). Loss and recovery.Journal of Social Issues, 44, 37-52. Zajonc,R. B. (1968). Attitudinaleffects of mere exposure.Journal of Personalityand Social Psychology MonographSupplement, 9(2, Pt. 2), 1-27.

22