Você está na página 1de 13

This article was downloaded by: [McGill University Library]

On: 03 February 2015, At: 03:23


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Psychological Inquiry: An International Journal for the


Advancement of Psychological Theory
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hpli20

Deeper Into Attachment Theory


Cindy Hazan & Philip R. Shaver
Published online: 10 Dec 2009.

To cite this article: Cindy Hazan & Philip R. Shaver (1994) Deeper Into Attachment Theory, Psychological Inquiry: An
International Journal for the Advancement of Psychological Theory, 5:1, 68-79, DOI: 10.1207/s15327965pli0501_15

To link to this article: http://dx.doi.org/10.1207/s15327965pli0501_15

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Psychological Inquiry Copyright 1994 by
1994, Vol. 5, No. 1,68-79 Lawrence Erlbaurn Associates, Inc.

AUTHORS' RESPONSE

Deeper Into Attachment Theory


Cindy Hazan
Cornell University
Phillip R. Shaver
University of California, Davis
Downloaded by [McGill University Library] at 03:23 03 February 2015

One of the explicit goals of our target article was These questions are not only of theoretical interest;
to provoke the kind of integrative thinking and con- they have obvious implications for operationalization
ceptual debate that can help advance the science of and measurement and, therefore, are relevant to
relationships. Surely that goal was achieved! We commentators' concerns about our extensions and ap-
were challenged and inspired by what many of the plications of the theory. For example, what is the nature
commentators wrote in response to o w proposed and form of attachment behavior in adulthood, and how
theoretical framework. For this and for the time and should individual differences be conceptualized and
thought they put into their commentaries, we thank measured? Last, several questions posed by the com-
them. They highlighted the strengths and limitations mentators pertain specifically to the study of adult
of attachment theory and our extensions of it and relationships. Is an integrated theory of close relation-
forced consideration of some very complex issues ships desirable or possible? Does attachment theory
that must be settled if research on adult attachment provide an adequate framework for such an integra-
is to proceed in fruitful directions. Their insights and tion? We see these as the major questions, to which we
even their criticisms made it easier for us to see just respond in turn.
how to proceed.
Most of the criticisms fall into one of two major
Theoretical Issues
categories. On the one hand are concerns about the
validity and utility of attachment theory itself. On the
other hand are concerns about our application of the Are Attachments Singular or Multiple?
theory to adult relationship phenomena. An import-
ant caveat to be stated at the outset is that our article Attachment researchers agree that, given the oppor-
should not be seen as providing a complete explica- tunity, all normal human infants become attached to
tion of attachment theory. Bowlby required well their primary caregiver, typically within the first 8
more than a thousand pages for his initial presenta- months of life. Whether secondary attachments to other
tion of it, whereas the present format allowed only a people are formed simultaneously or only after a pri-
sketch of the key concepts. T o respond t o mary attachmenthas been established is open to debate,
commentators' concerns about the theory necessi- but there is no doubt that infants and children do form
tates a detailed discussion of some of its neglected multiple attachments. Bonds that satisfy the criteria for
features, all directly or indirectly tied to three ques- being attachments-that is, that include proximity
tions raised repeatedly in the commentaries: maintenance and safe-haven and secure-base behav-
i o r s a r e commonly developed with other adults as
1. If individuals form multiple attachments, and if well as with older siblings. But, are these relationships
these attachments vary in quality, then what does it of equal importance to the attached infant or child?
mean to refer to an individual as being of a particular Empirical evidence strongly suggests not. Although a
attachment type? plurality of attachmentfigures is the norm, these figures
2. Is attachment best conceptualized as a character- are not treatedequivalently. Infants show clear discrim-
istic of an individual or of a relationship between two ination and consistent preferences (Colin, 1985,1987;
individuals? Cummings, 1980; Farran & Ramey, 1977). If several
3. If individual differences exist, when do they be- caregivers are available, infants reliably seek and main-
come stabilized and just how stable are they? tain proximity to one, especially if tired or ill (Ains-
AUTHORS'RESPONSE

worth, 1967,1982). This is the case even in polymatric the primary attachment figure is either unavailable or
settings. Infants exhibit more intense protest on being unable to provide sufficient reassurance. As Crowell
left by the primary attachment figure as compared to and Waters put it, "any port will do in a storm."
others (Schaffer & Emerson, 1964). In unfamiliar set- We agree with Lewis that "adults have needs other
tings, they are more reassured by the presence of the than attachment and security." And, we might add,
primary figure than by others (Ricciuti, 1974; Shill, children have other needs too. Bowlby clearly stated
Solyom, & Biven, 1984). Bowlby (196911982)referred that there exist a multiplicity of needs for which multi-
to this tendency of a child to form a special attachment ple social relationships are required. His postulation of
to one figure as monotropy. distinct behavioral systems is an explicit acknowledg-
Multiple attachments are hypothesized to be hier- ment of this fact (e.g., he distinguishesattachmentfrom
archically arranged. At the top of this hierarchy is the affiliation). Lewis makes reference to "the idea that
primary attachment figure. For good or ill, this figure these needs are all to be satisfied by a single significant
is usually the infant's mother. Although we share other in our adult life." Neither we nor Bowlby claimed
Lewis's concerns regarding attachment researchers' that the primary attachmentfigure is expected to satisfy
focus on the mother, we must stress that there is all needs. In fact, the attachment figure may not even
nothing in attachment theory requiring that mother satisfy needs for security, despite being preferred over
Downloaded by [McGill University Library] at 03:23 03 February 2015

be the primary attachment figure. Infants form a other sources. Some needs are best satisfied through
primary attachment to the person who most consis- social relationships, and these are regulated by what H.
tently provides care and responds to their distress F. Harlow and M. K. Harlow (1965) called the affective
signals. Theoretically, this could be almost anyone systems. R. S . Weiss (1974) specified a total of six
in the infant's environment. In actuality, it tends to relational provisions and corresponding needs, not all
be the mother, but there are exceptions. In one inves- of which are likely to be handled by any single person
tigation, 24% of the studied infants directed more or or relationship. In all such formulations, including
stronger attachment behavior to their fathers than to Bowlby's, attachment is just one of many needs.
their mothers (Colin, 1987). In almost every case, The needs and the systems that regulate them, how-
these special fathers had spent substantial time with ever, may not be as independent as Lewis implies.
their infants and had participated actively in their Indeed, in infants, activation of the attachment system
routine care. What might be seen as a "dangerously overrides the activation of systems such as affilitation
narrow" focus on mothers, then, is more a result of and exploration (Ainsworth, Blehar, Waters, & Wall,
societal organization than of attachment theory. 1978; H. F. Harlow, 1961). This suggests that needs
Over the course of development, many changes may may be better conceptualizedas hierarchically ordered,
occur in the content and structure of an individual's rather than as columns in the kind of person x needs
attachment hierarchy. According to Bowlby, parental matrix proposed by Lewis. Some needs are hypothe-
figures tend to be permanent members of the hierarchy, sized to be naturally directed toward the same individ-
but their positions naturally change as a child matures. ual--as may be the case with security and sexual needs
Others are added to or dropped from the hierarchy. in adulthood. Regardless of their arrangement, there is
Eventually, with the formation of a pair bond in adult- no doubt that the needs and their potential satisfiers are
hood, a peer-usually a sexual partner-assumes the multiple. That attachment theory deals primarily with
position of primary attachment figure and ascends to attachment-relatedneeds might be judged a limitation.
the top of the hierarchy. In an interview study of more In our view, however, this focus makes it ideally suited
than 100 adults---nearly all with extensive social net- for the study of close relationships. The closest of close
works-80% of subjects who had been involved in a relationships are those between parents and children
romantic relationship for at least 2 years were primarily and between adult lovers. In both, the provision of
attached to their partners (Hazan, 1992). Most of the security and comfort are of central importance.
rest were still primarily attached to a parent. This is not
to say that their attachment needs were not distributed
across multiple relationship figures. In fact, for the Are Attachment Patterns Properties of
majority, they were. All could name other people who Individuals or of Relationships?
were important sources of comfort and support. How-
ever, the overwhelming majority reported clear prefer- According to Lewis, if we reject the notion of singu-
ences for a single other. That Bowlby, along with other larity (i.e., that one relationshipsatisfiesall needs), then
attachment theorists and researchers, frequently re- "attachment cannot be a trait." Duck and the Hendricks
ferred to the attachment figure should not be taken to express similar opinions. We do reject the notion of
mean the only attachment figure, but rather theprimary singularity but still argue that attachment patterns
one. Subsidiary attachment figures are sought only if (types or styles) can be and are trait-like. And, in
HAZAN & SHAVER

response to Kobak, we argue that attachment patterns, that avoidant adolescents were more likely than secure
in addition to characterizing individuals, are also rela- or ambivalent adolescents to report childhood separa-
tionship constructs. How can this be? Sroufe and Flee- tions from their mothers, but not their fathers. Although
son (1986) proposed a model of relationships that these results do not provide conclusive evidence, they
integrates these two aspects of attachment. Considerthe strongly suggest that the traitlike aspects of attachment
case in which an infant develops an avoidant attach- develop mainly out of relationships with primary at-
ment to its primary caregiver. Avoidant, in this case, tachment figures.
describes the quality of the relationship between the
two. However, some key features of the relationship
and associated experiences eventually come to be rep- When Do Attachment Types Become
resented in the mind of, and habitually woven into the Stable, and How Stable Are They?
behavior of, the infant. The mental representationis one
of an avoidant attachment relationship in which the One of the most common misconceptions about at-
infant's bids for closeness and comfort have been fairly tachment theory is that it predicts 100% stability from
consistently rebuffed by the caregiver. The representa- infancy to adulthood. In its most simplistic form, "the
tion, which resides in the infant, is then carriedforward way the infant attaches at 1year of age mostly deter-
Downloaded by [McGill University Library] at 03:23 03 February 2015

into new relationships in which it may influence per- mines the way the adult attaches at age 21" (Hendricks
ceptions, feelings, and behaviors. In other words, the & Hendricks). To our knowledge, Bowlby never made
avoidant attachment has become a characteristic of the such a claim; we have never made such a claim, and
individual infant, even though it began as a feature of any responsible researcher apprised of the evidence
a dyadic relationship. How subsequent relationships would not make such a claim either. Such claims serve
unfold-whether they too become avoidant-will be a as a convenient "straw person" for critics who presume
joint function of the biases of the child (perhaps to that the theory predicts perfect continuity of individual
expect rejection and to avoid contact) and the behavior differences and, because there is abundant evidence
of new relationship partners. Experiences in subse- that continuity is not perfect, conclude that the theory
quent relationships will either confirm the child's must be wrong. Bowlby (1973) explicitly stated, in a
model of relationships and thereby reinforce avoidant passage we have quoted frequently in our writings, that
expectations and tendencies or disconfirm the preexist- working models of attachment are gradually con-
ing model and perhaps cause a revision of the model. structed out of experiences throughout infancy, child-
The resultant reinforced or revised model will then be hood, and adolescence. Only then do they become
carried into subsequent relationships, in which it may relatively resistant to, but still not impervious to,
be further elaborated, reinforce4 or modified. At any change. Our view is that they are sufficiently stable to
point in the individual's development, it would be warrant consideration and study.
possible to classify the quality of a current attachment Stabilityof attachmentclassificationsduring infancy
relationship and also the individual's representational depends, not surprisingly,on the stability of the infant's
biases. Bowlby (1988) presented attachment theory as environment. In stable environments, rates of stability
a theory of personality development within close rela- over a 6-month period have ranged from 81% (Connell,
tionships. To ask whether attachment is a trait or a 1976) to 96% (Waters, 1978). In unstable environ-
relationship construct is to imply a false disjunction. ments, stability rates tend to be lower, as would be
Of the many attachments that an individual forms, expected if they depend largely on the quality of inter-
which are the most influential in the construction of a actions with caregivers (and not, incidentally, if they
core internal working model of relationship? Research depend primarily on innate temperament). Even in an
indicates that the relationship with the primary attach- unstable environment, however, Egeland and Farber
ment figure has the greatest and most lasting impact on (1984) found 60% stability in attachment classifica-
later development and functioning. In a study of attach- tions. In one longitudinal study, subjects were classi-
ment among 6-year-olds, for example, Main, Kaplan, fied in infancy and then reassessed at age 6; high
and Cassidy (1985) found that the attachment relatian- stability rates were found (Main et al,, 1985). The study
ship established by a child during infancy with his or was replicated in Germany, with similar results (Wart-
her mother (who was assumed to be the primary attach- ner, 1986). In another longitudinal investigation, sig-
ment figure) influenced the child's representational nificant (although far from perfect) continuity was
models of self and others more than attachment to the found over the 10-year period following classifications
father did. Consistent with these findings, Bowlby at 12 months (Elicker, Englund, & Sroufe, 1992).
(1980) reported that loss of the mother before age 11 All these studies were conducted with infant or child
predicts much greater vulnerability to depression than subjects. If Bowlby was correct in his view that models
does loss of the father. Feeney and Noller (1990) found do not become stable until adolescence, studies exam-
AUTHORS'

ining stability from adolescence through adulthood uals select and sometimes create their own environ-
might provide more accurate estimates. More work is ments. Kirkpatrick and Davis (in press) suggested that
clearly needed before any firm conclusions about sta- partners are either chosen or retained for their tendency
bility can be drawn, and such work is underway. The to confirm attachment models. The data are equally
Hendricks say that "it would ...be asking too much to compatiblewith the interpretation that model-confirm-
require the kind of longitudinal studies necessary to ing styles emerged within the context of the relation-
assess attachment over the developmental span from ships, rather than operating as factors in partner
infancy to adulthood." Actually, Alan Sroufe and his selection. However, Caspi and Herbener's (1990) re-
colleagues have already followed a group of subjects sults indicate that assortative mating of spouses on
from infancy to adolescence, and Stuart Hauser and his personality characteristics promotes personality conti-
collaborators have followed a group of adolescents into nuity and that the continuity is not attributable to pre-
early adulthood Thus, the kind of longitudinal data existing stability tendencies. Swann, Hixon, and De La
needed to answer questions about stability and change Ronde's (1992) study, cited in our target article, pro-
across the life course will soon be available. For now, vides further evidence that partners who confirm self-
we can say that there is evidence for both continuity and models, even if negative, are preferred over partners
change, and researchers have begun to identify the whose feedback does not fulfill expectations.
Downloaded by [McGill University Library] at 03:23 03 February 2015

conditions under which each outcome is most likely. Given Bowlby's original formulation of attachment
Estimating stability of attachment styles in adult- theory and the empirical evidence that has accrued in
hood is rendered more difficult by a lack of consensus its support, it wouldbe inaccurate to say that singularity
about how such styles should be measured. (More on of attachment relationships is either predicted or ex-
this later.) But it is possible to speculate about the pected. Humans normally became attached to multiple
underlying mechanisms that might be expected to pro- individuals and even to inanimate objects. However,
mote stability. In an elegant and detailed account of these attachmentsare equivalent neither in their import-
personality across the life course, Caspi and Bem ance nor in their effects on developmental outcomes.
(1990) operated on the assumption that both continuity We accept that Lewis is attached to his pipe, but we
and change in traits occur as a result of person-envi- suspect that, in his hierarchy of attachments, the pipe
ronment interactions. Caspi and Bem identified three ranks somewhere below his wife and children. When
interaction types that appear to be particularly import- attachment researchers speak of an individual's attach-
ant. Reactive interaction refers to the well-documented ment type, they refer to a mental representation--corn-
finding that mental representations of the self and oth- plete with experience-based beliefs, expectations,
ers determine in large part what people select to attend emotions, and action tendencies-that is probably in-
to and how they interpret it. Adults seem to be particu- fluenced, to some degree, by all important relationships
larly attentive to information that confirms their expec- the individual has been involved in, but especially by
tations about the world and matches their views of attachment relationships that can be called primary.
themselves (Markus, 1977). Individuals with avoidant Our views of the social world are no doubt multiply
working models who, by definition, believe that others determined, but the experiences we have with the per-
are unreliable providers of support should be particu- son on whom we most depend for comfort and security
larly attentive to information indicative of a lack of will form the foundation of our model of the world as
trustworthiness. When such information is ambiguous, a place in which comfort and security can be reliably
as social information tends to be, an avoidant bias could counted on or not. To the degree that this model is
easily lead to a conclusion that is expectation confirm- repeatedly confirmed through theprocess of interaction
ing. In an anxiety-provoking experimental setting, with the environment, it should be stable. Disconfirm-
avoidant females were less likely than secure or ambiv- ing experiences, however, can occur and can produce
alent females to seek emotional support from their change (Hazan & Hutt, 1991). Behavior in all domains,
partners (Simpson, Rholes, & Nelligan, 1992). Evoca- including close relationships, is assumed to be guided
tive interaction refers to the finding that individuals by dispositional factors (e.g., attachment style) and
evoke different responses from their environments. In contextual factors (e.g., the actual behavior of the part-
the Simpson et al. (1992) experiment, avoidant subjects ner). As with the singularity of attachments, perfect
did not communicate their anxiety or seek support. As stability is neither predicted nor expected. It is an unfair
a consequence, they evoked less supportive behavior characterization of attachment theory to say that it
from their partners, which could then be interpreted as assumes either, On the basis of empirical evidence and
confirming expectations of nonsupport. logical argument, it would be all too easy to dismiss any
As children mature and social networks are extended theory that did.
beyond the family to the peer group, there is ample This brings us to two additional and closely related
opportunity forproactive interactions in which individ- questions: Is primary caregiver responsiveness the
IIAZAN& SHAVER

most influential determinant of mental models of at- (1989) found strong evidence for the importance of
tachment? If so, how responsive does a caregiver need contingent responding. What matters most in the long
to be in order for an infant to construct a secure model? term is not only how a caregiver responds, but under
The Hendricks "wonder whether any parent approaches what circumstances. Most mothers interact frequently
100% consistency in either response or nonresponse." and positively with their infants, but what seems to
Duck reminds us of how "extremely wearing" infants determine infants' internal working models is whether
can be and goes on to ask on whose perceptions- and how the caregiver responds to distress.
parents' or infant's--an attachment model is based. On Does anyone who has ever come into contact with
the infant's, of course. The mental representation that human infants really believe that the way they are
an individual constructs is necessarily based on that treated by the people on whom they are utterly depen-
individual's perceptions of his or her own experiences. dent for survival has no implications for their subse-
The perceptions may be similar to those of the social quent development or view of the world?
partner or depart dramatically from them. Bowlby
(1973) argued that the models would be "tolerably
acccurate reflections of the experiences those individ- Parenting or Temperament?
uals have actually had" @. 235), and so far there is no
Downloaded by [McGill University Library] at 03:23 03 February 2015

reason to doubt this. How important is caregiver responsiveness com-


The question of how responsive a caregiver needs pared to other factors, such as temperament? The Hen-
to be for his or her infant to learn to expect respon- dricks state that "temperament of the infant does make
siveness is, of course, an empirical question. Never- some difference," implying that this fact somehow
theless, for the very reasons Duck and the Hendricks negates the caregiver-responsiveness hypothesis. Of
suggest, it is unlikely that 100% responsiveness is course temperament matters; it is likely to have effects
required. If perfect consistency were necessary, we that are both direct and indirect. Anything that influ-
doubt that the majority of infants would develop ences the quality of the caregiver's response, as infant
secure attachments to their caregivers, as scores of temperament surely does, can be expected to influence
studies, beginning with Ainsworth's, indicate they the quality of the attachment between them. Despite
do. We also suspect that the absolute frequency of good evidence for differences among newborns on
response is less important that the context in which temperament dimensions such as irritability, sociabil-
the response occurs or fails to occur. ity, and threshold for fear, security of attachment can-
With full recognition of the limitations of analo- not (yet) be reliably predicted from such differences. In
gies, we offer one here. Imagine you are the owner contrast, it can be predicted from caregiver behavior.
of a car that, on most days, starts when you turn the Belsky and Rovine (1987) proposed a model of how
key and delivers you to your destination without temperament influences attachment classification.
incident. Imagine that this same car, on occasion, has Basedon their own and others' data, Belsky and Rovine
broken down. Both the frequency of breakdown and concluded that, although temperament is somewhat
the circumstances under which breakdown occurs predictive of the pattern of security or insecurity (i.e.,
can be expected to influence your evaluation of the the secure or insecure subtype) exhibited by an infant,
car as reliable or unreliable. If, for example, it breaks it is not predictive of whether the attachment will be
down only once-but while you are racing to the secure or insecure. The Fox, Kimmerly, and Schafer
airport on your way to deliver an invited address- (1991) meta-analysis cited by the Hendricks also
you might decide that the car is unreliable. However, showed that security is not predictable from tempera-
if it breaks down more than once-but never under ment variables. In a dramatic demonstration of the
pressing circumstances-you might be better able to contribution of caregiver behavior to attachment clas-
maintain the view that it is reliable. If it breaks down sification when temperament is controlled, van den
often or most of the time, you would consider it Boom (1990) identified 100 babies who had been la-
generally unreliable. Although acknowledging that beled temperamentally difficultand randomly assigned
models of our attachment figures differ in important them to an experimental or control group. When the
ways from models of our cars, both are cognitive infants were 6 to 9 months old, mothers in the experi-
representations that develop from concrete experi- mental group received three individualized training
ences. Models of attachment figures as responsive, sessions in sensitive responding. At 12months, 68%of
unresponsive, or unpredictable are assumed to be the infants whose mothers had been trained to be re-
based on the frequency of contingent tesponding as sponsive were securely attached, compared with only
well as on the context in which a response is desired. 28% of those in the control group.
In a very careful and detailed analysis of mother-in- We think that infants would be better served if re-
fant interactions, Isabella, Belsky, and von Eye searchers, rather than resisting theory and evidence that
AUTHORS' RESPONSE

underscore the importance of caregiver behavior for Bartholomew, Noller & Feeney) or mention measure-
social and personality development, would focus on the ment as a concern (e.g., Crowell & Waters, Duck,
factors that facilitateand foster optimal care. More than Stevenson-Hinde, R. S. Weiss). How should individual
one such factor has already been identified-for exam- differences be conceptualized? How should they be
ple, sensitivity training (van den Boom, 1990), close measured? How do the various existing measures relate
bodily contact (Anisfeld, Casper, Nozyce, & Cuming- to one another? These important questions must be
ham, 1990; Main, 1990), and social support for car- satisfactorily answered if the field of adult attachment
egivers (Crockenberg, 1981). research is to make further contributions to our under-
The Hendricks outline seven assumptions one is standing of close relationships.
"asked to accept" before accepting attachment theory. For goodreasons,social scientists are rarely satisfied
They summarize these assumptions by saying that with the quality of their measures, and we are not
"there is continuity of attachment patterns and that exceptions. When we composed our simple three-cate-
adult modes of attaching and loving are largely deter- gory measure in 1985,we thought of it as a first attempt
mined by the type and degree of maternal responsive- to capture what appeared to be distinct patterns of
ness in infancy." They go on to state that, if attachment individual differences that kept cropping up in studies
theory is to work as a framework for the study of adult of relationshipphenomena. We started with Ainsworth
Downloaded by [McGill University Library] at 03:23 03 February 2015

relationships, "these assumptions must hold true." As et al.'s (1978) three infant categories, not because we
should be clear by now, their assumptions are based on believed that adult differences were direct results of
a mistaken conception of attachment theory and so, of relationship experiences in infancy and not because we
course, need not hold true. Perhaps the many inaccura- thought that adult patterns would necessarily be the
cies in the Hendricks' analysis are due to their nearly same as those identified in studies of infants, but be-
total reliance on a 10-year-old, secondary account of cause we had to start somewhere and because it seemed
the theory--a thought-provoking but now outdated possible to translate Ainsworth's categories into adult
summary and critique by Camps, Barrett, Lamb, language. The types also had intuitive appeal and,
Goldsmith, and Stenberg (1983). The tone of the based on an examination of the literature on relation-
Hendricks' commentary is curious from start to finish. ships, seemed to capture many of the reported differ-
They say they "wanted to believe in attachment the- ences in the way adults think, feel, and behave in close
ory." Why would two researchersworking on their own relationships. The volume of informative studies and
very different approach to adult love "want" to "believe the coherence of results from investigations around the
in" a rival theory? Belief in strikes us as an odd phrase world (reviewed by Shaver & Hazan, 1993) indicate
in this context in any case. Why would a scientist want that our measure was a good first step.
to "believe in" or accept any theory on blind faith? The When the measure was first published (Hazan &
Hendricks say the invitation to comment on our article Shaver, 1987), and we began to receive requests for
finally prompted them to investigate attachment the- copies, we responded with encouragement to modify,
ory and that, when they did, "previously unknown expand, and test it. Whenever we gave talks about our
controversies" were made apparent. The controver- work, we urged researchers to use attachment theory
sies to which they refer were discussed in some detail and not our measure as their starting point. From the
in our first article on the topic (Hazan & Shaver, outset, we expressed concern that researchers would
1987); in that article, we cited the Camps et al. accept our simple self-report measure of attachment
(1983) critique on which the Hendricks base the bulk style as the defining or only measure of the construct.
of their comments. More recent research compendi- For these reasons, we are puzzled by Noller and
ums (e.g., Belsky & Nezworski, 1988; Greenberg, Feeney's claims that the adoption of our model by some
Cicchetti, & Cummings, 1990; Parkes, Stevenson- researchers has "hampered attempts to look at crucial
Hinde, & Marris, 1991) have indicated that many of issues" and "[created] a research environment marked
the so-called controversiesof a decade ago have been by a reluctance to question and explore the basic tenets
empirically addressed, and many of the criticisms of the model." Noller and Feeney go on to say that it is
have been reformulated or laid to rest. now "crucial for researchers to start from basic theoret-
ical principles rather than from the perspective of a
given typology." We couldn't agree more. This is ex-
Research Issues
actly what we have encouraged all along, and we have
not noticed the "reluctance" to which Noller and
The theoretical and substantive issues discussed so Feeney refer. For example, Bartholomew (1990) ex-
far have obvious and important implications for the panded our typology to four categories. Callins and
measurement of adult attachment. Several commentar- Read (1990) and Simpson (1990) broke the measure
ies focus on measurement-related issues (e.g., into parts, added new pieces to incorporate more of
HAZAN& SHAVER

the theory, and created factor-based dimensional mea- tions of attachment figures' responsiveness;
sures. If there is any reluctance to deconstruct, probe, Sperling's model is based on the interaction of affil-
and reconstruct our model and measure, we haven't iation and aggression (Sperling & Berman, 1991);
noticed it. Bartholomew's model is based on models of self and
other (Bartholomew & Horowitz, 1991); and Main's
model (e.g., Main et al., 1985) is based in part on
Measurement patterns of defensiveness revealed in face-to-face
talk. How do we decide which, if any, of these is
We agree with Noller and Feeney that better mea- correct? What does correct mean? The ideal, we
sures and methods are needed in the area of adult think, is to derive relevant dimensions from the
attachment, and we agree with Bartholomew's point theory while remaining open to empirically based
that theory and measurement are inextricably linked. changes in the theory.
As noted in the target article, debates about measure- A separate but related issue concerns the notion of
ment have taken many forms, including a focus on types. Although Ainsworth et al. (1978) identifiedthree
whether individual differences are best conceptualized major patterns of attachment, they also delineated eight
in terms of dimensions or types. The dimensional ap- subtypes. Distinguishing between subtypes is not sim-
Downloaded by [McGill University Library] at 03:23 03 February 2015

proach (e.g., Feeney, Noller, & Callan, in press; Kobak, ply a statistical problem but also a conceptual one. Are
Cole, Ferenz-Gillies, & Fleming, 1993) seems most Bartholomew's dismissing-avoidantand fearful-avoid-
promising for several reasons. For one, it potentially ant categories really two separate types or simply sub-
makes measurement more precise. Few human person- types of a more general avoidant pattern? So-called A1
ality differences are truly categorical (although think- infants in Ainsworth's classification scheme cry less
ing in terms of categories may be useful when than A2 infants and exhibit less conflict, but both are
theorizing and generating dynamic hypotheses). For classified as avoidant because both avoid their car-
another, a dimensional approach facilitates compari- egiver during Strange Situation reunion episodes.
sons between measures and combinations of measures Adult fearful avoidants are more conscious of their
(through methods such as factor analysis). As Noller fears of intimacy (or at least are more willing to express
and Feeney suggest, if there are several dimensions to such fears) than are dismissing avoidants, but the out-
be considered, it is possible to use various kinds of come is often the same-avoidance of intimacy. Theo-
profile analysis, which should be more precise than our retically, it should be possible to identify different
three-category scheme. In addition, dimensionalscores subtypes of ambivalence as well; Ainsworth noticed
may reveal small changes in attachment that might be two such subtypes, one more passive than the other. At
either masked or exaggerated by categorical measures. the adult level, it is not difficult to imagine different
Fortunately, it is always possible to recapture types strategies for keeping others close. One such strategy
from dimensions if one desires qualitative categories might be to master the art of caregiving and try to make
for some reason. oneself an indispensable source of support. Another
The trick, of course, is to figure out which dimen- might be to make oneself appear vulnerable and in need
sions to include. This phase of measurement develop- of constant care. Bowlby (1979) labeled these compul-
ment should, in our opinion, be theory driven. It would sive caregiving and compulsive careseeking, respec-
be a mistake to think that the "right" number of types tively. There may be other ways for a person who fears
or dimensions can be discovered through empirical abandonment to reduce the chances of its happening.
means alone. We agree with Noller and Feeney that Noller and Feeney propose that there may be multiple
factor analysis can be helpful in resolving some of these ways of expressinginsecurity but only one way of being
substantive issues, but it is important to remember, as secure. Even among infants, however, security takes at
always, that factors reflect the variables that go into an least four reliably coded forms (Ainsworth et al., 1978),
analysis. Factor analysis cannot tell us whether the most varying in terms of proximity seeking, contact mainte-
important aspects of attachment have been included or nance, crying, and strength of greeting on reunion. It
left out. If the measures of adult attachment are to be would be surprising if adults were any less variable.
used to validate and test Bowlby's theory, at least one Stevenson-Hinde also comments on the issue of
essential dimension must be i n c l u d e ~ o n f i d e n c e measurement, but she is interested primarily in behav-
in the availability and responsiveness of attachment ioral indicators. (Crowell & Waters and R. S. Weiss
figures. also raise this issue.) Stevenson-Hinde says that, to her
As Bartholomew notes, current models of individual knowledge, "there are no observational studies of at-
differences in adult attachment are based on very dif- tachment behavior and a correspondingattachment be-
ferent conceptual analyses and underlying dimensions. havior system in adults, as there are in infancy and early
Our model (Hazan & Shaver, 1987) is based on percep- childhood." We think several studies qualify as inves-
AUTHORS' RESPONSE

tigations of attachment behavior in adults. For example, and Shaver (in press). Belsky and Cassidy's hypothesis
Mikulincer and Nachshon's (1991) studies of self-dis- that avoidant adults might express relatively little dis-
closure, Simpson et al.'s (1992) investigation of s u p tress at the termination of a romantic relationship was
port seeking in an anxiety-provoking situation, and confirmed by Simpson (1990). Their predictions about
Kobak and Hazan's (1991) study of caregiving and the role of attachment differences in the workplace are
joint problem-solving in marital couples all count as consistent with the results of a study we conducted on
observational studies of adult attachment. Moreover, the links between love and work (Hazan & Shaver,
there are many studies of couple communication and 1990).
interaction (reviewedby R. L. Weiss & Heyman, 1990) We agree with Belsky and Cassidy's claim than an
and reactions to loss of a partner (reviewed by Hazan individual-differences framework can do as much to
& Shaver, 1992) that were not necessarily inspired by advance the theory and its application to adult relation-
attachment theory but are nevertheless clearly relevant ships as the normative framework we proposed in our
to it. Bowlby constructed his theory on the foundation target article. Our deliberate emphasis on the norma-
of a creative and integrative review of studies already tive aspects of the theory was intended to redress
lodged in the literature, most of which were not in- what we see as a dangerous imbalance in adult at-
formed by his ideas. Surely the same thing can and tachment research. Personality and social psycholo-
Downloaded by [McGill University Library] at 03:23 03 February 2015

should be done with respect to adolescent and adult gists who follow up our work tend to adopt our
attachment. individual-differences measure or one of the several
However, it is important to stress that, when study- adaptations of it and then proceed to conduct re-
ing adolescents and adults, we need not limit our- search without studying the theory very extensively.
selves to behavior. Main et al. (1985) noted that the (Noller & Feeney make this point.) We worried that
attachment system begins to include cognitive and this exclusive focus on individual differences meant
affective "representations" at least by age 6. The that people were failing to see the universal aspects
theory itself is as much about the role of internal of the theory and were thinking of it strictly as a
working models as it is about observable behavior theory of types. (Lewis goes so far as to say that, if
(Bretherton, 1990). Bowlby acknowledged the im- it's not a theory of types, then it's not a very compel-
portance of representations by including, in his the- ling theory--a position with which we disagree.)
oretical integration, studies of depression and grief Crowell and Waters also build on our proposed
based on interviews and questionnaires. framework by filling in some of the gaps we left open,
We all share Stevenson-Hinde's wish for "such and, like Belsky and Cassidy, they specify interesting
golden tools [as the Strange Situation] in adulthood." hypotheses that can be tested. As Crowell and Waters
We think the Simpson et al. (1992) paradigm is a good rightly note, our description of the phases in our model
first step in this direction. However, we do not wish for of attachment formation and transfer were too brief and
a single silver bullet. In fact, the infant attachment lacking in detail. We appreciatetheir insightfulanalysis
research area has been frequently criticized for over- and their compelling argument about why-in the
reliance on a single assessment procedure. It seems transfer of primary attachments from parents to peers
likely that the adult attachment research area will es- and in the formation of reciprocal peer attachments in
cape this criticism. Adult attachment includes thoughts, adulthood-the safe-haven function should precede the
feelings, and behaviors--all of which can and should secure-base function. One of us (Hazan) recently com-
be measured. pleted a study of children, adolescents, and adults that
supports this argument.
Crowell and Waters (along with Simpson) question
Fellow Travelers our separation of caregiving and attachment. We think
the separation of these two behavioral systems is justi-
Several commentators (e.g., Belsky & Cassidy, fied both logically andempirically. Infants form attach-
Crowell & Waters, Kobak, R. S. Weiss) offer very ments long before they become eapable of serving as
useful suggestions for how to proceed with the exten- attachment figures, so the systems emerge at different
sion of attachment theory to the study of adult relation- points in development. Adult partners may intermingle
ships. Belsky and Cassidy recast our core questions in attachment and caregiving behaviors, whereas typi-
terms of individual differences and thereby pose some cally only the caregiving system is active In relation to
intriguing questions and put forward some promising infants and children. It is quite plausible that the two
hypotheses. There are already data relevant to some of systemsco-evolvedand that, inpair-bond relatiomhips
these. For instance, the speculation that avoidant indi- (to use Simpson's favored term), they are likely to be
viduals would be most satisfied with a partner who intimately interconnected As Stevenson-Hinde states,
demands little care was tested and supported by Kunce "To the extent that caregiving involves reaching out-
HAZAN & SHAVER

ward, it is likely that a caregiving behavior system also (e.g., Duck, Levinger) criticize us for leaving out top-
would be deactivated by arousal of an attachment be- ics, issues, and theories that they believe are essential
havior system." If she is right (and we think she is), the ingredients of any comprehensive framework for re-
ability of one system to preempt the other can be taken search on close relationships. For example, Duck notes
as strong support for conceptualizing them as separate that we left out sociology and communication theories
(but interrelated) systems. and failed to specify the level of the theory we are
Crowell and Waters note that the intensity of grief proposing. That is partly because we are both psychol-
and mourning reactions is similar for caregivers who ogists, but also because attachmenttheory is not limited
have lost a child and children who have lost a caregiver. to a single level. Marris (1991), a sociologist who has
Although we call both reactions grieving, the cognitive long worked within a broad attachment framework,
and emotional action-tendency components are proba- wrote that "attachment theory powerfully links the
bly not entirely the same. A person who is attached social and psychological aspects of human behavior.
experiences fear and a loss of security, whereas a care- John Bowlby's contribution will, I believe, be seen to
giver experiences a sense of deep sadness and an urge be as central to the development of sociology as it has
to continue providing care to the lost loved one. Adults been to psychology" @. 77). As for communication,
who lose a partner to whom they were attached and for some of the most interesting aspects of the Adult At-
whom they also provided care may experience grief as tachment Interview have to do with coding dimensions
Downloaded by [McGill University Library] at 03:23 03 February 2015

a mixture of these emotions. This is a good topic for labeled coherence of discourse and coherence of mind.
future research. We think Main's discovery that coherence of mind, as
R. S. Weiss makes additional suggestions about is- reflected in coherent discourse, is a major hallmark of
sues deserving further study and, in so doing, exhibits secure attachment is one of the most important contri-
the creativity that has marked his own use of attachment butions to attachment theory since Ainsworth's pio-
theory in the study of loss and loneliness. He says that neering studies. Attachment theory, as it has evolved in
"adults rarely see their spouses as awesomely superior response to solid empirical evidence, is a communica-
in the same way [that children view their caregivers]." tion theory (as well as an ethological theory, a psycho-
This is probably true, but it isn't clear that an adult has logical theory, a sociological theory, etc.)-perhaps
to see his or her partner as "awesomely superior," just one of the most important communication theories
"stronger and wiser" in the moment of need. If adults around (Bretherton, 1990). Finally, Duck asks whether
did not depend strongly on their partners for safety (safe attachment theory can explain the "dark side" of rela-
haven) and emotional support, they presumably would tionships. Two of the volumes in Bowlby's attachment
not grieve their loss so deeply. The partner is as much trilogy deal in detail with anxiety, depression, and
an "irreplaceable figure" as the caregiver in the eyes of anger.
the individual who is attached. A partner is likely to Levinger (along with Noller & Feeney) expresses
understand one's needs and be able to help and comfort doubts that something as complex as human relation-
one in ways that no other person can. A partner is ships can ever be explained by a single theoretical
"superior" in at least this important way. framework. We are unsure but cautiously optimistic.
R. S. Weiss says that children's security depends on Rather than shy away from theorizing about the com-
fairly frequent reassurance of the attentive presence of plexities of close relationships, we work toward the
their attachment figure but that, in pair-bonds, it is creation of more comprehensive and complex theories.
"much more muted." Nevertheless, adults do need at- By pointing out some of the limitations of attachment
tention and reassurance from their partners. Most want theory and drawing attention to other theories that
to talk about their day, like to be hugged or touched on might fill some of the gaps, Levinger has helped us see
a regular basis, and maintain regular contact by tele- what a sufficiently complex theory might look like.
phone or mail during separations. Even in adulthood,
frequent, prolonged, or unexpected separations from
one" primary attachment figure tend to be distressing Do We Need a Comprehensive Theory
(Vormbrock, 1993). of Close Relationships?

Our argument that the field of close relationships is


Additional Criticisms in need of a comprehensive theory was based more on
intuition and experience with the field than on careful
Some commentators note our injudicious uses of the analysis. Peterson's survey of the field's premier jour-
terms behavioral system and function (Stevenson- nal provides empirical support for our position. His
Hinde), need (R. S. Weiss), and felt security (Kobak). review of the Journal of Personality and Social Psy-
We aren't sure we completely agree with every recom- chology turned up 22 of 65 articles (33.8% of the total)
mendation, but we appreciate the clarifications. Others "[offering] relatively straightforward empirical de-
76
AUTHORS' RESPONSE

scription but no clear implications for general theory." differences--will be incomplete at best. We hope that
This is compatible with our point that there are data the conversation begun here, for which we owe the
lying around detached from theory. Ten of the 65 commentators a great debt of gratitude, will continue
studies (another 15.3%) were "intended to clarify major and will affect the field's construction plans.
relational concepts." Fine, but most of those articles
were divorced ftom a comprehensive theory of rela- Note
tionships. For example, Fehr7s (1988) methodologi-
cally and descriptively useful research on everyday
conceptions of love (cited by Peterson) was largely Cindy Hazan, Department of Human Development
guided by Rosch's (1978) theory of the nature of cog- and Family Studies, Van Rensselaer Hall, Cornell Uni-
nitive categories, not by a theory of love itself. This versity, Ithaca, NY 14853-4401.
means that at least 49.2% of the relationship research
that Peterson examined is largely atheoretical in the
sense we care about. Eleven of 65 studies (another References
16.9%) "tested relatively specific relational hypothe-
ses." It was our impression that many of these hypoth- Ainsworth, M. D. S. (1%7). Infancy in Uganda:Infant care and the
eses would make more sense in the context of a larger growth of nttnchment Baltimore: J o b Hopkins University
Downloaded by [McGill University Library] at 03:23 03 February 2015

Press.
theory. Thus, 66.2% of the relationship work he sur- Ainsworth, M. D. S. (1982). Attachment: Retrospectand prospect. In
veyed was not guided by a comprehensive relationship C.M.Parkes & J. Stevenson-Hinde (Eds.), Theplace of attach-
theory. Even among the remaining third, which Peter- ment in human behavior @p. 3-30). New York: Basic.
son refers to as being "directly linked with compre- Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978).
Patterns of attachment: Assessed in the Strange Situation and
hensive theoreticalformulations," we think it likely that at home Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
some could be fruitfully combined. For example, what Anisfeld, E., Casper, V., Nozyce, M., & Cunningham, N. (1990).
Peterson calls the "Darwinian theory of mate selection" Does infant carrying promote attachment? An experimental
might be productively and parsimoniously combined study of the effects of increased physical contact on the devel-
opment of attachment. ChildDevelopment, 61,1617-1627.
with attachment theory, perhaps along the lines sug- Bartholomew, K. (1990). Avoidance of intimacy: An attachment
gested in Simpson's commentary. perspective. Journal of Social and Personal R e k a t k h i p , 7,
We agree with Peterson that attachment theory is 147-178.
potentially important not only to abstract social science Bartholomew, K, & Horowitz, L. M. (1991). Attachment styles
among young adults: A test of a four-category model. Journal
but also to the solution of some of society's pressing of Personaliw and Social Psychology, 61,226-244.
problems, such as child abuse, domestic violence, di- Belsky, J., & Nezworski, T. (Eds.). (1988). Clinical implications of
vorce, delinquency, drug abuse, depression, and teen attachment theory. Hillsdale, NJ: Lawrence ErIbaum Associ-
pregnancy. A field of developmental psychopathology ates, Inc.
Belsky, J., & Rovine, M. J. (1987). Temperament and attachment
is emerging (e.g., Cicchetti, 1984, Sameroff & Emde, security in the Strange Situation: An empirical rapprochement.
1989; Sroufe & Rutter, 1984), and attachment theory is Child Development, 58,787-795.
influential within it. We certainly hope that our exten- Bowlby, J. (1973).Attachmentandloss: Vol. 2. Separation:Anviely
sions of attachment theory will eventually have clinical and anger. New York: Basic.
Bowlby, J. (1979). The making and breaking of affectional bondr.
and policy implications. London: Tavistock.
Bowlby, J. (1980). Attachment and loss: VoL 3. Loss: Sadness and
depression New York: Basic.
Conclusion Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment. New
York: Basic. (Original work published 1969)
Bowlby, J. (1988). A secure base: Parent-child attachment and
If, as Peterson puts it, we are to build better theo- healthy human development. New York: Basic.
retical buildings in the close relationships field, we Bretherton, I. (1990). Open communication and internal working
need to have a solid foundation. We think the foun- models: Their role in the development of attachment relation-
ships. In R. A Thompson (Ed.), Nebraska Symposium on
dation must consist of a theory concerning the most Mothution (Vol. 36, pp. 57-113). Lincoln: University of Ne-
basic and universal aspects of human relationships. braska Press.
Regardless of whether the field ever arrives at a Camps, J. J., Barrett, K, Lamb, M. E., Goldsmith, H. H., &
comprehensive theory or whether the eventual Stenberg, C. (1983). Socioemotional development. In P. H.
Mussen (Ed.), Handbwk of childpsychol~gy:Vol. 2. Infancy
framework is based on Bowlby's formulations, we and developmental pcychobiology @p. 783-915). New York:
contend that any theory that fails to acknowledge that Wiley.
human beings are biological creatures whose rela- Caspi, A, & Bem, D. J. (1990). Personality coatinuity and change
tionship behavior has been influenced by evolution- acros the life course. In L. A. Pervin (Ed.), Handbook of
personality: Theory and reswrch @p. 54S575). New York:
ary processes-any theory that is adevelopmental or Guilford.
that fails to explain both normative aspects and the Caspi, A, & Herbener, E. S. (1990). Continuity and change: Assor-
origins and nature of relationship-relevantindividual tative marriage and the consistencyof personality in adulthood.
HAZAN & SHAVER

Journal of Personality andSocia1Psychology, 58,250-258. chology, 59,270-280.


Cicchetti,D. (1984). The emergenceof developmental psychopathol- Hazan, C., & Shaver, P. R. (1992). Broken attachments. In T. L.
ogy. Child Development, 55,l-7. Orbuch (Ed),Close relationship loss: Theoretical approaches
Colin, V. (1985). Hierarchies andpatterns of infants' attachmentsto @p. 90-108). New York: Springer-Verlag.
parents and day caregivers:An erploration Unpublished doc- Isabella, R. A, Belsky, J., & von Eye, A. (1989). Origins of infant-
toral dissertation, University of Virginia, Charloffesville. mother attachment: An examination of interactional synchrony
Colin, V. (1987, April). Infant'spreferences betweenparents before during the infant's first year. Developmental Psychology, 25,
andajier moderate stressactivates behavior. Paper presented at 12-21.
the meeting of the Society for Research in Child Development, Kirkpatrick, L. A., & Davis, K E. (in press). Attachment style,
Baltimore. gender, and relationship stability: A longitudinal analysis.Jour-
Collins, N. L., & Read, S. J. (1990). Adult attachment, working MI of Personality and Social Psychology.
models, and relationship quality in dating couples. Journal of Kobak, R., Cole, H., Ferenz-Gillies, R., & Fleming, W. (1993).
Personali@andsocial Psychology, 58,644663. Attachment and emotion regulation during mother-teen prob-
Connell, D. B. (1976). Individual differences in attachment: An lem-solving. Child Development, 64,231-245.
investigatwn into stability, implications and relationships to Kobak, R. R., & Hazan, C. (1991). Attachment in mamage: The
struc&e of early langua& de~lopment. unpublished douoral effects of security and accuracy of working models. Journal of
dissertation, University of Syracuse. Personality andSocial Pschology, 60,861-869.
Crockenberg, S. B. (1981). infancir~itabiiit~, mother responsiveness, Kunce, L. J., & Shaver, P. R. (in press). An attachment-theoretical
and social support influences on the security of infant-mother approach to caregiving in romantic relationships. In K.
attachment. ChildDevelopment, 5 2 , 8 5 7 4 9 . Bartholomew & D. L. Perlman (Eds.), Advances in personal
Cummings, E. M. (1980). Caregiver stability andattachmentin infant relationships: Vol.5. London: Jessica Kingsley.
Downloaded by [McGill University Library] at 03:23 03 February 2015

day care. Developmental Psychology, 16,31-37. Main, M. (1990). Parental aversion to infant-initiated contact is
Egeland, B., & Farber, E. A (1984). Infant-mother attachment: correlated with theparent'sown rejection during childhood: The
Factors related to its developmentand changes over time. Child effects of experience on signals of security with respect to
Development, 55,753-771. attachment. In K. E. Barnard & T. B. Brazelton (Eds.), Touch:
Elicker, J., Englund, M., & Sroufe, L. A. (1992). Predicting peer The foundation of esperience @p. 461495). Madison, CT:
competence and peer relationships in chiidhood from early International Universities Press.
parent-child relationships. In R. Parke & G. Ladd (Eds.), Fam- Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy,
ily-peerrelatwns: Modes of linkage@p.77-106). Hillsdale, NJ: childhood, and adulthood: Amove to the level of representation.
Lawrence Erlbaum Associates, Inc. Monographs of the Societyfor Research in ChildDevelopment,
Farran, D. C., & Ramey, C. T. (1977). Infant day care and attachment 50(1-2, Serial No. 209), 66-104.
behavior towards mothersand teachers. ChifdDevelopment,48, Markus, H. (1977). Self-schemata and processing information about
112-116. the self. Journal of Personality and Social Psychology, 35,
Feeney, J. A, & Noller, P. (1990). Attachment style as a predictor of 63-78.
adult romantic relationships. Journal of Personality and Social Manis, P. (1991). The social construction of uncertainty. In C. M.
Psychology,58,281-291. Parkes, J. Stevenson-Hinde, & P. Manis (Eds.), Attachment
Feeney, J. A., Noller, P., & Callan, V. J. (in press). Attachment style, a c m s the life cycle @p. 77-90). London: TavistocWRoutledge.
communication and satisfactionin the early years of marriage. Mikulincer, M., & Nachshon, 0. (1991). Attachment styles and
In K. Bartholomew & D. L. Perlman (Eds.), Advances in per- patterns of self-disc1osure. Journal of Personality and Sociol
sonal relationships: VoL 5. A h l t attachment relationship. Psychology, 61,321-331.
London: Jessica Kingsley. Parkes, C. M., Stevenson-Hinde, J., & Marris, P. (Eds.). (1991).
Fehr, B. (1988). Prototype analysis of the concepts of love and Attachment across the life cycle. London: TavistocWRoutledge.
commitment.Journal ofPersonalityandSocialPsychology 55, Ricciuti, H. N. (1974). Fear and the development of social attach-
557-579. ments in the first year of life. In M. Lewis & L. Rosenblum
Fox, N. A, Kimmerly, N. L., & Schafer, W. D. (1991). Attachment (Eds.), Origins of fear @p. 73-104). New York: Wiley.
to motherlattachment to father: A meta-analysis. Child Devel- Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B.
opment, 62,210-225. Lloyd (Eds.), Cognition and categorization (pp. 27-48).
Greenberg, M . T., Cicchetti, D., & Cummings, M. E. (Eds.). (1990). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Attachment in thepreschoolyears: Theory, research, andinter- Sameroff, A. J., & Emde, R. N. (1989). Relationshipdisturbances in
vention. Chicago: University of Chicago Press. early 1ife:A developmental approach. New York: Basic.
Harlow, H. F. (1961). The development of affectional patterns in Schaffer, H. R., &Emerson, P. E. (1964). The development of social
infant monkeys. In B. M. Foss (Ed.), Determinants of infnnt attachments in infancy. Monographs of the SocietyforResearch
behavior(Vol.1). New York: Wiley. in Child Development, 29(3).
Harlow, H. F., & Harlow, M. K. (1965). The affectional systems. Shaver, P. R, & Hazan, C. (1993). Adult romanticattachment: Theory
In A. M. Schrier, H. F. Harlow, & F. Stollnitz (Eds.), Behavior and evidence. In W. H. Jones & D. Perlman (Eds.),Advancesin
of nonhuman primates p o l . 11, pp. 287-334). New York: personal relationship (Vol. 4, pp. 29-70). London: Jessica
Academic. Kingsley.
Hazan, C. (1992, July). Transitions and transformations in attach- Shill, M. A, Solyom, P., & Biven, C. (1984). Parent preference in
ment. Paper presented at the meeting of the International Society the attachment exploration balance in infancy: An experimental
for the Study of Personal Relationships, Orono, ME. psychoanalytic approach. Child Psychiatry and Human Devel-
&an, C., & Hutt, M. J. (1991). Continuity and chanae in internal o l ~ t ~ n15.34-48.
t.
working models of a&hhent. unpublished m&uscriPt, Cor- simpson,J. k (1990). Theinfluenceof attachment styles on romantic
nell University, Department of Human Development and Fam- relationships.Journa1 ofPersonalityandsocial Psychology, 59,
ily Studies. 971-980.
Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as Simpson,1.A, Rholes, W. S., &NeUigan, J. S. (1992). Support-seek-
an attachment process. Journal of Personality and Social Psy- ing and support-giving within couple members in an anxiety-
c h a m , 52,511-524. provoking situation: The role of attachment styles. Journal of
Hazan, C., & Shaver, P. R. (1990). Love and work: An attachment- Pwsonality andsocial Psychology, 62,434-446.
theoretical pers~ve.Jour~~ofPersona~~andSocin[Psy-Sperling, M. B., & Berman, W. H. (1991). An attachment
AUTHORS' RESPONSE

classification of desperate love. Journal of Personality As- Uitgeverij Eburon.


sessment, 56,4555. Vormbrock, J.K (1993). Attachment theory as applied to war time
Sroufe, L.A, & Fleeson, J.(1986). Attachment and the construction and job-related marital separation. PsychologicnlBulletin, 114,
of relationships. In W. W. Hartup & Z. Rubii (Eds.) Relation- 122-144.
s h i p and development @p. 51-72). Hillsdale, NJ: Lawrence W a e r , V. G. (1986). Attachment in infancy and at age sir, and
Erlbaum Associates, Inc. children's self-concept:A follow-up of a German longitudinal
Sroufe, L. A, & Rutter, M. (1984). The domain of developmental shuty. Unpublisheddoctoraldissertation, University of Virginia,
psychopathology. Child Development, 48,1184-1 199. Charlottesvifle.
Swann, W. B., Jr., %on, J. G., & De La Ronde, C. (1992). Embrae Waters, E. (1978). The reliability and stability of individual differ-
ing the bitter '%uth2': Negative self-concepts and marital com- ences in infant-mother attachment. Child Development, 49,
mitment. PsychologiurlScience, 3,118-121. 483-494.
van den Boom, D. (1990). Preventive intervention and the quality Weiss, R L,& Heyman, R. E. (1990). O b ~ e ~ a t i oofn marital
of mother-infant interaction and infant exploration in initable interaction. In F. D. Fincham & T. N. Bradbury (Eds.), The
infants. In W. Koops, H. J. G. Soppe, J. L.van der Linden, P. psychology of marriage @p. 87-117). New York: Guilford
C. M. Molenaar, & J. J. F. Schroots (Eds.), Developmental Weiss, R S. (1974). The provisions of social relationships. In Z
psychology behind the dikes: An outline of developmental Rubin (Ed.), Doing unto others. Englewood Cliffs, NJ:Pren-
psychology research in The Netherlands. The Netherlands: tice-Hall.
Downloaded by [McGill University Library] at 03:23 03 February 2015

Você também pode gostar