Clinical Psych&~ Reuim~, Vol. 8, pp. 61 l-636, 1988 027%7358/88 $3.00 + .

OO
Printed in the USA. All rights reserved. Copyright 0 1988 Pergamon Press plc

ATTACHMENT THEORY, PERSONALITY

DEVELOPMENT, AND PSYCHOTHERAPY

Randolph 1. Paterson
Greg Moran

University of Western Ontario

ABSTRACT. Attachment tbzory, currently a dominant thme in the study of ear& social
development, is beginning to have a presence in clinical psycholou as well. A review of the theory
as proposed by Bowlby is followed by an examination of the related research literature. Research
using Ainsworth? Strange Situation paradigm suggests that the infant’s attachment status is a
genuine product of the intiraction between mother and child rather than a measure of infant
temperament, and is related to a broad array of developmental variables. Relatively little research
has effectively exam&d th link between the rrwthr-infant relationship and adult socialfunction-
ing, however, due in part to difftiulties with the measurement of attachment-related constructs. It
is concluded that while the theory continues to evolve and some critical issues remain unresolved,
attachment may be a useful construct for conceptualizing many disorders, and bus the potential to
provide valuable insights regarding the process and techniques of psychotherapy.

In recent years attachment theory has been applied with growing frequency to
the practice of psychotherapy, particularly by those who favor a psychodynamic
approach (e.g., Turco, 1982). In its ethological origins and expanding research
base, attachment theory seems to provide a welcome empirical justification for a
central tenet of psychodynamic theory: That early life experiences strongly influ-
ence subsequent adult functioning and vulnerability to psychopathology. The
empirical literature on attachment is immense; it is presently a major force in the
study of early social development. But practitioners not overly familiar with the
recent developmental literature may find themselves somewhat at sea when at-
tempting to evaluate this new influence in the clinical setting.
The present article is divided into two main sections in order to address the
issues of theory and research. First, a brief review of the history and nature of
attachment theory is presented. Primary emphasis is given to the writings of John
Bowlby. In the second and larger section the empirical literature is examined. An

for reprints should be sent to Randolph J. Paterson, Department of Psychology,

fwestern Ontario, London, Ontario, Canada N6A 5C2.

611
612 R. J. Paterson and G. Moran

exhaustive review is beyond the scope of this article or, for that matter, any
presentation short of a multi-volume work. Instead, the focus is limited to issues
that have provided the primary impetus for research or have strong implications
for clinicians. Although at present the research literature offers the clearest guid-
ance to those working with children, such readers may be more familiar with the
attachment literature than specialists in adult work and may wish to pursue these
issues in more depth than can be presented here. Consequently, greater emphasis
is given to issues dealing with adulthood than is often the case in discussions of
attachment theory.

THE ORIGINS AND PRINCIPLES OF ATTACHMENT THEORY

The origins of attachment theory can be traced to John Bowlby, a psychiatrist

whose work in residential facilities for children and adolescents centered on the
role of environmental factors in development. Trained in the psychoanalytic tra-
dition, he became convinced of the paramount importance of the relationship in
the first years of life between the infant and the primary caretaker in the develop-
ment of the self and of later social behavior (Bowlby, 1958). Bowlby joined with
the learning theorists of the time in criticizing psychoanalytic theory’s inadequate
empirical basis, but he also appreciated the difficulty inherent in testing complex
psychoanalytic concepts. In his search for a way out of this conundrum, he turned
to ethology.
Bowlby (1957) f ound that ethologists had discarded simple catalogs of “in-
stincts” as being oversimplifications of complex behavioral processes which are
dependent upon extensive environmental feedback and serve multiple goals.
They had progressed to thorough descriptions of these behavior systems and the
analysis of their adaptive function and evolutionary origins. Psychoanalytic theo-
ry shared ethology’s early reliance on the concept of instinct. Bowlby concluded
that a more rigorous description of human behavior and consideration of its
adaptive function would serve psychoanalysis as it had ethology. His subsequent
work drew heavily on ethological concepts, culminating in the publication of his
influential trilogy, Attachment and Loss (1969, 1973, 1980).
Two animal research programs are of particular interest in association with
attachment theory. Lorenz’ (1952) studies of imprinting demonstrated the tenden-
cy of the young of some species to bond to a specific individual soon after birth.
Related work by Harlow and his colleagues extended these findings to the attach-
ment behavior of monkeys. The best known of these studies demonstrated that,
given a choice, infant rhesus monkeys spend more time with a non-lactating cloth-
covered surrogate mother than with a lactating one made of wire (Harlow, 1958).
This finding indicates that the infant monkey’s attachment to its mother does not
rest solely on her status as a secondary reinforcer based on an association with
food. It has become clear that infants of many species are biologically prepared to
develop an attachment to the mother or a mother-figure independent of the direct
reinforcement of known physiological needs. Mothers are likewise prepared to
respond in specific ways to the presence or absence of their infants, who are
equipped with behavior patterns (e.g., crying) designed to elicit caring responses
(Harlow, 1971; Jensen, 1965).
Bowlby (1957) interpreted these data in terms of their adaptive significance.
Infant monkeys are relatively helpless (and infant humans even more so), yet they
 Attachment Theory 613

have a wide variety of needs. They must have nourishment, water, protection
from predators, and so on. Without the support of others the infant would perish.
It is adaptive for the infant to develop an attachment to another individual who
will satisfy these needs until the infant can meet them him/herself. So too, the
reproductive success of the mother is enhanced if she helps the infant. The stage is
set for the development of a behavior system in which the mother is biologically
“ready” (or prepared to learn) to perform specific caring actions and the infant is
ready to perform actions that will elicit such caregiving. Harlow (1971) and
Bowlby (1984) h ave both presented accounts of this system, and where they cover
the same ground they are largely in agreement. But it is Bowlby whose work has
provided the basis for much subsequent research with humans, and his influence
that is at the core of attachment theory today.
Bowlby defines an attachment as a bond developed with “some other differenti-
ated and preferred individual, who is usually conceived as stronger and/or wiser”
(1977a, p. 203). Attachment behavior is “any form of behavior that results in a
person attaining or retaining proximity” (1977a, p. 203) to an attachment figure.
Bowlby (1977a) has listed the defining features of attachments and attachment
behavior: Attachments are directed toward specific individuals, are characterized
by their long duration, and serve the biological function of survival. They have a
strongly positive emotional component while they are being formed (falling in
love), maintained (security), or renewed (joy), and a strongly negative one when
they are threatened (anxiety and/or anger) or lost (sorrow and grieving). Since
attachment is a primary and not a secondary drive, attachments may develop
independently of direct reinforcement by the attachment figure (as was the case
for Harlow’s monkeys). Finally, attachment behavior is controlled by a specific set
of conditions for activation (such as fear) and termination (such as interaction
with the attachment figure).
The theory clearly states that the attachment system is a central feature of the
developing relationship between infants and their caretakers. Bowlby further sug-
gests that attachment plays a role beyond infancy. He states that attachment
behavior is most easily activated before the age of three, and with declining ease
thereafter (1984). But, while an adult is clearly more self-sufficient than an infant
and not in need of constant attention from a caregiver, there are certain situations
in which a more knowledgeable or experienced individual is of great help. The
range of situations in which attachment behavior is activated may become more
narrow, but will not vanish altogether. An attachment figure may be perceived as
stronger and/or wiser in only some situations, thus allowing for attachment to
occur among relative equals, as in strong friendships and romantic relationships
(1977a).
Attachment may also affect adult functioning in a more significant way.
Through repeated transactions with its attachment figures, the infant is said to
form internal representations both of the self and of others. These “internal
working models” (Bowlby, 1973) subsequently form a heuristic basis for future
relationships. The model of the self structures expectations about one’s own role
in relationships. For example, an infant whose needs are typically left unmet may
develop a model of the self as being unworthy of the attention of others and
incapable of influencing their behavior. The model for others serves as a template
for future relationships, forming the fundamental assumptions and beliefs about
how others will behave. The adult neglected by mother in infancy may believe that
614 R. J. Paterwn and G. Moran

each new person met will prove to be inaccessible, uncaring, and unresponsive.
Bowlby (1980) suggests that these models act outside conscious awareness and so
are resistant to change. They are most relevant in close relationships involving
attachment, and relationships simulating the caretaker-recipient bond (such as
relations with authority figures), but they may also affect relations with friends
and acquaintances.
Some attachment behaviors, particularly the more passive varieties such as
crying, are present from birth. Others, notably the “social smile,” appear over the
ensuing weeks and months. Initially the attachment “relationship” is somewhat
one-sided, as at birth the infant is incapable of discriminating the primary care-
taker (usually mother, and who will be referred to as such henceforth) from others
and so does not develop an exclusive attachment relationship with her immediate-
ly. From an ethological perspective this is nonessential, as the infant is not yet
mobile. At this stage attachment behavior may instead encourage the develop-
ment of the mother’s bond to the infant, thereby promoting her proximity and
caregiving.
By about 6 months, however, the infant recognizes and has developed an
exclusive relationship with the mother. This is now an adaptive feature, since new
behaviors such as crawling, walking, and speech begin to enter the repertoire and
can be used in the service of attachment if the infant can recognize a particular
individual to whom he or she can approach and call. Between 6 months and 1 year
the infant commonly develops “stranger anxiety”; unfamiliar figures are met with
resistance and anxiety. While this typically appears soon after recognition of the
mother as a separate person, it is not considered an index of the quality of the
attachment with her. Instead it seems to serve the adaptive function of encourag-
ing the infant to avoid potentially dangerous, noncaregiving others in favor of the
familiar caregiver. Over subsequent months and years stranger anxiety diminish-
es. Others frequently in the presence of the child may also become attachment
figures, but the mother usually remains primary (Bowlby, 1977a).
Once an attachment is established, the infant uses the mother as a secure base
from which to explore the environment. Ainsworth (1982) emphasizes the impor-
tance of the exploratory behavior system which, like the attachment behavior
system, is essential to the individual’s future welfare. When the child “is rested,
healthy, and not hungry, his attachment system is likely to be inactive”
(Ainsworth, 1982, p. 4). Rather than exhibiting attachment behavior designed to
increase proximity to mother, the exploratory system is activated and the infant
(utilizing many of the same skills employed by the attachment system) moves away
to examine and manipulate the surroundings. In this way skills in dealing with the
environment are developed.
As time passes, the distance from mother increases, or an apparent danger
arises, the attachment system is reactivated and the infant attempts to reestablish
contact by moving toward the mother, by emitting attachment behavior (calling or
crying) designed to bring her, or simply by some form of social referencing such as
establishing eye contact. Once the presence and availability of the secure base has
been established, the infant moves away again and recommences exploration. The
exploratory system is most likely to be activated by novelty, and the attachment
system by threat. The more strongly activated system gains expression
(Ainsworth, 1982). Th ese complementary, mutually inhibitory systems ensure
 Attachment Theory 615

that the infant learns about the environment but does not become lost or exposed
to danger unnecessarily.
The theory’s claim that exploratory behavior occurs more often when the secure
base is available than when it is not raises an important point. Rather than
conceptualizing dependence on mother and the development of autonomy as
being antagonistic dynamics, the two are seen as mutually facilitative. Harlow
(197 1) found that infant monkeys placed in a strange test room would first cling to
their surrogate cloth mothers, then explore their surroundings. In contrast, those
placed alone in the room froze and did not explore. Although attachment and
exploratory behaviors are incompatible, the attachment figure must be available
for the exploratory system to find outlet.
It follows that the unwilling separation of the young child from the mother
should activate the attachment system. Based on both casual observation and
studies in residential nurseries and hospitals, Bowlby (1958, 196 1) has described a
consistent behavioral sequence following such separations. The initial phase is
characterized by crying and demands for mother’s return that potentially function
to draw her attention or the attention of others who will bring her. If this fails,
then despair sets in, characterized by continued longing for mother but without
hope that she will actually return. The attendant depression and inactivity may
give way, on occasion, to outbursts of rage. Finally, the child becomes detached
from the problem, appearing to have forgotten the mother. When she eventually
returns, the child may initially be unresponsive or avoidant of the mother. Provid-
ed the separation has not been prolonged (such as over 6 months), this avoidance
gives way to intense clinging, anxiety, and anger. Exploratory behavior is sharply
curtailed, as though the child does not trust the mother to remain available. Over
time the relationship may gradually return to normal.
Inadequacy of the attachment figure may be expressed in ways other than
physical absence. The primary function of a parent, according to Bowlby, is to
provide a secure base and encourage exploration from it (1977a). The parent
must be readily available to the child, responsive to its needs, and accepting of the
child as a person. The child should not be subjected to threats of withdrawal of
love or caring, or threats to abandon the family. Finally, the child should not be
made to feel responsible for the parent’s illnesses or death. All of these interfere
with the child’s perception of the parent as a secure base and may, Bowlby sug-
gests, have far-reaching consequences. If the course of normal attachment is
disrupted in these ways or by traumatic separation (either repeated, extended, or
permanent), then the child can be considered at risk to develop a generalized fear
of losing its attachment figures.
Several patterns of disorder can develop from this anxiety. Bowlby (1977a)
suggests that the most common is that of “anxious attachment,” in which the
individual is constantly fearful of losing attachment figures with the result that
attachment behavior is activated too easily. According to Bretherton (1985), this
pattern may result when the child “has to take more than optimal responsibility
for maintaining proximity” (p. 11). The pattern is analogous to the clinging
response demonstrated toward parents upon reunion after a separation. The
anxiously attached individual is said to become excessively clinging and jealous in
relationships, and often attempts to test and verify the strength of bonds by being
manipulative. Of course, such a pattern tends to disrupt attachments and so may
616 R. J. Paterson and G. Moran

become self-perpetuating. Such individuals, continually fearing desertion, are

liable to harbour high levels of anger toward their attachment figures. At the same
time, they are likely to be unable to express anger effectively because they fear this
could lead to the feared event itself- loss of the attachment. In combination, these
features result in the displacement of much anger onto others. The pattern is
reminiscent of the DSM-III-R (American Psychiatric Association, 1987) diag-
noses of borderline and dependent personality disorders.
If anxious attachment is similar to the clinging stage of childhood reunion, then
“compulsive self-reliance” (Bowlby, 1977a) is parallel to the more pathological
withdrawal period. Compulsively self-reliant individuals have essentially lost hope
of finding an adequate attachment figure, and avoid close relationships. Their
apparent independence is largely illusory, motivated not by free choice or an urge
to explore and develop their own resources, but by a fear of attachments and their
anticipated consequences. The absence of attachments in adulthood may suppress
growth-oriented exploratory behavior in these individuals, just as absence of the
attachment figure does in childhood. The description of these individuals suggests
the DSM-III-R diagnoses of schizoid and avoidant personality disorders.
Bowlby suggests other patterns of disorder which find their origins in disturbed
attachment histories, but they are variations on these two themes. For example,
compulsive caregiving (1977a) refers to an individual who denies personal needs
to fulfill those of others. The endlessly self-sacrificing mother (the “don’t-worry-
about-me” syndrome) is a salient example. Bowlby sees this pattern in part as
compulsive self-reliance, but it may also be a manipulative strategy to elicit
gratitude in the form of a sustained relationship-a pattern related more to
anxious attachment. Each of the problems can arise if the base is not secure or if
the parent fails to encourage the child to explore. The specific form of the attach-
ment deficits will determine the pattern that arises. However, Bowlby does not
fully elaborate on this mechanism or the specific links between early attachment
and adult disorder.
Early attachment is most clearly disrupted when no primary attachment figure
is available, or when attachments are formed only to have the attachment figure
become permanently unavailable. Consequently, much of the clinically oriented
literature on attachment has focussed on the results of separation from and loss of
attachment figures. Such loss is more easily identified than less understood or
quantifiable aberrations of the early attachment relationship. Bereavement thus
provides an excellent illustration of the application of attachment theory to human
functioning and psychopathology.
The normal course of mourning a lost attachment has been described by
Bowlby (1961; Bowlby & Parkes, 1970) and by Parkes (1972). Bowlby and Parkes
(1970) describe four stages, of which the latter three closely parallel the stages of
the child’s reaction to separation. The first and usually shortest stage (lasting from
several hours to several days) is characterized by numbing and a lack of affect,
occasionally broken by sudden displays of anger or anxiety. The second stage
features intense longing for the lost figure and can often last for several months.
Parkes (1972) suggests that the representation of the world at this stage continues
to include the presence of the lost figure. The individual automatically interprets
various stimuli within that context. A face on the street may cause momentary
“recognition” of the deceased, or an extra place may be set for a meal. Alarm and
 Attachment Theory 617

searching responses are manifestations of attachment behavior designed to regain

the lost figure. In situations of temporary loss such behavior would be adaptive.
As in anxious attachment, this stage of loss parallels descriptions of the child’s
initial response to separation.
Eventually the world view in which the lost figure is alive breaks down and is
replaced by disorganization and despair, the third stage (Bowlby & Parkes, 1970).
Feelings of emptiness are common, perhaps because interaction patterns asso-
ciated with the figure are not activated and have not been replaced by other
activities. Anger and other emotions become milder, and pangs of grief are re-
placed by longer periods of apathy and depression (Parkes, 1972). This stage
resembles the withdrawal reported in longer childhood separations. In the final
stage, the individual reintegrates the world view into a new model excluding the
lost figure. New roles are adopted, and new behaviors not dependent on the
other’s presence are produced. Here again, this pattern reflects the final stage of
childhood separation, when the attachment is broken and attachments with others
may be formed in its place.
The process of bereavement is complicated by a number of variables that may
lead to significant problems of adjustment. Maladaptive patterns of attachment,
such as anxious attachment or compulsive self-reliance may arise particularly if
the loss occurs at an early age. Bowlby and Parkes (1970) argue that grieving a
death is more difficult for children than adults. Children seem to find it difficult to
understand the finality of death, and their extreme dependence imparts a much
greater threat to the loss of the attachment. In addition, the strength of childhood
attachments (Bowlby, 1977a) predisposes them to more difficulty when a central
attachment figure dies. This can only be exacerbated by the frequently inade-
quate understanding of the child’s position by surviving adults. Raphael (1982),
in a study of families in which one parent had died, found that many children had
difficulty understanding the concept of death. Many more were not taken to the
funeral, and some had not even been told of the death. The surviving parent often
believed that the child had recovered quickly, attributing behavioral disturbances
to other causes, or missing such problems completely in the haze of their own loss.
It seems that while children are often the most affected in these situations, they
frequently have the least support. If loss can lead to subsequent disorder, as
attachment theory suggests, then loss in childhood is likely to be particularly
damaging.
Both adults and children sometimes avoid grief altogether, or fail to progress
through the stages of the grief process (Parkes, 1972). Both patterns can lead to
unusually prolonged disturbances of functioning. Several factors can predispose
the individual to a pathological grief reaction (Parkes, 1972). These include a pre-
existing problem with attachments, such as the excessive dependency of the anx-
iously attached individual. The experience of loss could make an anxiously at-
tached person even more conscious of the precariousness of attachments, and the
compulsively self-reliant individual yet more reluctant to engage in them. Anoth-
er predictor of problems in grieving is pre-existing ambivalence toward the lost
figure, such that the expression of emotions essential to successful grieving would
necessitate the recognition of hostile affect just when such bond-threatening feel-
ings are likely to be avoided most.
While the above discussion is directed specifically at instances of bereavement,
618 R. J. Paterson and G. Moran

analogous processes are postulated for other types of loss. To the extent that the
primary attachment figure fails to provide a secure base, inadequate caretaking is
also a situation of deprivation. Attachment theory thus attempts to relate early
interactions with caretakers to later development, and to certain patterns of pathol-
ogy (Bowlby, 197713). Given that most psychopathology involves a social compo-
nent, attachment theory could have at least indirect relevance to most clinical
cases.
RESEARCH ON ATTACHMENT

Failure to Thrive

One of the more basic principles of attachment theory is that the infant has a
primary need to develop a relationship with a caretaker. Harlow’s early studies
(1958) led him to assert an inborn desire for contact comfort in rhesus monkey
infants. If the need is indeed primary, then failure to form an attachment should
produce adverse consequences, even when other physiological needs are met.
Such is the situation of some children who have either been neglected by their
parents (Pollitt, Eichler, & Chan, 1975), or who have been raised in institutional
settings where individualized attention was not provided (Provence & Lipton,
1962; Spitz, 1945). These infants may exhibit normal attachment behaviors, such
as crying, smiling, and babbling, up to 6 months of age. But in the absence of an
appropriate object for their attachment, the behavior patterns disappear. At 1
year these children rarely vocalize and become rigid when picked up. They also
typically display restricted growth, delayed developmental milestones (including
speech), poor motor control, greater vulnerability to disease, and general weak-
ness (Tibbits-Kleber & Howell, 1985). Disturbances of feeding and digestion are
common (Call, 1984), and social responsiveness is very poor (Tibbits-Kleber &
Howell, 1985).
This cluster of symptoms attendant upon failure to develop attachments is now
well accepted. The associated diagnostic category, Reactive Attachment Disorder
of Infancy (RAD), is attributed in DSM-III-R to a lack of adequate caretaking
(American Psychiatric Association, 1987). There is disagreement over whether
RAD is directly attributable to inadequate stimulation or more specifically to the
absence of a unique attachment to a single caretaker. Provence and Lipton (1962)
have shown that institutionalized infants given toys and other sensory stimulation
but only limited attention by caregivers were substantially as impaired as infants
not given such stimulation. Further, several studies have indicated that individu-
alized caretaking can overcome the symptoms of RAD, provided that such care is
initiated before the age of 3 years (Clark & Hanisee, 1982; Dennis, 1973; Rutter,
1979). Attachment itself, rather than stimulation, appears to be critical. Some
studies have found that infants given adequate care by multiple caregivers seem
perfectly normal (Levy-Shiff, 1983). This suggests that a unique relationship may
not be essential, although some infants raised with multiple caregivers are found
to develop specialized attachments to one or two of them (e.g., Marvin, 1977).
In sum, there is substantial evidence that simply satisfying the physiological
needs of infants in the absence of adequate social stimulation is insufficient. The
specific requirements of relationships with caregivers are not well defined. How-
ever, it is clear that the absence of such relationships has a significant detrimental
impact on many areas of early development.
 Attachment T%zory 619

Ma ternal Bonding

The past 15 years have witnessed the popularization of the notion that the period
immediately after birth is critical to the development of the mother’s bond to her
infant. The theory is sometimes likened to imprinting- mistakenly so, since
imprinting involves the newborn fixating on a being it identifies as “mother”
independently of what the mother may feel about it. Such bonding is useful in
precocial animals, as it prevents mobile young from straying too far from their
caretaker, but is less important in altricial species such as humans. Since human
babies are not capable of discriminating one individual from another until several
months after birth, any bonding that occurs at this stage must be from m.other to
infant.
The most influential proponents of the critical period hypothesis have been
Marshall Klaus and John Kennell. In a longitudinal study, control mothers were
shown their babies briefly following birth and had only short visits for the next 3
days, while experimental group mothers were given 1 hour of contact within 2
hours of birth and spent extra time with them for the next 3 days. At 1 month and
1 year, several measures indicated that the experimental mothers were more
closely involved with their babies (Klaus et al., 1972; Kennel1 et al., 1974) and at
1 year experimental group babies scored significantly higher on the Bayley devel-
opmental assessment. At 2 years, mothers in the experimental group used more
elaborate and varied child-directed speech than controls (Ringler, Kennell, Jar-
vella, Navojosky, & Klaus, 1975). At age 5 the groups did not differ in intelligence
(Ringler, Trause, Klaus, & Kennell, 1978).
Klaus and Kennel1 (1976, 1982) h ave interpreted the results to be supportive of
the critical period hypothesis. However, subsequent reviews (Lamb, 1982; Myers,
1984) have pointed out a number of methodological problems, including a failure
to adequately account for Type I errors in view of the large number of variables
examined, a trend toward the decay of effects over time, and the failure to repli-
cate findings across studies.
A number of recent studies have further undermined the hypothesis. Gomes-
Pedro, de Almeida, da Costa, and Barbosa (1984) found that initial differences
favoring an early-contact group tended to decline over time, and Svejda, Campos,
and Emde (1980) found no differences on measures of mother-infant interaction
only 36 hours after delivery. A supposed consequence of the failure of bond forma-
tion is that the mother will be less motivated to take good care of her child. But
Egeland and Vaughn (198 1) found that delivery complications, prematurity, hospital-
ization time, and other indications of potential bonding failure all failed to distinguish
between mothers identified as providing suboptimal versus superior care.
In her review of the literature, Myers (1984) points out that if the critical period
is indeed critical, then experiences during that period should have effects that
amplify as the child develops, “fanning out” across measures and traits. In con-
trast, any gains attributable to early mother-infant contact seem to disappear over
time. It would appear that the period soon after birth may simply be a relatively
sensitive (rather than critical) period, perhaps due to a primacy effect or the
ordeal of birth. To the degree that the critical period hypothesis has encouraged
hospital maternity wards to allow early contact and humanize the birthing pro-
cess, it is to be applauded. There is, however, little evidence that early contact
plays a critical role in bond formation.
620 R. J. Paterson and G. Moran

Ainsworth’s Strange Situation

Infants are presumed to use a variety of behaviors in the service of attachment,

including crying, smiling, establishing eye contact with the attachment figure,
and crawling or walking toward her (Bowlby, 1969). In past years several studies
have assessed the practicality of using these behaviors as indicators of individual
differences in attachment status (e.g., Coates, Anderson, & Hartup, 1972). Un-
fortunately, disappointingly low correlations between these supposedly related
measures were typically obtained. Lamb (1974) interpreted these results by ob-
serving that different behaviors are called into play in different situations, each
acting independently in the service of the same goal: increasing proximity to the
mother. An infant seeking to increase proximity would not cry, smile, and run to
mother all at the same time, but would select whichever behavior was perceived to
be most appropriate and most likely to succeed under the circumstances at hand.
Given this, low correlations between the different behaviors would be expected.
According to this view, differences in attachment would have to be assessed
according to overall patterns of behavior in carefully delimited, replicable
circumstances.
Enter Ainsworth and Wittig’s (1969) longitudinal study of attachment in a
Baltimore sample. They developed a scenario, the Strange Situation, designed to
allow the assessment of the 12- to 24-month-old infant’s relationship to a caretaker
in a series of circumstances:

An infant and mother are brought into a comfortable laboratory room; a stranger
enters and sits talking to the mother and then to the infant; the mother leaves the
room unobtrusively; the mother returns and the stranger leaves them together; the
mother leaves the infant alone in the room; the stranger returns; the mother returns
once more. Each of these seven episodes lasts three minutes unless the infant is more
than mildly distressed. (Ainsworth, 1982, pp. 6-7)

The episodes typically become increasingly stressful for the infant, thus serving to
elicit attachment behaviors. The procedure provides a standardized context with-
in which attachment and exploratory behavior systems can be assessed. The
session is videotaped and is later scored on scales of proximity and contact seek-
ing, contact maintaining, resistance, avoidance, search, and interaction at a dis-
tance. The infant’s responses to reunions with mother are frequently regarded as
the most telling clues to attachment status.
Infants are classified into one of three groups. Infants classified as type B are
said to be securely attached, using the mother as a secure base, perhaps becoming
upset at separation, and greeting her upon reunion, returning to exploration
afterward. Those classified as A or C are said to be anxiously attached. A’s are
considered avoidant: they substantially ignore the caretaker when playing and
ignore or avoid her when she returns. There is little proximity-seeking. This
seems to resemble an early form of compulsive self-reliance as described by
Bowlby (1977a). Type C’s are called resistant or ambivalent, mixing contact-
seeking with angry rejection of the caretaker, and are so focused on the caretaker
that they do not play independently. This is reminiscent of Bowlby’s description of
anxious attachment in the adult. In addition, there are eight attachment subclas-
sifications, but the majority of studies use only the more general categories.
 Attachment Theory 621

More recently, some researchers have proposed the use of a fourth category that
soon may have an important impact on research in this area: the insecure-disor-
ganized/disoriented or D infant (Main, Kaplan, & Cassidy, 1985). These infants
display considerable apparent confusion and behavioral disorganization upon
reunion with the parent, including undirected affect, contradictory patterns and
incomplete movements. This category is only now being employed in studies of
attachment and therefore does not figure in this review.
Many researchers collapse categories A and C to form a dichotomous classihca-
tion of securely versus anxiously attached infants. In their review Lamb, Thomp-
son, Gardner, Charnov, and Estes (1984) g ive some support to this practice,
suggesting that there is more evidence supporting the distinction between B and
non-B infants than between A and C infants. However, to the degree that A and C
groups vary in underlying dynamics, the practice may have the disadvantage of
introducing more variance than necessary to the insecure group, perhaps cloud-
ing potential relationships with other variables.
The controlled setting of the Strange Situation is an advantage over more
naturalistic observation techniques, but prompts questions of external validity.
However, several well-controlled studies have shown that behavior in the Strange
Situation is reflective of the mother-infant relationship at home (Ainsworth,
1982; Gaensbauer et al., 1985).

Attachment, Infant Temperament, and Internal Models. A more serious debate in the
literature has concerned whether infant behavior in the Strange Situation is truly
attributable to the nature of the relationship with mother or simply reflects infant
temperament. Kagan (1982) has taken the latter position, arguing that an infant’s
own predisposition to respond to desertion and reunion in specified ways, regard-
less of the status of the relationship with the caregiver, may be the primary
determinant of attachment classifications received. Sroufe (1985) disagrees, argu-
ing that the bulk of evidence supports the interpretation of Strange Situation
behavior as a product of the caregiver-infant relationship. A number of findings
which are of interest in their own right can be brought to bear on this issue.

Attachment Stability Over Time. One line of research relevant to this debate is the
degree to which attachment status remains consistent over time. A widely ac-
cepted view states that if Strange Situation behavior reflects the relationship, then
attachment status should remain stable in the absence of major stresses on the
relationship or either partner, but should show less stability in the face of such
stress. The majority of studies demonstrate high levels of stability over time. In a
summary of nine studies that have examined this issue that was compiled by
Lamb, Thompson, Gardner, & Charnov (1985) the average concordance
(weighted for sample size) for middle-income families and families rated as pro-
viding excellent care was 77.1% .
Other studies have examined less stable families. Vaughn, Egeland, Sroufe,
and Waters (1979) assessed attachment in low SES families at 12 and 18 months,
finding only 62 % concordance overall, and an association between stressful events
acting on the family and shifts away from the optimal B classification to an
insecure attachment. Egeland and Farber (1984) report a 60% 12- to 18-month
concordance in a low SES sample. Egeland and Sroufe (1981a, 1981 b) drew from
622 R. J. Paterson and G. Moran

this pool two subsamples identified as providing seriously inadequate and excel-
lent care. The concordance rate in the inadequate care group was 48 % , compared
to 81% for the excellent care group. Thompson, Lamb, and Estes (1982) found
that maternal employment and the start of nonmaternal care can significantly
shift attachment classification. Although some infants were observed to change
from secure to insecure, others went from insecure to secure.
Overall, the data on temporal stability tend to support those arguing that
attachment classification is a reflection of the interaction between mother and
infant, rather than infant temperament alone. When family life is reasonably
stable, attachment classifications also tend to be stable. In more stressful environ-
ments attachment status seems to change more often, and tends to change toward
insecure attachment. This evidence is more difficult to explain from the tempera-
ment perspective.

Attachment Status Across Caretakers. Research examining the security of attach-

ment status across different caretakers also supports the relationship view. If
classification was a product of infant temperament, then one would expect the
infant to achieve the same classification with mother and with father. Several
studies have shown a moderate tendency for the attachment classifications
achieved with different parents to be concordant at higher than chance levels
(Belsky, Garduque, & Hrncir, 1984; Lamb, 1978; Lamb, Hwang, Frodi, & Frodi,
1982; Owen & Chase-Lansdale, 1982). However, the majority of studies find no
significant relationship between the two classifications (Grossmann & Gross-
mann, 1981; Grossmann, Grossmann, Huber, & Wartner, 1981; Lamb, 1977;
Main & Weston, 1981; Sagi et al., 1985). In fact, Main, Kaplan, and Cassidy
(1985) report a zero correlation between attachment security toward mother and
toward father.

Maternal Personality and Disturbance. Finally, some researchers have examined

whether mothers’ personality characteristics or caretaking styles can be reliably
related to the attachment classifications achieved by their infants. If the attach-
ment classification is the product of the relationship, one would expect it in turn to
be related to maternal caretaking style and, perhaps more indirectly, to maternal
personality characteristics. Tronick, Ricks and Cohn (1982) found that mothers of
securely attached infants, as assessed at 12 months, rated themselves as higher in
self-esteem, more likable, and more competent than mothers of insecurely at-
tached infants. In a prospective study of maternal personality characteristics,
Egeland and Farber (1984) found few differences between groups, although moth-
ers of securely attached infants were found to be less aggressive than mothers of
insecure infants. Benn (1986) h as reported a significant association between a
global measure of maternal emotional functioning and security of attachment,
with better-functioning mothers having a higher proportion of securely attached
children.
Several studies have looked more specifically at attachment to mothers suffering
from psychological disorders. Radke-Yarrow, Cummings, Kuczynski, and Chap-
man (1985) found insecure attachment to be more common among children of
mothers with unipolar and bipolar depression than among children of mildly
depressed or normal mothers. Naslund, Persson-Blennow, McNeil, Kaij, and
Malmquist-Larsson (1984), using an abbreviated version of the Strange Situa-
 Attachment Theory 623

tion, found a significantly greater incidence of anxious attachment among the

children of schizophrenic mothers relative to controls. D’Angelo (1986) found that
the regular Strange Situation used with the same sample revealed a significant
tendency for children of schizophrenics to be avoidant, and for children of de-
pressed mothers to be ambivalently attached.

Carefaking Style and Attachment Status. The effects of personality and psychiatric
disturbance, if casual, may be mediated by caretaking style. Ainsworth, Blehar,
Waters, and Wall (1978) f ound that avoidant attachment at 1 year was associated
with maternal aversion to physical contact, nonresponsiveness to signals such as
crying, and a lack of emotional expressiveness. In contrast, ambivalent attach-
ment was associated with inconsistency in responding to infants’ signals, rather
than simple low responding. Infants whose mothers responded to their cries in the
first quarter cried less in the fourth quarter. This is strictly counter to what social
learning theory would predict, since “reinforcement” of crying by responding
reliably would be expected to lead to increased rates of crying later. Positive
relationships between maternal responsivity and security of attachment have since
been found by other researchers (Bates, Maslin, & Frankel, 1985; Belsky, Rovine,
& Taylor, 1984) although Belsky et al. found no relation between security of
attachment and frequency of close physical contact with their mothers. Egeland
and Farber (1984) found that mothers of securely attached infants were more
sensitive, perceptive, and empathetic than others.
Not surprisingly, outright abuse or neglect also relates to attachment security.
Lamb, Gaensbauer, Malkin, and Schultz (1985) found that infants maltreated by
their mothers were more likely to develop insecure attachments with them and
with subsequent foster mothers, most often becoming avoidant; although children
abused by others demonstrated no greater than usual tendency toward insecure
attachment with their natural mothers. Others have also found a tendency for
abuse or neglect to lead to insecure attachment (Schneider-Rosen & Cicchetti,
1984; Schneider-Rosen, Braunwald, Carlson, & Cicchetti, 1985). Lamb et al.
(1985) have suggested that the overstimulation of chronic abuse may encourage
the development of avoidant patterns, while the understimulation of neglect
might lead to ambivalent attachment. Egeland and Sroufe (1981 a), however,
found that poor mothering was associated with different patterns over time. At 12
months such infants exhibited the ambivalent pattern, as though unsure how to
respond to the parent; while at 18 months avoidant attachment was more com-
mon, as though by then they had given up attempting to relate.

Summary. These studies underline a relation between maternal characteristics and

attachment classifications. Mothers of more securely attached infants appear to
be more confident, emotionally stable, and responsive to their infants’ attachment
behaviors than mothers of insecurely attached infants. Futher, mothers suffering
from mental illness and those who neglect or abuse their children tend to have
insecurely attached infants. These patterns do not follow from the view that infant
temperament is a major determinant of attachment. In summary, although infant
temperament may have a role in attachment, the research on stability, concor-
dance, and maternal influence suggests that attachment status primarily reflects
the quality of the relationship between mother and child.
 6.24 R. J. Paterson and G. Moran

But while the accumulated evidence calls the temperament position into serious
question, it also poses a problem for attachment theory. Evidence that attachment
status changes over time and differs by attachment figure appears to conflict with
the position that early attachment experience forms the basis of a unitary and
stable internal working model that determines the form of subsequent relation-
ships. While far from being resolved, some directions for approaching this issue
are discernable from the existing literature. Multiple changing attachments might
contribute to a single but very complex working model which is then applied
universally. Alternatively, different attachment experiences might create a num-
ber of distinct working models, which might differ in their influence on subse-
quent relationships depending on the importance of the experience and the type
of relationship involved.
Bretherton (1985) supports the view that the individual internalizes different
working models corresponding to different formative relationships. As an illustra-
tion she offers Cottrell’s (1969) case history of a man who seems to have based his
harsh actions toward his son on his stormy experience with his father, and his
loving relationship with his wife on his history with his mother. In a similar vein,
psychodynamic clinicians often view the patient’s behavior toward a male thera-
pist to be reminiscent of the past relationship with father, and that toward a female
therapist to be suggestive of the history with mother.
To summarize this position, the individual may internalize many early relation-
ships; their influence on an adult relationship may depend on the similarity
between the person concerned (or the position they hold) and the original proto-
type. In addition, stronger attachment relationships (such as that with mother)
may have more wide-ranging effects. Unfortunately, these points are difficult to
investigate and have not been the subject of much discussion in the theoretical
literature, despite their centrality. In consequence, the process of internalization,
the number and form of internal models, and the way these models influence
future behavior are all inadequately understood issues.

Attachment Status and Other Childhood Characteristics. Many studies have exam-
ined the relationship between attachment status and other concurrent infant char-
acteristics. Great caution must be exercised in inferring causality on the basis of
such evidence. Other studies have looked for relations between attachment status
and later social or cognitive characteristics. These studies are more relevant to the
hypothesis that early attachments become internalized and shape later develop-
ment. Evidence emerging from prospective designs cannot be accepted uncritical-
ly, however. Attachment status and later developmental characteristics may both
be products of a common third factor, such as maternal sensitivity or the quality
of the home environment. If this is the case, the temporal order of measurement
may provide only the illusion of causality. Nevertheless, this type of evidence is
currently the best available.
A variety of studies have assessed the cognitive correlates of secure attachment.
No differences in cognitive development across attachment categories are typically
found using the Bayley Developmental Scales (Matas, Arend, & Sroufe, 1978;
Pastor, 1981), although Main (1983) f ound that securely attached infants had
higher Developmental Quotients than insecure infants of the same age. Frodi,
Bridges, and Grolnick (1985) f ound that at 12 months ambivalent babies demon-
 Attachment Theory 625

strated less persistence on mastery-related play tasks than secure or avoidant

babies, and that avoidant babies demonstrated less overall competence than se-
cure or ambivalent babies. Neither of these differences were sustained at 20
months. Matas, Arend, and Sroufe (1978) o b served that securely attached infants
engaged in more symbolic play and were more enthusiastic in a problem-solving
task than insecure children. Studies have similarly indicated that securely at-
tached infants exhibit more sophisticated spontaneous play than avoidant or am-
bivalent infants (Belsky, Garduque, & Hrncir, 1984; Slade, 1987). Others find
that securely attached infants perform better on an object permanence task than
insecure groups (Ahmad & Worobey, 1984; Bell, 1970), though Levitt, Antonuc-
ci, and Clark (1984) f ound no group differences in awareness of object or person
permanence. Finally, secure infants have been found to recognize themselves in a
mirror both more readily (Schneider-Rosen & Cicchetti, 1984) and less readily
(Lewis, Brooks-Gunn, & Jaskir, 1985) than insecure infants.
More directly related to the predictions of attachment theory than general
cognitive development is competence at exploring the environment, since the
provision of a secure base by the attachment figure should promote exploration.
Cassidy (1986) f ound that secure infants made significantly fewer errors in nego-
tiating their environment during the Strange Situation than avoidant or ambiva-
lent infants, though deriving both attachment status and the competence measure
from the same videotapes makes interpretation difficult. Hazen and Durrett
(1982) examined exploration and cognitive mapping abilities in 28 children (who
had been classified at 12 months) at play in the laboratory when they were
between 30 and 34 months. Although the idiosyncratic grouping of infants ac-
cording to sub-categories of attachment makes comparisons with other studies
difficult, they found that secure infants were favored on five of eight measures;
however, on only two (one each of exploration and cognitive mapping) was this
group superior to both anxiously attached groups. Overall, the evidence suggests
that secure infants have a greater propensity to explore their environment than
their insecure counterparts.
From a clinical perspective, the most important predictions of attachment theo-
ry concern ego functioning and social relations, rather than cognitive abilities.
Securely attached infants should be more confident and show more adaptive and
harmonious relationships with others. Many studies have examined this area and
to a great extent the contention is supported. One paradigm examines the infant’s
response to adult strangers. Lamb et al. (1982) f ound that sociability toward a
stranger was highest among infants securely attached to father, but was unrelated
to attachment status with mother. Main and Weston (1981) reported that secure
attachments both with mother and with father predicted more positive interac-
tions with a clown at 12 months, but that the relationship to mother was the
strongest predictor. In a prospective study, Lutkenhaus, Grossmann, and Gross-
mann (1985) assessed attachment to mother at 12 months and interaction with an
adult visitor to the home at 3 years. Readiness to interact with the visitor was
lower among avoidant children than the other two groups, which did not differ.
Other studies have examined interactions with peers. Jacobson and Wille
(1986) observed that 18-month attachment status to mother did not predict socia-
bility or responsiveness to a securely attached peer at 2 or 3 years, but did relate to
behavior directed toward the focal children by the peer. At age 3 securely attached
children received more positive responses than insecure children. Avoidant chil-
dren received fewer positive initiations than others, while ambivalent children
received more disruptive responses, more combative initiations, and more resis-
tance to their own combativeness. Patterns at age two were generally weaker.
Pastor (1981) found that 20- to 23-month-old children who earlier had been
classified as securely attached scored higher on measures of sociability and orien-
tation toward a peer than did members of the two insecure groups. Secure chil-
dren also redirected their attention after a dispute over a plaything more often
than avoidant children. Finally, avoidant and secure children made more social
bids than ambivalent children, and ignored fewer sociaf bids made by their
playmates.
Several studies have evaluated social characteristics more generally. Main
(1983) Found secure infants assessed at I2 months to be more sociable and playful
in interactive play with an adult woman at 23 months, although the differences in
cognitive development noted earlier may confound these findings. Waters, Wipp-
man, and Sroufe (1979) classified infants at 15 months, then observed the chil-
dren at preschool over a 5-week period at 3.5 years. Five of 12 items regarding
confidence and effect.iveness favored the secure group, as did 11 of 12 measures of
the childrens competence in interactions with peers.
Erickson, Sroufe, and Egeland (1985; see also Sroufe, Fox, & Pancake, 1983)
assessed a sample of 96 children in the Strange Situation at 12 and 18 months,
then in a variety of preschool settings (including a special laboratory preschool for
40 of the children) at 4.5 to 5 years. Secure children were rated higher than
ambivalent children on a measure of confidence/assertiveness, and avoidant chil-
dren were rated higher than secure children an dependency. No differences were
found on measures of ego control, social skills, positive or negative affect, or
compliance, although a more restricted study of the laboratory preschool sample
found secure infants to have more positive affect, better social skills, more empa-
thy, and less dependency than anxiously attached infants (Sroufe et al., 1983). On
measures completed by teachers, avoidant infants were rated more withdrawn,
exhibitionistic and impulsive than secure infants, and gave up at tasks and cried
more easily. Avoidant infants in the subsample were described as more hostile and
isolated, while ambivaient infants were described as helpless and having low ego
controf. In contrast, Bates, Ma&n, and Franket (1985) found no relationship
between early attachment status and subsequent behavioral dif~~u~t~es in a study
of 3-year-olds who had undergone the Strange Situation at 13 months. They
attribute this to their use of a more stable, supportive sample of parents and to
having fewer insecurely attached children than in other studies. Their question-
naire method of assessment may also have been less sensitive than the use of
trained observers in situ.
En a study by I,ewis, Feiring, McCuffog, and Jaskir (1984), avoidant 6-year-old
boys (assessed at 12 months with an abbreviated Strange Situation) were rated
more schizoid on the Child Behavior Profile (CBP) than secure boys, while
ambivalent boys ranked higher on depression and social withdrawal. Both inse-
cure groups showed more uncommunicative behavior and somatic complaints
than secure boys. There were no group differences for girls. Using CBP scores it
was determined that risk for pathology was 40% for avoidant and ambivalent
males compared to 6% of secure males, while girls again showed no differences.
 Attachment Theory 627

Lewis et al. interpreted these figures to indicate that while early insecure attach-
ment is not sufficient in itself to produce pathology, these boys may be more
subject to difficulty when placed under environmental stress.
In summary, while existing evidence is somewhat ambiguous regarding the
relationship of attachment status to cognitive faculties, attachment may be rele-
vant to some narrowly defined areas, particularly those involving motivation and
confidence in cognitive tasks. In keeping with the central principle that attach-
ment figures serve to enable and encourage exploration, securely attached infants
appear to be more accomplished explorers than their insecure counterparts. Se-
cure infants also seem to be more amenable to social contacts, and as older
children are more competent at social interaction than their insecure counter-
parts. They seem to be less easily threatened, and may exhibit fewer behavior
disorders than avoidant or ambivalent children.
These findings emerge from studies examining functioning both concurrent
with Strange Situation classification and subsequent to it, up to 6 years. Observa-
tions of these children into the school years and beyond will be very telling.
Findings that early attachment status relates to later adaptation, even as the direct
influence of parents is in decline, would be far more compelling evidence of
attachment’s causal role than is currently available.
The Strange Situation has proven to be an important and useful research tool in
the examination of the relationship between the infant and its caretakers. Unfor-
tunately it is sharply limited, to an extent matched by few psychological instru-
ments in wide use, in the age range of subjects to which it can be applied. This
makes the assessment of shifting patterns of attachment over time very difficult. If
progress is to be made beyond examining the dynamics of attachment in early
childhood, then other reliable measures with greater flexibility and for use with
older populations must be developed. Promising directions include the recent
development of a Q-sort measure of security of attachment (Waters & Deane,
1985) and of a classification procedure that can be applied to 6-year-olds (Main,
Kaplan, & Cassidy, 1985). Until such approaches are applied more widely, clini-
cal histories and imperfect subjective measures will limit insight into the influence
of attachment on later life.
Despite limitations on the assessment of attachment beyond infancy, there
remain questions which must be asked of the theory as it applies to later childhood
and adulthood. The final issue to be addressed in the present section is the
evidence from studies concerning the internalization of the early attachment
relationship and its influence on later functioning.

ATTACHMENT INFLUENCES BEYOND EARLY CHILDHOOD

The research examining Strange Situation behavior and social functioning in

infants and children is consistent with the suggestion that early attachment rela-
tionships contribute to the individual’s representation of others. But, as noted
above, problems of assigning causality limit our conclusions. Proponents of this
aspect of attachment theory must rely on convergent evidence from multiple
perspectives. Although limited, evidence is available to suggest a link between
early attachment experience and later parenting, romantic relationships, and
depression.
628 R. J. Paterson and G. Moran

Attachment and Subsequent Parenting

Several studies have examined the relationship between women’s parenting behav-
ior (and, in some cases, the attachment status of their children) and their recollec-
tions of their own relationships with their mothers. Central to this approach is the
often-replicated finding that abusive mothers were frequently themselves abused
as children (Belsky, 1978, 1980; DeLozier, 1982; Spinetta & Rigler, 1972). At first
glance, a simple modeling process seems the most parsimonious explanation of
these observations. Attachment theory does not provide an obvious account of
how early aversive interaction would cause individuals to re-enact the abusive
actions of their own parents. However, Sroufe and Fleeson (1986) argue that in
any relationship the individual learns not only his or her own role, but also the
role of the other person: both sides of the relationship are introjected. Thus, the
abused child comes to understand relationships as involving an abuser and an
abused, and structures (or understands) subsequent relationships in a like man-
ner. The role played may depend on the situation. In parenthood, the role of the
abused is occupied by the infant and so the role of abuser is adopted. Role
selection may not always be guided by context, however. Bowlby (1973) and
Bacciagaluppi (1985) d escribc situations in which mothers with overwhelming
dependency needs attempt to use their children as attachment figures, effectively
taking the child role and thrusting the parent role on their children.
Main (Main, Kaplan, & Cassidy, 1985; Main & Goldwyn, 1984) has developed
the Adult Attachment Interview as a tool to investigate attachment issues in
adulthood. This structured interview involves both a retrospective examination of
early attachments and an assessment of current attitudes toward attachment.
Initial research has revealed a striking relation between classifications of parents
based on this instrument and the Strange Situation categorization of their chil-
dren. Parents of securely attached children (at 12 and 18 months, and 6 years of
age) were comfortable with and valued attachments in adulthood. Recollections of
relationships with their own parents were not uniformly positive, but were rcalis-
tic and not idealized. Importantly, the representation of the relationship, rather
than specific early experiences, appeared to be the factor that distinguished them
from parents of less securely attached children.
Parents rated as insecure fell into three categories. One group was very dismis-
sive of the importance of attachments and tended to be parents of insecure-
avoidant children (A). A second group was preoccupied with and often idealized
their relationships with their own parents. These parents most frequently had
children categorized as insecure-ambivalent (C). A final group of parents had
experienced the death of an attachment figure in childhood and had not success-
fully resolved this loss. Such individuals were most frequently parents of insecure-
disorganized/disoriented infants (D). Overall, Main et al. (1985) reported a sig-
nificant correlation (r=.62) between infant classification and the mother’s work-
ing model of attachment as assessed using the Adult Attachment Interview. These
results provide perhaps the first empirical validation of the influential hypothesis
that the parent’s representations of past attachments influence behavior with their
own children and, ultimately, the form of the new parent-child relationship.
Ricks (1985) f ound that mothers of secure infants rated their own mothers
significantly higher on acceptance and encouragement of their independence than
mothers of insecure infants, and significantly lower on overprotection and re_jec-
 Attachment Theory 629

tion. Aggregate ratings of maternal encouragement and acceptance were also

higher for mothers of secure infants. In addition, these mothers rated their fathers
more favorably on encouragement of independence, acceptance, and both aggre-
gate scales, but showed no differences on overprotection or rejection. No differ-
ences on rated idealization of the mother’s mother were found, but mothers of
secure infants tended to idealize their fathers more than mothers of insecure
infants. In a followup study (1985), mothers of insecure infants were found to be
more defensive and idealized both parents more than mothers of secure infants.
Sroufe and Fleeson (1986) d escribe a study by Morris (1980) in which 9 of 36
mothers exhibited seductive behavior toward their infants, defined as fondling,
sensual teasing, kissing on the lips, and so on. This behavior was directed almost
entirely toward sons. Of the nine mothers, eight were identified as having been
the subject of emotional manipulation or incest by their fathers. Only 4 of 12 such
victims failed to act in a seductive manner toward their children; of these, three
were observed only with daughters. Sroufe and Fleeson argue that modelling does
not appear to be an adequate explanation for these findings, as the sex of child
and the specific behaviors engaged in by the mothers and their fathers differed.
Instead, these women imitated the sexual nature of their earlier relationships
while tailoring the behaviors and selection of subject to the female sex-role. This
suggests that it was the pattern of relating, not the behavior, that was learned in
the earlier relationship.
Although scarce, the available research suggests that early experience influ-
ences adult parenting behavior. The prospective mother seems to introject early
relationships and uses these as templates for relationships with her children. This
conclusion rests on retrospective data, however, and maternal recollections may
be inaccurate. Reports may be biased, perhaps in part by the salience of the
present relationship with the infant, creating spurious correlations. Recall bias
might be particularly attractive to abusive parents, who might find in such strate-
gies justifications or rationales for their own behavior.
Helpful support for the attachment view comes from studies in which the past
relationship variables were externally verifiable or, at least, were reasonably dis-
crete and therefore perhaps less subject to retrospective distortion than ratings of
parental style. Frommer and O’Shea (1973a, 1973b) found that mothers who
reported that they had been separated from either parent before the age of 11
exhibited difficulties in parenting, marital problems, and depression. Hall and
Pawlby (1981) f ound that women from disrupted families seemed less involved
with their babies. Finally, Rutter, Quinton, and Liddle (1983) observed that
mothers whose children had been taken into government care on different occa-
sions were more than three times as likely as matched controls to have been in
government care in childhood themselves, and were also far more likely to have
been separated from their parents due to apparent conflict or rejection.

Attachment and Subsequent Romantic Relationships

The available literature on attachment as it relates to romantic and peer relation-

ships is more limited. Weiss (1982) notes that some relationships between adults
(most commonly romantic relationships and relationships in situations of extreme
stress) seem to satisfy the requirements of attachment bonds: the individual is
motivated to remain within range of the attachment figure, the presence of
630 R. .J. Paterson and G. Moran

the figure is accompanied by increased comfort and security, and separations or

threats to access to the figure may be perceived as threats to the individual’s own
well-being and may be met with protest. On the other hand, Weiss observes, adult
relationships differ from childhood attachment in at least three ways: these rela-
tionships are with peers, not older and wiser caretakers; threats to continued
access do not overwhelm other behavior as easily as in infants; and in the context
of romantic relationships there is the added explicit component of sexuality. In a
subsequent article Weiss (1986) presents a typology of bonds in adulthood and
points out that only a few qualify even as approximations of attachment relation-
ships. It should be noted, however, that the internal working models developed in
childhood may influence even relationships in which attachment behavior is not
elicited.
Empirical research on this topic is noticeably sparse. Hetherington (1972)
found that adolescent girls who had lost a father to divorce sought more attention
from an adult male-perhaps demonstrating a dependent, or anxiously attached,
style-than did girls without such loss. Girls who had lost a father through death
were more avoidant of an adult male than the other two groups. Hepworth,
Ryder, and Dreyer (1984) examined courtship patterns among college students
who had or had not experienced recent parental loss. They observed a trend
among the loss sample toward either accelerated courtship or to increased hesitan-
cy regarding close relationships relative to the control sample. As with
Hetherington, those with a loss due to death more often avoided close relation-
ships, while those with a loss due to divorce had accelerated courtship and more
interest in relationships. The recency of these losses (generally within the previous
5 years) suggests that the effects might bc attributable to residual mourning rather
than formative influences on internal working models, although such influcnccs
may not end in early childhood.

Attachment and Subsequent Psychopathology

Research has also related depression and suicidal behavior and ideation to early
childhood experiences with caretakers. As described above, Frommer and O’Shea
(1973b) found that early separation predicted self-reported depression. Lindeman
(1960) found that patients with early separation also had high rates of both
suicidal ideation and depressive episodes. In a sample of 79 women whose moth-
ers had died while the subjects were children, Parker and Hadzi-Pavlovic (1984)
found that subsequent lack of caring by fathers and stepmothers significantly
predicted high trait depression in adulthood. Tennant, Hurry, and Bebbington
(1982) found that while none of four types of parental separation from birth to age
four predicted psychological morbidity in adulthood, several measures of separa-
tion at age 5-10 predicted both overall morbidity and depression, as did separa-
tion due to parental illness from age 11 to 15.
With university students, Adam (1982) f ound that parental loss due to death
and divorce was strongly related to later self-reported suicidal ideation and suicide
attempts. A subsequent study revealed family instability to be a significant factor
related to loss, with greater instability in the histories of subjects who had at-
tempted suicide. Adam concludes that the literature on loss in childhood supports
a predictive link to later attempted suicide, although the type of loss and age at
loss associated with greatest vulnerability remain a matter of debate.
 Attachment Theory 631

What little evidence exists suggests that early attachment status and subsequent
loss influence later psychopathology and behavior in relationships. The studies
available are inevitably flawed, however, and there is not the breadth of evidence
from different perspectives utilizing different paradigms that would be necessary
to offset this shortcoming. Although the evidence is substantially supportive of
this critical aspect of attachment theory, the point cannot be said to have been
established with the rigour one would prefer as a basis for clinical intervention.

SUMMARY

Attachment theory has spawned a large and ever expanding body of research,
most of it concerned with functioning in infancy. A number of the fundamental
propositions of the theory have been supported. The available data suggest that
human infants have a primary need for social interaction of some kind, perhaps
specifically involving an exclusive relationship with a single caretaker perceived as
“older and/or wiser.” Not supported by the data (and not a part of Bowlby’s
original theory) is the concept of a critical period soon after birth for the formation
of the mother’s bond to the child. The attachment status of the child appears to
depend, as would be predicted, on the personality and caretaking style of the
mother, which typically seem to remain constant in the absence of unusual
stresses. The infant’s attachment pattern may differ with different caretakers.
Caretaker variables associated with secure attachment in the infant are psycho-
logical health, responsiveness to the infant’s attachment cues, and the absence of
abuse or neglect. Secure attachment appears to promote certain limited aspects of
cognitive development and more wide-ranging aspects of social development,
with securely attached infants being more sociable and eliciting more positive
responses from others. Finally, early attachment relationships may contribute to a
more adaptive model of the self in relation to others, both in caretaking situations
and close adult relationships, though few studies have investigated the latter. In
addition to gaps in the research literature, a major difficulty is a lack of consensus
regarding the issue of how and when internal working models are formed and
altered.
The lack of direct evidence in support of the influence of early attachment on
later interpersonal functioning is not accompanied by a body of research search-
ing for such evidence and failing to find it. The difficulty lies in formulating and
carrying out practical research designs that can adequately test the theory. Nor
can it be said that the theory is untestable by its very nature. Practical and ethical
concerns, however, limit the types of study which can be undertaken. This repre-
sents one of the many instances in clinical practice in which we may be forced to
content ourselves with indirect evidence from a number of different sources,
perspectives, and paradigms. Where this is available the tenets of attachment
theory are supported to a degree. We may never have direct evidence that internal
working models developed in early life form the basis of adult relationships.
Instead we may have to assess attachment theory from a practical perspective,
determining with each client whether it provides a useful model of adult
functioning.

Acknowledgements-Preparation of this manuscript was supported by a Medical Research

Council of Canada Doctoral Studentship to the first author, Natural Sciences and Engi-
632 R. IJ. Paterson and G. Moran

neering Research Council of Canada Grant A7046 to the second author, and Medical
Research Council of Canada Operating Grant MA5028 to Richard W. J. Neufeld.
The authors thank Joel Herscovitch and Richard W. ,J. Neufeld for their helpful com-
ments on an earlier draft of this article.

REFERENCES

Adam. K. S. (1982). I,oss, suicide, and attachment. In C. M. Parkes &J. Str\rnson-Hinde (Eds.),
7%~ plucf r?f’athchmmt zn human behacior (11~. 2699294). New York: Basic Books.
Ahmwd. A, & Worobey, ,J. (1984). Attachment and cognition in a naturalistic context. Child S’tu&
Journal, 14. 185-203.
Ainsworth, M. D. S. (1982). Attachment: Retrospect and prospect. In C. M. Parkcs &J. Stcvcn-
son-Hinde (Eds.), Theplace ofattachm~~nt in human hehaG (pp. 3-30). New York: Basic Books.
Ainsworth, M. D. S., Blehar. M. C., Waters, E., & Wall. S. (1978). Pat!em.~ ~faattnchment: A
psycholqical Jtudy ofthe Strange Sztuation. Hillsdale, NJ: Earlbaum.
Ainsworth. M. D. S., & Wittig. B. A. (1969). Attachment and exploratory behavior ofone year olds
in a strange situation. In B. M. Foss (Ed), Determinants of infant beha~~zor(Vol. 4, pp. 11 l-136).
London: Mcthuen.
.4mcrican Psychiatric Association. (1987). Diqmrtic and rt&tical manual rf m~ntul disorders (3rd ed.,
revised). Washington, DC: Author.
Bacciagaluppi, M. (1985). I nversion of parent-child relationships: A contribution to attachment
theory, British~/ournal ofMedical Psycholqy. 58, 369-373.
Bates. <J. IX., Maslin, C. A., & Frankel, K. A. (1985). Attachment security, mother-child interac-
tion. and temperament as predictors of behavior-problem ratings at age three years. Monqraph.~ of
the Societyfor Research in Child Dezaelopment, 50. 167-193.
Bell. S. M. (1970). The development of thr concept of object as rclatcd to infant-mother attach-
ment. Child Deuelopm&, 41, 291-31 1.
Belsky, J. (1978). Three theoretical mod& of child abuse: A critical review. ChildAbut and 1Vegiect. 2,
37-49.
Belsky. J. (1980). Child maltreatment: An ecological integration. Am~rzcnn P+ycho/ogict, 35, 32%
335.
Belsky, J.. Garduque. L., hi Hmcir, E. (1984). Assessing performance, compctcncc’. and executive
capacity in infant play: Relations to home environment and security of attachment. Droelopmental
Psychology. 20. 406-417.
Bclsky, J,, Ravine, M, & Taylor, D. G. (1984). l‘he Pennsylvania Infant and Family Development
Project, III: The origins of individual diffcrcnces in infant-mother attachment: hlaternal and
infant contributions. Child Development, 55, 718-728.
Berm, R. K. (1986). Factors promoting secure attachment relationships between employed mothers
and their sons. Child Development. 57, 12241231.
Bowlby, J. (1957). An ethological approach to research in child dcvclopment. HrztithJournul ufMedicn1
PJychology, 30, 230-240.
Bowlby, J. (1958). Psychoanalysis and child cart. In .J, D. Sutherland (Ed.), P!ychonna&.si~ and
contempora~ thqht (pp. 33-57). London: Hogarth Press.
Bowlby. J. (1961). Childhood mourning and its implications for psychiatry. ilvvrican Journc~l OJ
P~_ychiat~, 118, 481-498.
Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. New York: Basic Books.
Bowlby, J. (1973). Attachment and 1~s~. Ml. 2. Separation: Anxiety and Arqer. New York: Basic Books.
Bowlby, ,J. (1977a). The making and breaking of affcctional bonds. I. Aetiology and psychopatholo-
gy in the light of attachment theory. British Journal ofPsych+y, 130, 201-210.
Bowlby, J. (1977b). The making and breaking of affectional bonds. II. Some principles of psycho-
therapy. Rritish~/ournal ofP~ychiatry, 130, 421-431.
Bowlhy, J, (1980). Attarhment and 10s~. Vol. 3. Loss, sadness, and depression. New York: Basic Books.
Bowlby, J. (1984). Attachment and loss: Vol. 1. Attachment (rev. cd.). New York: Basic Books.
Bowlby, J., & Parkes, C. M. (1970). Separation and loss within the family. In E. .J. Anthony & C.
Koupernik (Eds.), The child in hisJ&ni!y (Vol. 1, pp. 197-216). N ew York: John Wiley & Sons.
Bretherton, I. (1985). Attachment theory: Retrospect and prospect. ;[email protected] of thr SocieQ for
Research m Child Development, 50, 3-35.
 Attachment Theory 633

Call, J. D. (1984). Child abuse and neglect in infancy: Sources of hostility within the parent-infant
dyad and disorders of attachment in infancy. Child Abuse and Neglect, 8, 185-202.
Cassidy, J. (1986). The ability to negotiate the environment: An aspect of infant competence as
related to quality of attachment. Child Development, 57, 331-337.
Clark, E. A., & Hanisee, J. (1982). Intellectual and adaptive performance of Asian children in
adoptive American settings. Developmental Psych&Q, 18, 595-599.
Coates, B., Anderson, E. P., & Hartup, W. W. (1972). Interrelations in the attachment behavior of
human infants. Developmental Psycholou, 6, 218-230.
Cottrell, L. S., Jr. (1969). Interpersonal interaction and the development of the self. In D. A. Goslin
(Ed.), Handbook of socialization theory and research (pp. 543-570). Chicago: Rand-McNally.
D’Angelo, E. J. (1986). Security of attachment in infants with schizophrenic, depressed, and
unaffected mothers. Journal of Genetic Psychology, 147, 421-422.
DeLozier, P. (1982). Attachment theory and child abuse. In C. Parkes &J. Stevenson-Hinde (Eds.),
The place of attachment in human behavior (pp. 95-l 17). New York: Basic Books.
Dennis, W. (1973). Children of the creche. New York: Appleton-Century-Crofts.
Egeland, B., & Farber, E. A. (1984). Infant-mother attachment: Factors related to its development
and changes over time. Child Development, 55, 753-771.
Egeland, B., & Sruofe, L. A. (1981a). Attachment and early maltreatment. Child Deuelopnent, 52,
44-52.
Egeland, B., & Sroufe, L. A. (1981b). Developmental sequelae of maltreatment in infancy. In R.
Rizley & D. Cicchetti (Eds.), Development&perspectives on child maltreatment (pp. 77-92). San Francis-
co: .Jossey-Bass.
Egeland, B., & Vaughn, B. (1981). Failure of “bond formation” as a cause of abuse, neglect, and
maltreatment. American Journal of Orthopsychiatry, 51, 78-84.
Erickson, M. F., Sroufe, L. A., & Egeland, B. (1985). The relationship between quality of attach-
ment and behavior problems in preschool in a high-risk sample. Monographs of the So&y for Research
in Child Development, 50, 147-166.
Frodi, A., Bridges, L., & Grolnick, W. (1985). C orrelates of mastery-related behavior: A short-term
longitudinal study of infants in their second year. Child Deuelopment, 56, 1291-1298.
Frommer, E. A., & O’Shea, G. (1973a). Antenatal identification of women liable to have problems
in managing their infants. British Journal of Psychiatry, 123, 149-156.
Frommer, E. A., & O’Shea, G. (197313). The importance of childhood experience in relation to
problems of marriage and family building. British Journal ofpsychiatry, 123, 157-160.
Gaensbauer, T. J., Harmon, R. J., Culp, A. M., Schultz, L. A., van Doorninck, W. J., & Dawson,
P. (1985). Relationships between attachment behavior in the laboratory and the caretaking
environment. Infant Behavior and Development, 8, 355-369.
Gomes-Pedro, J., de Almeida, J. B., da Costa, C. S., & Barbosa, A. (1984). Influence of early
mother-infant contact on dyadic behaviour during the first month of life. Developmental Medicine and
Child Neumlou, 26, 657-664.
Grossmann, K., & Grossmann, K. (1981). P arent-infant attachment relationships in Bielefeld. In
K. Immelmann, G. W. Barlow, L. Petrinovich, & M. Main (Eds.), Behavioral development: The
Bielejld interdisciplinary pro&t (pp. 694-699). N ew York: Cambridge University Press.
Grossmann, K. E., Grossmann, K., Huber, F., & Wartner, U. (1981). German children’s behavior
towards their mothers at 12 months and their fathers at 18 months in Ainsworth’s Strange
Situation. International Journal ofBehavioral Development, 4, 157-181.
Hall, F., & Pawlby, S. (1981). Continuity and discontinuity in the behavior of British working-class
mothers and their first-born children. International Journal of Behavioral Development, 4, 13-26.
Harlow, H. F. (1958). The nature of love. American Psychologist, 13, 673-685.
Harlow, H. F. (1971). Learning to loue. San Francisco: Albion.
Hazen, N. L., & Durrett, M. E. (1982). Relationship of security of attachment to exploration and
cognitive mapping abilities in 2-year-olds. Developmental Psychology, 18, 751-759.
Hepworth, J., Ryder, R. G., & Dreyer, A. S. (1984). The effects of parental loss on the formation of
intimate relationships. Journal of Marital and Family Therapy, 10, 73-82.
Hetherington, E. M. (1972). Effects of father absence on personality development in adolescent
daughters. Developmental Psychology, 7, 313-326.
Jacobson, J. L., & Wille, D. E. (1986). The influence of attachment pattern on developmental
changes in peer interaction from the toddler to the preschool period. Child Development, 57, 338-
347.
634 R. J. Paterson and G. Moran

Jensen, G. D. (1965). Mother-infant relationship in the monkey Macaca nemestrina: Development of

specificity of maternal response to own infant. Journal ofComparative
and Physiologtcal Psychology, 59,
305-308.
Kagan, J. (1982). Psychological rtwarch on the human infant: An evaluative summary. New York: W. T. Grant
Foundation.
Kennell, J. H., Jerauld, R., Wolfe, H., Cheslcr, D., Kreger, N. C., McAlpine, W., Steffa, M., &
Klaus, M. H. (1974). Maternal behavior one year after early and extended post-partum contact.
Developmental Medicine and Child Neurology, 16, 172-I 79.
Klaus, M. H., Jerauld, R., Kreger, N. C., McAlpine, W., Steffa, M., & Kennell, J. H. (1972).
Maternal attachment: Importance of the first post-partum days. New England Journal of Medicine,
286, 460-463.
Klaus, M. H., & Kennell, J. H. (1976). Matxrnal-inznt bonding. St. Louis, MO: Mosby.
Klaus, M. H., & Kennell, J. H. (1982). P arent-infant bonding. St. Louis, MO: Mosby.
Lamb, M. E. (1974). A defense of the concept of attachment. Human Development, 17, 376-385.
Lamb, M. E. (1977). Father-infant and mother-infant interaction in the first year of life. Child
Development, 48, 167-181.
Lamb, M. E. (1978). Qualitative aspects of mother- and father-infant attachments. In@ Behavior
and Development, 1, 265-275.
Lamb, M. E. (1982). Early contact and maternal-infant bonding: One decade later. Pedzatrik, 70.
763-768.
Lamb, M. E., Gaensbauer, T. J., Malkin, C. M., c(r Schultz, L. A. (1985). The effects of child
maltreatment on security of infant-adult attachment. Zn$znt Behavior and Drvelopmcnt, 8, 35-45.
Lamb, M. E., Hwang. C. P., Frodi, A. M., & Frodi, M. (1982). Security of mother- and father-
infant attachment and its relation to sociability with strangers in traditional and nontraditional
Swedish families. Inyant Behavior and Development, 5, 355-367.
Lamb, M. E., Thompson, R. A., Gardner, W., & Charnov, E. I,. (1985). Infant-motheratlachrnent. The
ori+ and developmental .signiffcance of indizjidual dzyferrntes in \tran,Fe situation behavior. Hillsdalc. NJ:
Erlbaurn.
I,amb, M. E., Thompson, R. A., Gardner, W. P., Charnov, E. I,., & Estes, D. (1984). Security 01
infantile attachment as assessed in the Strange Situation: Its study and biological interpretation.
Behavioral and Brain Science,<, 7, 12 7- 147.
Levitt, M. J., Antonucci, T. C., Br Clark, M. C. (1984). Object-person permanence and attach-
ment: Another look. Merrill-Palmer Quarterly, 30, l-10.
Levy-Shiff, R. (1983). Adaptation and competence in early childhood: Communally-raised kibbutz
children versus family raised children in the city. Child Deuelopment. 54, 1606-1614.
Lewis. M,, Brooks-Gunn, _J,, hi Jaskir, _J. (1985). Individual differences in visual self-recognition as
a function of mother-infant attachment relationship. Developmental Pjycholo~u, 21, 1181-l 187.
Lewis, M., Feiring, C., McGuffog, C., & ,Jaskir, ,J, (1984). Predicting psychopathology in six-ycar-
olds from early social relations. Child Development, 55, 123-l 36.
Lindeman, E. (1960). Psychosocial factors as stress agents. In ,J. M. Tanner (Ed.), Stre.\\ and
psychiatrtc disorder (pp. 112-137). Oxford: Blackwell.
Lorenz, K. (1952). King Solomon? ring. London: Methuen.
Lutkenhaus, P., Grossmann, K. E., & Grossmann, K. (1985). Infant-mother attachment at twelve
months and style of interaction with a stranger at the age of three years. Child Development, 56,
1538-1542.
Main, M. (1983). Exploration, play, and cognitive functioning related to infant-mother attach-
ment. Infant Behavior and Development, 6, 167-t 74.
Main, M., & Goldwyn, R. (1984). Predicting rejection of her infant from mother’s representation
of her own experience: Implications for the abused-abusing intergenerational cycle. Child Abuse
and Neglect, 8, 203-217.
Main, M., Kaplan, N., & Cassidy, J. (1985). S ecurity in infancy, childhood, and adulthood: A
move to the level of representation. MonographA ojthr Societyfor Research in Child Development, 50, 66-
104.
Main, M. B., c(r Weston, D. R. (1981). The quality of the toddler’s relationship to mother and to
father: Related to conflict behavior and the readiness to establish new relationships. Child Develop-
ment, 52, 932-940.
Marvin, R. S. (1977). An ethological-cognitive model for the attenuation of mother-child attach-
ment behavior. In T. Alloway, P. Pliner, & L. Kramcs (Eds.). AdDanceJ in the study qfrommunkation and
q&t: W .7. Attachment behavior (pp. 25-60). N <w York: Plenum Press.
 Attachment Theory 635

Matas, L., Arend, R. A., & Sroufe, L. A. (1978). Continuity of adaptation in the second year: The
relationship between quality of attachment and later competence. Child Development, 49, 547-556.
Morris, D. (1980). Infant attihment and problem solvinE in the toddler: Relations to mother’s farnib history.
Unpublished doctoral dissertation, University of Minnesota.
Myers, B. J. (1984). Mother-infant bonding: The status of this critical-period hypothesis. Develop-
mental Review, 4, 240-274.
Naslund, B., Persson-Blennow, I., McNeil, T., Kaij, L., & Malmquist-Larsson, A. (1984). Off-
spring of women with nonorganic psychosis: Infant attachment to the mother at one year of age.
Acta Psychiatrica Scandinavia, 69, 231-241.
Owen, M. T., & Chase-Lansdale, L. (1982, April). Similarity between infant-mother ana’ infant-&her
attachments. Paper presented at the meeting of the Southwestern Society for Research in Human
Development, Galveston, TX.
Parker, G. & Hadzi-Pavlovic, D. (1984). Modification of levels of depression in mother-bereaved
women by parental and marital relationships. Psycholo&zl Medicine, 14, 125-135.
Parkes, C. M. (1972). Bereavement.- Studies of grief in adult life, New York: International Universities
Press.
Parkes, C. M. (1982). Attachment and the prevention of mental disorders. In C. M. Parkes &J.
Stevenson-Hinde (Eds.), The place of attachment in human behavior (pp. 295-309). New York: Basic
Books.
Parkes, C. M. (1985). Bereavement. British JournalofPsychtitry, 146, 11-17.
Pastor, D. L. (1981). The quality of mother-infant attachment and its relationship to toddlers’
initial sociability with peers. Developmental Psychology, 17, 326-335.
Pollitt, E., Eichler, A. W., & Chan, C. K. (1975). P sy c h asocial development and behavior of
mothers of failure-to-thrive children. American Journal of Orthopsychiatry, 45, 525-537.
Provence, S., & Lipton, R. C. (1962). Infants in institutions. New York: International Universities Press.
Radke-Yarrow, M., Cummings, E. M., Kuczynski, L., & Chapman, M. (1985). Patterns of
attachment in two- and three-year-olds in normal families and families with parental depression.
Child Development, 56, 884-893.
Raphael, B. (1982). The young child and the death of a parent. In C. M. Parkes &J. Stevenson-
Hinde (Eds.), Theplace ofattachment in human behavior (pp. 131-150). New York: Basic Books.
Raphael, B. (1983). The anatomy of bereavement. New York: Basic Books.
Ricks, M. H. (1985). The social transmission of parental behavior: Attachment across generations.
Monographs of the Socieyfor Research in Child Development, 50, 2 1 l-227.
Ringler, N., Kennell, J. H., Jarvella, R., Navojosky, B. J,, & Klaus, M. H. (1975). Mother-to-child
speech at 2 years: Effects of early postnatal contact. Journal of Pediatrics, 86, 141-144.
Ringler, N., Trause, M. A., Klaus, M., & Kennell, J. (1978). The effects of extra postpartum
contact and maternal speech patterns on children’s IQs, speech, and language comprehension at
five. Child Development, 49, 862-865.
Rogers, C. (1951). Client-centired therapy. Boston: Houghton Mifflin.
Rutter, M. (1979). Maternal deprivation, 1972-1978: New findings, new concepts, new approach-
es. Child Development, 50, 283-305.
Rutter, M., Quinton, D., & Liddle, C. (1983). Parenting in two generations: Looking backwards
and looking forwards. In N. Madge (Ed.), Families at risk (pp. 60-98). London: Heineman.
Sagi, A., Lamb, M. E., Lewkowicz, K. S., Shoham, R., Dvir, R., & Estes, D. (1985). Security of
infant-mother, father, and metapelet attachments among kibbutz-reared Israeli children. Mono-
graphs oftheSociety for Research in Child Development, 50, 257-275.
Schneider-Rosen, K., Braunwald, K. G., Carlson, V., & Cicchetti, D. (1985). Current perspectives
in attachment theory: Illustration from the study of maltreated infants. Monographs ofthe Societyfor
Research in Child Development, 50, 194-210.
Schneider-Rosen, K., & Cicchetti, D. (1984). The relationship between affect and cognition in
maltreated infants: Quality of attachment and the development of visual self-recogmtion. Child
Development, 55, 648-658.
Slade, A. (1987). Quality of attachment and early symbolic play. Developmental PsycholoQ, 23, 78-85.
Spinetta, J. J., & Rigler, D. (1972). The child-abusing parent: A psychological review, Psychological
Bulletin, 77, 296-304.
Spitz, R. A. (1945). Hospitalism: An inquiry into the genesis of psychiatric conditions in early
childhood. Psychoanalytic Study of the Child, 1, 53-74.
Sroufe, L. A. (1985). Attachment classification from the perspective of infant-caregiver relation-
ships and infant temperament. Child Development, 56, l-14.
636 R. J. Paterson and G. Moran

Sroufe, L. A., & Fleeson, J. (1986). Attachment and the construction of relationships. In W. W.
Hartup & Z. Rubin (Eds.), Relationships and development (pp. 51-71). Hillsdale, NJ: Erlbaum.
Sroufe, L. A., Fox, N. E., & Pancake, V. R. (1983). Attachment and dependency in developmental
pcrspectivc. Child Deuelopmnt, 54, 1615-1627.
Stephen, T. D. (1984). Symbolic interdependence and post-break-up distress: A reformulation of
the attachment construct. Journal ojDiwrce, 8, l-16.
S\rejda, M. J,, Campos, J, J., & Emde, R. N. (1980). Mother-infant “bonding”: Failure to general-
ize. Child Deuelopment, 51, 775-779.
Tennant, C., Hurry, J., & Bebbington, P. (1982). l’he relation of childhood separation experiences
to adult depressive and anxiety states. Briti.ch Journal ofPsychiatry, 141, 475-482.
Thompson, R. A., Lamb, M. E., & Estes, D. (1982). Stability of infant-mother attachment and its
relationship to changing life circumstances in an unselected middle class sample. Child Dewlop-
Tent, 53, 144-148.
Tibbits-Kleber. A. L., 6i Howell, R. J. (1985). Reactive attachment disorder of infancy (RAD).
&urnal of Clinical Child Pgxhologp, 14, 304-3 10.
rlionick, E. Z., Ricks, M., & Cohn, J. F. (1982). Maternal and infant affective exchange: Patterns
of adaptation. In T. Field & A. Fogel (Eds.), Emotion and ear!y interaction (pp. 83-100). Hillsdalc,
N,J: Erlbaum.
T urco, R. (1982). The influence of previous life events and dwelopmcntal cxperirnces on the
resolution of grief as dwreloped in psychotherapy. Psychiatric Qurterb, 54, 143-148.
Vaughn, B., Egeland, B., Sroufe. L. A., & Waters, E. (1979). Individual diffcrcnces in inf;tnt-
mother attachment at twelve and eighteen months: Stability and change in families under sttws.
Child Deuelopment, 50, 971-975.
Waters, E., & Deane, K. E. (1985). Defining and assessing individual differences in attachment
relationships: Q-methodology and the organization of behavior in infancy and early childhood.
Monographs oj the SocieQ for Research in Child Deuelopment, 50, 4 l-65.
Waters, E., Wippman, J,, & Sroufe, L. A. (1979). Attachment, positive affect, and competence in
the peer group: Xvo studies in construct validation. Child Development, 50. 821-829.
Weiss, R. S. (1982). Attachment in adult lift. In C. M. Parkes &J. Stevenson-Hindc (Ed.), Theplace
qfattachment in human behavior (pp. 171-184). New York: Basic Books.
Weiss, R. S. (1986). Continuities and transformations in social relationships from childhood to
adulthood. In W. W. Hartup & Z. Rubin (Eds.), Relationships and dewlopment (pp. 95-l 10). Hills-
dale, N,J: Erlbaum.