Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Recent Developments in Attachment Theory and Research*

SUSAN GOLDBERG, Ph.D. I

The history and development of attachment theory are the caregivers as teachers, attachment theory focuses on
reviewed. Research has focused on four major patterns of parents as protectors and providers of security. (All three
attachment in infancy: one pattern of secure attachment and theories acknowledge that parents play multiple roles, that of
three patterns of insecure attachment (avoidant, resistant,
teacher, caregiver, playmate, etc.; they differ with respect to
and disorganized). These patterns have been shown to reflect
different histories of parent-child interaction and affected which role is considered most influential.) Furthermore,
subsequent development up to age eight. More recently, meth- while the psychoanalytic and learning theories view children
ods have been developed for identifying similar patterns of as initially passive, Bowlby's view credited infants with
attachment in preschoolers.five to seven year olds and adults. active participation. Prior theories considered infants to be
Future research is likely to focus on the development of dependent on caregivers and dependency as a state which
attachment patterns and their transmission from one gener- must be outgrown. Attachment theory, however, considers it
ation to another. New data on the relationship between at- possible for individuals to be reciprocally attached. Attach-
tachment and behaviour problems has generated mutual ment is therefore a quality of relationships which lasts one's
respect and collaboration between clinicians and
researchers. lifetime. The nature of attachments may be transformed as
children develop, but an attachment can endure.
The concept of attachment includes social components (it
I t is widely accepted that the parent-child relationship plays
a central role in a child 's development, but empirical data
to support this hypothesis are very recent and remarkably
is a property of social relationships), emotional components
(each participant in the relationship feels emotional bonds
scant. The most common theoretical approaches to the study with the other), cognitive components (each participant forms
of parent-child relationships are psychoanalytic theory (ob- a cognitive scheme - a "working model" of the relationship
ject relations), social learning theory (dependency), and at- and its participants), and behavioural components (partici-
tachment theory. There has been research into each of these, pants engage in behaviours that reflect and maintain the
but it is attachment theory that has given rise to a recent wave relationship). It is the nature and interrelationship of these
of empirical studies that has excited both clinicians and components that reflect developmental change.
researchers. The goal of this article is to review the basic Over the first year, the infant's proximity-promoting
constructs of attachment theory and empirical research. behaviours (orienting signals such as cries and vocalizations,
and direct actions such as approaching and clinging to the
A CapsuleViewof Attachment Theory caregiver) become organized into a goal-oriented system
focused on a specific caregiver. The mother is usually the first
Attachment theory was originally described by Bowlby such figure, but others can play this role. When the attachment
(1-4), combining ideas from psychoanalysis and ethology. system is in its goal state (i.e., there is adequate proximity and
Bowlby argued that affectional ties between children and contact), attachment behaviours subside; when the goal state
their caregivers have a biological basis which is best under- is threatened, attachment behaviours are activated. Further-
stood in an evolutionary context. Since children's survival more, because the attachment system operates in the context
depends on the care they receive from adults, there is a genetic of other related systems (for example, exploration), the goal
bias among infants to behave in ways which maintain and is adjusted to fit the context. For a healthy infant in a familiar
enhance proximity to caregivers and elicit their attention and (safe) environment, the goal may be to remain in the same
investment. A complementary evolutionary history biases room with the attachment figure; if the infant is tired or ill or
adults to behave reciprocally. Thus, while psychoanalytic the environment is unfamiliar, the goal becomes greater prox-
theory emphasizes the caregivers' initial roles in reducing imity and contact.
physiological arousal, and social learning theory emphasizes As the child's locomotor, linguistic and social skills de-
velop, the goals of the attachment system are modified to
allow for longer separations over greater distances. Cognitive
*Manuscript received May 1991. components playa more dominant role and proximity plays
I Acting Director, Psychiatric Research Unit, The Hospital for Sick Children; a less important role in moderating attachment behaviour.
Professor of Psychiatry and Psychology, University of Toronto, Toronto, Individuals' working models of a particular relationships
Ontario.
Addressreprintrequeststo: Susan Goldberg, PhD, Psychiatric Research Unit, include concepts of themselves and others. A more general
The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario "working model" of relationships also develops which re-
MSG IX8 flects individuals' experiences in relationships. The quality of
both early and later attachments influences self-concepts as
Can. J. Psychiatry Vol. 36, August 1991 well as expectations and attitudes toward social relationships.

393
394 CANADIAN JOURNAL OF PSYCHIATRY Vol. 36, No.6

Individuals whose primary attachment relationships in child- feelings of security are the ontogenetic function of the attach-
hood were satisfying and provided emotional security view ment system. Bowlby acknowledged both the phylogenetic
themselves as lovable, expect positive interactions with oth- goal of protection from predators and external danger and the
ers, and value intimate relationships. Individuals who expe- ontogenetic function of psychologically perceived security.
rienced rejection or harsh treatment as children view Ainsworth elaborated on the latter in the Strange Situation.
themselves as unworthy oflove, expect further rejections, and The procedure involves observing the infant, caregiver (usu-
act in ways that elicit rejections. These predictions may not ally the mother) and a friendly but unfamiliar adult in a series
differ radically from those of other theoretical approaches to of eight semi-structured episodes in a laboratory playroom
relationships; however, the evaluation of a theory depends on (5,6). It relies on the observation of the balance of attachment
its ability to be empirically tested. and exploratory behaviours in response to the manipulations
in the eight episodes. The crux of the procedure is a standard
History of Research into Attachment: An Overview sequence of separations and reunions between the infant and
each of the two adults. It is felt that over the course of the
Attachment theory has generated 25 years of productive
eight episodes the child experiences increasing distress and a
empirical research. Empirical research was made possible by
greater need for proximity. The extent to which children cope
Ainsworth and her students (5,6), who classified infant-par-
with these needs and the strategies they use to do so are
ent attachment between 12 and 18 months of age into three
considered to indicate the quality of attachment.
distinct patterns. The first ten to 15 years of research based
Scoring depends on a detailed review of videotapes. The
on this method were devoted to collecting normative data and
infant is rated on behaviour directed at the caregiver: seeking
documenting the precursors and sequelae of different pattems
contact, maintaining contact, distance interaction, avoidance,
of attachment. Initial efforts to study patterns of attachment
and resistance to contact. From this information, the dyad is
in clinical populations followed (for example, maltreated
classified into one of eight subtypes that fall into three broad
infants, medically ill infants, infants of depressed mothers).
categories. Although behaviour during the entire session is
These efforts resulted in the addition of a fourth pattern of
considered, reunion behaviours have been shown to be the
attachment, which was thought to be potentially more patho-
most salient feature distinguishing between these patterns (6).
logical (7,8). However, the inability to examine patterns of
These three patterns reflect strategies used by the infant to
attachment beyond infancy soon became a recognized limi-
manage affective arousal during interactions with, separa-
tation, and more recent work has included the development
tions from, and reunions with the caregiver. In the secure
and validation of methods for assessing attachment in pre-
strategy, the attachment system is activated only when the
schoolers (9), five to seven year olds (10), and adults (11). An
infant's security is threatened (for example, the caregiver
outline of these schemes is shown in Table I. This conceptual
departs and the child is left in an unfamiliar place) and
scheme is likely to be completed in the future.
subsides to give the exploratory system free rein when the
attachment figure (the secure base) returns. In the avoidant
Patterns of Attachment
(dismissing) strategy, the attachment system is defensively
Since theories of infant-parent attachment were developed suppressed so that the child appears to be exploring without
before other age groups were looked at and have been most concern for security, although he carefully monitors the at-
extensively studied, they will be reviewed in detail as a tachment figure. In the ambivalent/resistant (preoccupied)
prelude to introducing the analogues for other age groups. strategy, the attachment system is continuously activated at
The assessment of infant-mother attachment developed by the expense of the exploratory system, even when to all
Ainsworth et al is based on their hypothesis that the infant's outward appearances the child should be safe and comfort-

Table I
Patterns of Attachment at Different Stages of Life
Patterns of Attachment
Age Assessment Method Secure (B) Dismissing(A) Preoccupied (C) Disorganized(D)
12 to 18 months Structured observation Secure Avoidant Ambivalent! Disorganized
(Strange Situation) resistant
2 1/2 to four years Structured observation Secure Avoidant Dependent Controlling/
(reunions) disorganized
Five to seven years Structured observation Secure Avoidant Dependent Controlling
(reunions)
Adult Interview Autonomous/ Dismissing Preoccupied Unresolved
secure mourning (loss)
August, 199L ATIACHMENT THEORY 395

able (i.e., the attachment figure is present). Another way of the infant's style of coping with distress, in particular use of
understanding this is to consider the threshold for activating the attachment figure as a source of comfort. Studies of parent
attachment behaviour: in the avoidant strategy the threshold reports of temperament have generally found little associa-
is very high, while in the ambivalent/resistant strategy, it is tion between scores on questionnaires of temperament and
very low. In both insecure strategies, the threshold is set security of attachment (22). However, more careful examina-
primarily to meet internal needs and is not adapted to the tion of the full classification scheme (including all the sub-
environment. However, in the secure strategy, the threshold groups) suggests that there isa temperament dimension
is both moderate and sensitive to environmental conditions. inherent in the scheme. Figure 1 illustrates the full classifica-
In Ainsworth's original study (5), 65% of the babies ex- tion scheme. Avoidant babies and some secure babies who are
hibited a secure pattern of attachment, 21 % an avoidant slightly avoidant (B 1) are less easily distressed than resistant
pattern and 14% an ambivalent/resistant pattern (6). A recent babies, very secure babies (B3), and secure babies who are
meta-analysis of nearly 2,000 infants from 39 studies con- slightly resistant (B4) . If we divide infants according to the
ducted in eight different countries showed almost exactly the degree to which they are likely to become distressed
same distribution, although there were some cultural varia- (high/low distress groups), infants in the high distress group
tions (12). Furthermore, under stable life conditions, patterns are more likely to be reported as temperamentally "difficult"
of attachment are relatively stable over both the short term by parents, and those in the low distress group are more likely
(six months) (13) and the long term (up to five years) (10). to be reported as "easy" (18). Thus, the distress experienced
These data refer to attachment to the same caregiver assessed during the Strange Situation is related to the type of security
on different occasions. or insecurity, not to security per se.
A number of studies have shown that patterns of attach- In the meta-analysis conducted by Fox et al (17), a high
ment to the mother and to the father are independent (14-16). degree ofconcordance was found between the infant's attach-
This demonstrates that attachment patterns derived from the ment to the mother and father in the high distress (A1-B z) and
Strange Situation reflect qualities of distinct relationships low distress (B3-CZ) groups. However, within each of these
rather than a trait of the child. In a recent meta-analysis of 11 groups, security to the mother and the father were not concor-
studies of attachment to the mother and the father, Fox and dant. Thus, if an infant falls into the A1-Bz group with one
his colleagues (17) found a high degree of concordance parent, s/he is likely to fall into the same group with the other
between patterns of attachment to mother and father, which parent, but within the A,-B z group, having an A (avoidant) or
raises some questions about the trait-versus-relationship in- B (secure) pattern of attachment with one parent does not
terpretation of attachment inferred from the Strange Situa- predict security or insecurity with the other parent. These
tion. These findings can be better understood if we consider findings show an association between temperament and
the relationship between temperament and attachment. behaviour on the Strange Situation Test, but they also indicate
There has been an ongoing discussion in the literature over that security is a characteristic of relationships, rather than an
the extent to which the infant's temperament affects attach- individual trait.
ment (18-22). It would seem the infants' propensity to become Patterns of attachment behaviour during the Strange Situ-
distressed is the primary determinant of behaviour during the ation are related to both prior and current behaviour in the
Strange Situation. However, the most salient behaviours for home. Mothers of secure infants have been rated as being
classification are not the presence or absence of distress, but more sensitive, responsive, accessible and cooperative during

Insecure ..........~~

Marginally Secure

(
I I
Fully Secure

Avoidant Secure Secure Secure Ambivalent


Avoidant Ambivalent

Low Contact High Contact

Low Distress High Distress

Figure 1. Attachment classifications.


396 CANADIAN JOURNAL OF PSYCHIATRY Vol. 36, No.6

the first year than mothers of insecure infants (23-26). Al- using this new category indicated a high proportion of disor-
though there is some evidence that characteristics and/or ganized attachment (38,41,42). These data lead to the specu-
behaviours of the infant may also predict the quality of lation that disorganization is a very insecure pattern of
attachment, these findings have been less consistent than attachment and that some children in previous risk popula-
those on maternal behaviour. This suggests that the mother tions who were actually disorganized were initially "misclas-
plays a more influential role than the infant in shaping the sified" as secure. This is one area where we can expect further
quality of the relationship. development, and it seems likely that disorganization of
An increasing number of studies have found that secure attachment may have implications for subsequent psychopa-
infants are more competent than insecure infants in a variety thology. However, some observed "disorganization" may
of subsequent cognitive and social skills (27-33). The most also reflect transient responses to events that are normally
ambitious undertaking in this domain is the longitudinal stressful to a young child (for example, the birth of a sibling).
research conducted by Sroufe and his colleagues (31-36) at Studies of the disorganized pattern of attachment are needed
the Institute for Child Study in Minneapolis. A large sample to provide normative data on the precursors and sequelae of
of children in whom attachment was assessed in infancy are disorganization. Psychiatric populations have not been
now reaching puberty. Published reports of data up through widely used in studies of attachment. Such studies may make
the early school years consistently document the influences major contributions to both clinical understanding and attach-
of early attachment on social skills in later life. ment theory.
Coherent relationships between behaviour at home and
attachment assessed in the laboratory and between the latter Beyond Infancy
and subsequent development in normative samples have been
demonstrated. This has led researchers to use the Strange While a few studies used the Strange Situation on children
Situation to explore possible "derailing" ofthe infant-mother up to age four (39,41), the absence of methodology for
relationship under unusually stressful conditions (for exam- assessing attachment beyond infancy has limited the devel-
ple, premature birth, maltreatment, maternal depression). It opment of attachment theory. What Bowlby had intended to
was thought that there would be fewer infants with secure be a life-span construct became an infancy construct because
attachment relationships with the parent in these populations infancy was the only developmental period in which attach-
at-risk. In general, medically ill infants are not more likely to ment could be measured. Although it is important to study the
develop insecure patterns of attachment (24,37,38). Some sequelae of early attachment in later development, it is
early studies found that maltreated infants are less likely to equally important to study the development of attachment
develop secure attachment relationships (38,39). However, it itself.
seems implausible that the majority of maltreated infants or The most important new directions concern the develop-
even a substantial minority should be securely attached (7) . ment of assessment methods for preschoolers, five to seven
These studies of "atypical" populations provided the first year olds, and adults (see Table I). As with the procedures
indication that there might be potential problems in attach- used for infants, those for preschoolers and five to seven year
ment theory or in the assessment methodology. Another intri- olds rely on videotaped observations of reunions. (The nature
guing finding was that while there had been a small number and duration of separations are not specified, but typically the
of cases that could not be classified into the three pattern child is engaged in a series of tasks with an experimenter.)
scheme, Crittenden's observations of maltreated toddlers (39) Adult procedure relies on interviews about the individual's
and Radke-Yarrow's assessments of preschoolers of de- early attachments. A detailed transcript of the interview is
pressed mothers (40) uncovered a substantial number of scored. The goal is not to make an objective determination of
children who seemed to be both avoidant and ambivalent/re- the nature of these relationships, but rather to assess the
sistant. In both studies infants with a mixed pattern of attach- individual's current working model of attachment. As shown
ment were from the most potentially damaging conditions. in Table I, the four patterns previously described for infants
Subsequently Main and Solomon (7,8) reviewed a large can be applied to the other age periods. These classification
number cases which had not previously been classified or schemes are relatively new, and research to validate them is
were anomalous (for example, securely attached infants who only beginning. Nevertheless, these preliminary descriptions
had been maltreated), by studying video tapes and identified provide some sense of the developmental continuity in attach-
a fourth pattern, which they called "disorganized". ment patterns which might be possible.
Unlike the previous patterns, which could be described as
Secure/Autonomous
strategies, the infants in the "disorganized" group did not
have a coherent strategy for coping with the separations and Secure preschoolers and five to seven year olds, like
reunions during the Strange Situation. In addition, they en- secure infants, greet a returning parent happily and are able
gaged in unusual and inexplicable behaviours which only to coordinate attention to the parent and exploration. Separa-
made sense if interpreted to reflect confusion or fear of the tion is less stressful to these children than to infants because
caregiver. The "mixed" pattern described by Crittenden (39) their cognitive schemas, which include recognition of parents
and Radke-Yarrow (40) was also included in this new disor- as independent individuals, are more sophisticated. Older
ganized group. Subsequent studies of maltreated children children may engage in enjoyable and absorbing activities
August, 1991 ATIACHMENT THEORY 397

during the separation. Thus greeting, while happy, is casual Disorganized/Controlling/UnresolvedMourning


and children usually continue their activities but find a way This is the least consistent category. Some signs of disor-
to involve the parent either by inviting the parent to join the ganization similar to those in infants can be seen in preschool-
activity or by volunteering information about what they have ers, especially the youngest preschoolers. However, in older
been doing. Conversation is fluent, and both participants are children the disorganized pattern seems to emerge as disor-
relaxed. There is comfortable eye contact, and children may ganization in the relationship rather than the individual. The
initiate physical proximity or contact. child takes control of the parent in one of two ways. The first
Secure adults (described as autonomous) value attach- is related to caregiving. The child appears to feel responsible
ments and are able to talk coherently and realistically about for making the parent happy and is overly bright and enthu-
them. If they had positive relationships with their parents, siastic at reunions. The child works hard to engage the parent.
they can acknowledge and accept their parents' imperfec- The second is punitive. The child is directly hostile toward
tions. If they had a difficult or harsh childhood, they can the parent in a style that conveys assurance that the parent
acknowledge the unhappiness, have come to some under- will comply or meet demands. The child may also ignore the
standing of it, and can establish new and important relation- parent when it is clear that such ignoring is a flagrant violation
ships for themselves. of the social conventions.
AvoidantlDismissing The significant feature of adults in the "disorganized"
category is unresolved mourning over the loss of an attach-
Avoidant preschoolers and five to seven year olds, like ment figure. The loss may be a death or loss through divorce
avoidant infants, appear to be more interested in other activ- or a loss of trust through abuse or neglect. Initial mourning is
ities than in the parent's return. However, they have learned typically characterized by disorganizing and disorienting ex-
that social conventions require greetings as well as responses periences. Unresolved mourning is inferred from interview
to initiations by the parent. Avoidance in older children is transcripts where signs of continuing cognitive disorganiza-
therefore more subtle and is shown by a lack of eye contact, tion are shown when the attachment figure is discussed (for
lack of social initiatives' and a minimal response to parental example, disbelief in the loss, persisting inappropriate guilt).
overtures. The strategy of avoidance is to maintain neutrality At all ages, disorganized individuals are also given an
- to do nothing either positive or negative that would draw alternative or "forced" classification of one of the other three
attention to the relationship. alternatives (avoidant, secure, ambivalent) .
Avoidant or "dismissing" adults likewise downplay the
importance of intimate relationships. They may idealize their Development of Attachment
childhood experiences without being able to supply support-
We have already noted that a core group of studies on
ing details or, in fact, give contradictory examples. Some may
infants found prior observations in the home to be predictors
speak of negative experiences but do not acknowledge the
of attachment status based on the Strange Situation. Indeed,
effects of these experiences. They attempt to limit the influ-
behaviour during the Strange Situation is a marker or indica-
ences of attachments on themselves. The reader of the inter-
tor of the relationship history. Since the "disorganized" clas-
view transcript can infer a history of lack of closeness or
sification was developed, studies of maltreated children have
support, or significant rejections.
consistently shown that disorganized attachment is more
Dependent/Preoccupied common in maltreated infants than in controls (38-40,42). In
addition, initial data from an ongoing longitudinal study of
Among older children, the insecure/ambivalent pattern is the children of clinically depressed mothers document a high
labelled "dependent" and, like the infant pattern, its hallmark rate of "mixed insecure" attachment (A/C, now considered a
is preoccupation with the relationship at the expense of other form of disorganized attachment) among children with moth-
activities. The parent and child may be engaged in a constant ers who have major bipolar affective disorder (40). These
struggle for control, the conversation marked by "put-downs" studies support the assertion that secure attachment is the
and disagreements. The child is whiny and contentious. Al- result of appropriately responsive parental care and that inad-
ternatively, the child may emphasize his dependence with equate care is the result of very insecure attachment. Further-
extreme coyness (for example, whispering) and feigned help- more, in the Minnesota longitudinal study (43), it was shown
lessness. Even at this age, dependent children are more likely that when infants are tested in the strange situation, at 12 and
to be visibly upset at the parent's departure than children in 18 months, changes in quality of attachment are related to
other groups. changes in family environment. If the mother's quality oflife
The adult analogue of this pattern is labelled "preoccu- improves, the infant-mother relationship is more likely to
pied". These individuals are caught in old struggles with become more secure; if the mother's quality of life deterio-
parents, lack a sense of personal identity apart from family or rates, infant-mother attachment is likely to become less se-
parents, and are unable to evaluate their own role within cure.
relationships. They are unable to move beyond details ofearly Now that methods are available to assess attachment in
memories or current interactions with parents to an objective preschool and early school years, it is possible to evaluate
overview. stability and change in attachment relationships over a longer
398 CANADIAN JOURNAL OF PSYCHIATRY Vol. 36, No.6

period if time. In the first such study, Main and her associates Implications for Child Psychiatry
(10) found that the correlation between security of attachment
The methods developed by researchers studying attach-
to the mother at one year and security of attachment to the
ment can be applied to studies of the relationship between
mother at six years was r = .76 (p < .001); the analogous
attachment (particularly in infancy) and behaviour problems.
correlation for attachment to the father at 18 months and six
Most of the existing studies have looked at relatively large
years was r = .30 (p < .05). Several other studies of this type cohorts of normally developing children. The number of
are under way. From these we can expect to leam about children likely to have behaviour disorders that can be diag-
factors contributing to stability and change in attachment nosed would be small in this group. Therefore, the measure
status. of outcome is generally a parent or teacher report checklist
(47-49) or a composite score on experimental observation and
Transmission of Attachment Patternsbetween standardized social competence measures (34-36).
Generations The findings of these studies are equivocal. For example,
With the development of the Adult Attachment Interview Sroufe and his colleagues (34-36) have shown that the quality
of early attachment is related to later behaviour problems in
(AAI) (11), it is also possible to test whether individuals'
children up to early school age. However, this relationship is
experiences of nurturing determine their ability to nurture
stronger in boys than in girls. Lewis and his colleagues (48)
their own children. Indeed, the first studies of this type show
reported similar findings, but two other studies (47,49) failed
a high degree of concordance between adult "states of mind"
to find such an association.
regarding attachment and the attachment status of their in-
fants (44-46); parents who are "secure/autonomous" tend to In our own recent preliminary data, we have found that
while there is no consistent relationship between quality of
have securely attached infants, those who are "dismissing"
attachment in infancy and scores on the Achenbach's Child
tend to have avoidant infants, and those who are preoccupied
Behavior Checklist (CBCL), a different pattern emerges if we
tend to have ambivalent/resistant infants. While disorganized
consider only children whose parental report scores were high
attachment in infants has been associated with maltreatment enough to place them in the clinical range. For example, at
and maternal bipolar affective disorder, it also occurs among three years of age, seven percent of children who had had
infants in low risk families. In these circumstances, the initial insecure attachment relationships as infants were scored in
indications are that disorganized attachment in the infant is the clinical range of the CBCL by their mothers, compared
associated with unresolved mourning in the parent (46). Al- with two percent of securely attached infants. Figures based
though it has not yet been clearly demonstrated, it is possible on the fathers' reports were similar: 12% of insecurely at-
that parents who maltreat children have experienced trau- tached infants in the clinical range and five percent of se-
matic and unresolved loss as a result of inadequate care as curely attached infants. The relative odds of scoring in the
children. clinical range were therefore 2.4 to 3.5 times higher if the
child had had an early insecure attachment than if the child
The AAI does not purport to provide an accurate picture
had had a secure attachment.
of childhood. It represents the adult's working model (i.e,
present attitudes and feelings) of important relationships dur- Thus, measures of outcome are important in this type of
ing their childhood. Even the preliminary data now available study. When the actual number of diagnosable cases is ex-
indicate that a significant number of adults with unhappy pected to be small, a full psychiatric assessment for each
child in a large cohort is difficult to justify. Standardized
childhoods are secure/autonomous adults (10). Likewise,
parent and teacher reports, however, do not provide psychi-
some adults who are insecure were probably securely at-
atric diagnoses. It may be that differences in scores below the
tached during earlier periods of life. The indications from the
clinical cutoff are not clinically relevant measures.
current generation of studies are that an adult's cognitions and
emotions (i.e., working models) about relationships influence An alternative approach, and one that may be of greater
their caregiving, which in tum affects their childrens' attach- use in associating specific diagnoses with particular attach-
ment patterns, is the use of measures of attachment on an
ment status. While those who had secure childhoods clearly
experimental basis as part of routine assessment of children
have a better chance of becoming secure adults than those
referred to clinics for behaviour problems. When attachment
whose initial experiences were harsh and rejecting, interven-
was measured (using the measure used to assess preschoolers)
ing experiences can change working models of attachment in 25 children referred for one of the DSM-III-R disruptive
and lead to behaviour consistent with the new model. This is behaviour disorders and 25 children who were not referred,
good news for clinicians, whose goal is to bring about a 84% of the clinic children were classified insecure compared
positive change through therapeutic intervention. The avail- with only 25% of the non clinic group (50). Furthermore, a
ability of a method for assessing adult attachment now makes significantly greater proportion of the children who were
possible the use of pre- and post-therapy assessments to referred to the clinic fell into the insecure/controlling cate-
ascertain whether therapy has resulted in changes in working gory (40% versus 12%). This is the first study to demonstrate
models of attachment. an association between clinic status and attachment quality
August, 1991 AITACHMENT THEORY 399

and between a specific diagnosis and a specific form of interdisciplinary collaborations are a necessary step in mak-
attachment. ing that future a reality.
References
Collaboration between Researchers and Clinicians
Much of the work reviewed above has been carried out by 1. Bowlby J. The nature of the child's tie to his mother. Int J
psychologists studying child development who communicate Psychoanal 1958; 39: 350-373.
2. Bowlby J. Attachment and loss: attachment. New York: Basic
with each other and publish their work in journals not rou- Books, 1969.
tinely read by psychiatrists. Many developmental psycholo- 3. Bowlby 1. Attachment and loss: separation. New York: Basic
gists remain cautious and skeptical of this work (51). Yet, Books, 1973.
whenever psychiatrists, particularly child psychiatrists, have 4. Bowlby J. Attachment and loss: loss, sadness and depression.
been exposed to these ideas, methods and findings, there has New York: Basic Books, 1980.
been enthusiastic reception. Why is this so? 5. Ainsworth MDS, Wittig BA. Attachment and exploratory be-
havior of one-year-olds in a strange situation. In: Foss BM, ed.
First, the concepts of attachment theory are familiar to Determinantsof infant behavior. London: Methuen, 1969.
many child psychiatrists. John Bowlby was a psychiatrist and 6. Ainsworth MDS, Blehar MC, Waters E, et aI. Pattems of
wrote from a psychiatric perspective. However, there is now attachment: a psychological study of the Strange Situation.
considerable empirical evidence to support what child psy- Hillsdale NJ: Erlbaum, 1978.
chiatrists always knew: experiences with a primary caregiver 7. Main M, Solomon J. Discovery of an insecure-disorga-
influence important aspects of personality. There is the prom- nized/disoriented attachment pattern. In: Brazelton TB, Yog-
man MW,eds. Affectivedevelopmentin infancy.NorwoodNJ:
ise in these new findings of a firm scientific basis for some
Ablex, 1986.
aspects of common belief and practice. 8. Main M, Solomon J. Procedures for identifying infants as
Second, research on attachment combines clinical and disorganized/disoriented during the Ainsworth Strange Situa-
experimental techniques. These methods rely on standard tion. In: Greenberg MT, Cicchetti D, Cummings EM, eds.
manipulations or probes and evaluation of resulting behavi- Attachment in the preschool years. Chicago IL: University of
our by schemes that can be objectively described and repli- Chicago Press, 1990.
9. Cassidy J, Marvin RS. Attachment organization in three- and
cated. However, an important component of this evaluation four-year-olds: coding guidelines (unpublishedmanual).Char-
is clinical in nature - reliance on detailed sensitive observa- lottesville VA: Department of Psychology, University of Vir-
tions of the patterning and organization of behaviour. The ginia, 1990.
training of coders, whether to score reunions from videotape 10. Main M, Kaplan N, Cassidy J. Security in infancy, childhood
or review interview transcripts, is primarily training in clini- and adulthood: a move to the level of representation. Monogr
cal judgement. Detailed descriptions of salient behaviours are Soc Res Child Dev 501985; (1-2, Serial No. 209): 6-104.
provided, but the coder must also learn how to use clinical 11. George C, KaplanN, Main M. The Berkeley Adult Attachment
Interview(unpublishedprotocol). BerkeleyCA: Departmentof
intuitions. Thus, developmental psychologists working in this
Psychology,Universityof California, 1985.
field are gaining more appreciation of clinical approaches to 12. Van Ijzendoom MH, Kroonenberg PM. Cross cultural pattems
human behaviour. of attachment: a meta-analysis of the strange situation. Child
Clinicians have been quick to see the possible applications Dev 1988; 59: 147-156.
of the methods used in research on attachment to clinical 13. Waters E. The stability of individual differences in infant-
practice. Such efforts could lead to important clinical insights mother attachment. Child Dev 1978; 49: 483-494.
14. Grossmann KE, Grossmann K. Parent-infant relationships in
as well as information relevant to attachment theory. How- Bielefeld.In: Immelma.n K, Barlow G, PetrinovichL, et al, eds.
ever, caution is warranted. These experimental methods are Behavioral development: the Bielefeld interdisciplinary proj-
time consuming and costly in clinical practice. Furthermore, ect. New York: Cambridge University Press, 1981.
the available data are group data, not individual data. We do 15. Lamb ME. Qualitativeaspectof mother-infantandfather-infant
not know, for example, how frequently classification errors attachments.Infant Behav Developm 1978; 1: 265-275.
are made or what factors influence them. This is less impor- 16. Main M, Weston DR. The quality of the toddler's relationship
tant for group research data than for the clinical assessment to mother and to father: related to conflict behavior and readi-
ness to establish new relationships. Child Dev 1981; 52: 932-
or treatment of individuals. In addition, all but the methods
940.
used with infants must be considered to be in a very early 17. Fox N, Kimmerly NL, Schafer WO. Attachment to mother/at-
stage of development. Nevertheless, a number of projects tachment to father: a meta-analysis. Child Dev 1991; 62: 210-
supported by research funding are using these methods in 225.
clinical settings as part of assessment and treatment (50,52), 18. Belsky J, Rovine M. Temperament and attachment security in
and these projects promise to provide insight into the devel- the strange situation: an empirical rapprochement. Child Dev
opmental history of psychopathology. 1987; 8: 787-795.
19 Chess S, Thomas A. Infant bonding: mystique and reality.Am
The result of this research has been a growing collabora- J Orthopsychiatry 1982; 52: 213-222.
tion between researchers and clinicians, and between psy- 20. CrockenbergSB. Infant irritability,mother responsiveness, and
chologists and psychiatrists. The history of attachment social support influences on security of infant-mother attach-
research is short but it has shown early promise. These ment. Child Dev 1981; 52: 857-865.
400 CANADIAN JOURNAL OF PSYCHIATRY Vol. 36, No.6

21. Goldsmith HH, Alansky JA. Maternal and infant temperamen- 41. Lyons-Ruth C, Connell D, Zoll D, et al. Infants at social risk:
tal predictors of attachment: a meta-analytical review. J Consult relations among infant maltreatment, maternal behavior, and
Clin Psychol1987; 55: 806-816. infant attachment behavior. Dev Psychol 1987; 23: 223-232.
22. Sroufe LA. Attachment classification from the perspective of 42. Carlson V, Cicchetti D, Barnett D, et al. Disorganized/disori-
infant-caregiver relationships and infant temperament. Child ented attachment relationships in maltreated infants. Dev Psy-
Dev 1985; 56: 1-14. chol 1989; 25: 525-531.
23. Belsky J, Rovine M, Taylor DG. The Pennsylvania Infant and 43. Egeland B, Farber EA. Infant-mother attachment: factors re-
Family Development Project. Ill: the origins of individual lated to its development and changes over time. Child Dev
differences in infant-mother attachment: maternal and infant 1984; 55: 753-771.
contributions. Child Dev 1984; 55 : 718-728. 44. Levine L, Ward M, Carlson B. Attachment across three gener-
24. Goldberg S, Perrotta M, Minde K, et al. Maternal behavior and ations: grandmother, mother, and infants. Paper presented at the
attachment in low birthweight twins and singletons. Child Dev World Association of Infant Psychiatry and Allied Disciplines,
1986; 57: 34-46. Lugano, Switzerland, 1989.
45. Van Ijzendoorn MH. Intergenerational transmission of parent-
25. Grossmann K, Grossmann KE. Maternal sensitivity and
ing: a review of studies in non-clinical populations. Developm
newborns' orientation responses as related to quality of attach-
Rev (in press).
ment in northern Germany. Monogr Soc Res Child Dev 1985;
50(1-2, Serial No. 209): 233-256. 46. Main M, Hesse E. Parents unresolved traumatic experiences are
related to infant disorganized attachment status: is frightened or
26. Main M, Tomasini L, Tolan W. Differences among mothers of frightening behavior the linking mechanism? In: Greenberg M,
infants judged to differ in security. Dev Psychol 1979; 15: Cicchetti D, Cummings EM, eds. Attachment in the preschool
472-473.
years: theory, research and intervention. Chicago IL: University
27. Arend R Gove F, Sroufe LA. Continuity of individual adapta- of Chicago Press, 1990: 161-184.
tion from infancy to kindergarten: a predictive study of ego 47. Goldberg S, Corter C, Lojkasek M, et al. Prediction of behavior
resiliency and curiosity in preschoolers. Child Dev 1979; 50: problems in 4-year-olds born prematurely. Developmental Psy-
950-959. chopath 1990; 2: 15-30.
28. Bell S. The development of the concept of object as related to 48. Lewis M, Feiring C, McGuffog C, et al. Predicting psychopa-
infant-mother attachment. Child Dev 1970; 41: 291-311. thology in sixyear-olds from early social relations. Child Dev
29. Lieberman A. Preschooler's competence with a peer: relations 1984; 55: 123-136.
with attachment and peer experience. Child Dev 1977; 48: 49. Bates JE, Bayles K. Attachment and the development ofbehav-
1277-1287. ior problems. In: Belsky J, Nezworski T, eds. Clinical implica-
30. Londerville S, Main M. Security of attachment, compliance, tions of attachment. Hillsdale NJ: Erlbaum, 1988.
and maternal training methods in the second year of life. Dev 50. Speltz ML. The treatment of preschool conduct problems: an
Psychol1981; 17: 289-299. integration of behavioral and attachment concepts. In:
31. Matas L, Arend RA, Sroufe LA. Continuity and adaptation in Greenberg M, Cicchetti D, Cumming EM, eds. Attachment in
the second year. The relationship between quality of attachment the preschool years: theory, research and intervention. Chicago
and later competence. Child Dev 1978; 49: 549-556. IL: University Press, 1990: 399-426.
32. Sroufe LA, Fox N, Pancake V. Attachment and dependency in 51. Lamb ME, Thompson RA, Gardner WP, et al. Security of
developmental perspective. Child Dev 1983; 54: 1335-1354. infantile attachment as assessed in the "strange situation": its
33. Waters E, Wippman J, Sroufe LA. Attachment, positive affect, study and biological interpretation. Behav Brain Sci 1984; 7:
and competence in the peer group: two studies in construct 127-147.
validation. Child Dev 1979; 50: 821-829. 52. Lieberman A, Pawl J. Clinical applications of attachment the-
ory. In: Belsky J, Nezworski T, eds. Clinical implications of
34. Erickson MF, Sroufe LA, Egeland B. The relationship between
attachment. Hillsdale NJ: Erlbaum, 1988: 327-351.
quality of attachment and behavior problems in a preschool
high-risk sample. Monogr Soc Res Child Dev 1985; 50(1-2, Resume
Serial No. 209): 147-166.
35. Renken B, Egeland B, Marvinney D, et al. Early childhood L' auteur trace l' historique de la theorie de l' attachement
antecedents of aggression and passive-withdrawal in early et passe en revue l' evolution de la recherche dans ce domaine.
elementary school. J Pers 1989; 57: 257-281. Cette derniere est axee principalement sur la determination
de quatre modes principaux d' attachement, observes d' abord
36. Sroufe LA, Egeland B, Kreutzer T. The fate of early experience
dans la prime enfance: un attachement solide et trois
following developmental change: longitudinal approaches to
attachements peu solides (evitement, resistance et
individual adaptation in childhood. Child Dev 1990; 61: 1363-
desorganisation). On a montre que ces modes d' attachement
1373.
traduisent diferentes interactions parent-enfant et influent
37. Goldberg S. Risk factors in attachment. Can J Psychol 1988; sur le developpement jusqu' a I' age de 8 ans. Plus
42: 173-188. recemment, on a mis au point certaines methodes permettant
38. Spieker SJ, Booth C. Maternal antecedents of attachment qual- d'identifier des modes d' attachement comparables chez les
ity. In: Belsky J, Nezworski T, eds. Clinical implications of en/ants en bas age, les enfants de 5 a 7 ans et les adultes. A
attachment. Hillsdale NJ: Erlbaum, 1988. l' avenir, it est probable que la recherche porte principalement
39. Crittenden P. Maltreated infants: vulnerability and resilience. J sur le developpement et la transmission intergenerationnelle
Child Psychol Psychiatry 1985; 26: 85-96. des modes d' attachement. Le present travail contribue au
40. Radke-Yarrow M, Cummings EM, Kuczynski L, et al. Patterns domaine de la pedopsychiatrie, car it fournit de nouvelles
of attachment in two- and three-year-olds in normal families donnees concernant le lien entre l' attachement et les
and families with paternal depression. Child Dev 1985; 56: problemes de comportement, et engendre un respect mutuel
591-615. et une collaboration entre cliniciens et chercheurs.

You might also like