In Development and Psychopathology, 1992, 4, 209-241.
Quality of Attachment in the Preschool Years
Patricia M. Crittenden
University of Miami
P.O. Box 016820(D-820)
Miami, FL 33101
Running head: Preschool attachment
I would like to thank Dante Cicchetti for his support and assistance throughout the writing of this
paper. In addition, I wish express appreciation to Robert Marvin and Mary Ainsworth as well as
the members of my research team, Mindy Cassell, Angelika Claussen, Sharon Heller, Michelle
Jean-Gilles, Claudia Lang, and April Vogel, for their contributions to the thinking developed
here and for their helpful comments during the development of this manuscript.
This is a theoretical paper about differences in quality of attachment in preschool-aged
children with emphasis on the development of the goal-corrected partnership. Inferences are
made about the processes underlying preschoolers' attachment behavior. Specifically, the notion
of quality of attachment is expanded to explicitly include strategy, regulation of affect,
negotiation, secure base behavior, and response to maternal behavior. The classificatory system
is expanded by adding two additional defended patterns, i.e., compulsive caregiving and
compulsive compliance, to the infant avoidant pattern. Furthermore, at the preschool age, the
infant ambivalent pattern is identified as having a coercive strategy. In addition, the disorganized
infant category is reconceptualized in terms of complex organization, reorganization, and
disorganization. Finally, the process of generating new theories and hypotheses through a
"participant observer" methodology is considered from the perspective of developmental
Quality of attachment has been a unifying construct for research on infants and one which
predicts differences in developmental pathways from infancy to the preschool and early school
years (Cicchetti, 1990; Main, Kaplan, & Cassidy, 1985; Sroufe, 1983). The construct has not yet,
however, been applied to the preschool years with as thorough a developmental perspective as is
available for infancy. Instead, the emphasis has been on identifying continuity of patterns from
infancy to the preschool years and beyond (Ainsworth, 1990). The understanding of the meaning
of such continuity awaits further development of theory, particularly of how attachment
functions in the preschool years and how its function is integrated into the development of the
preschool child (Cicchetti, Cummings, Greenberg, & Marvin, 1990).
There are several issues to be considered regarding quality of attachment in preschool-aged
children. The first is defining quality of attachment in terms that are relevant to the preschool
period. The second is incorporating developmental change into the aspects of behavior used to
assess quality of attachment in the preschool years. The third is grouping together children
whose quality of attachment, i.e., underlying strategy, is similar. This may involve both
identifying new patterns of behavior and also combining into one classification, i.e., quality,
patterns of behavior which fulfill the same strategy. The fourth is determining appropriate
situations with which to evoke attachment behavior and identifying the critical features of those
situations. The fifth is ensuring the development of classificatory systems sensitive to the range
of behavior shown by children who differ in culture, upbringing, and risk status. Because middle
class children are not representative historically or currently of most human children, the
development of a broader range of children must be incorporated in any universal theory. In
addition, the behavior of children experiencing abnormal physiological or psychological
conditions (roughly a quarter of all children) is necessary to describe the range of human
behavior as well as to elucidate aspects of development not easily observed under more
normative conditions (Cicchetti & Toth, 1991; Crittenden, 1990; Marvin & Pianta, in press).
In this paper I consider each of these issues. Theory, empirical studies, and my own clinical
observations form the basis of my discussion. With regard to theory, I follow Bowlby's lead in
attempting to integrate other perspectives with that of attachment theory. Current thinking by
social ecologists extends the basis of attachment theory in ethology beyond biological systems.
Similarly, work by cognitive psychologists on memory systems expands our understanding of
how series of experiences are linked together into coherent histories and predictable patterns of
behavior. Finally, the origins of attachment theory in psychopathology (Ainsworth & Bowlby,
1991; Bowlby, 1973, 1980) and in varied cultures, particularly non-industrial cultures
(Ainsworth, 1967; see special issue of Human Development, vol. 33, no. 2, 1990 on attachment
from a cross-cultural perspective), provide a reminder of the importance of considering the full
range of human behavior when constructing theory (see Cicchetti & Greenberg, 1991).
My clinical observations are drawn from work, over a period of more than twenty years,
with families in their homes, in long-term parent groups, in family support centers, and in the
context of research. Of particular importance to the development of my thinking has been my
opportunity to apply notions drawn from systems theories (i.e., social ecology, family systems,
attachment theory, communication theory) to extended families followed over several years and
generations. The families included substantial numbers of low income families, families who
were exemplary as well as those who maltreated their children, families with normal and
handicapped children, and families from several ethnic, racial, and national groups. This is in
contrast to the majority of empirical data on attachment which is drawn predominantly from
middle income, low risk, white American mother-infant dyads.
Expectations and the Goal-corrected Partnership
Critical to developing attachment theory at ages beyond infancy is consideration of a basic
premise of ethology, i.e., that universal behaviors often serve functions which promote survival.
In infancy, the biological function of attachment behavior is clearly related to the survival of the
infant; indeed, attachment behavior functions as a bid for protection. In the preschool period, the
urgency of that function is diminished (although certainly not nullified) by children's increasing
ability to protect themselves (Cicchetti, et al., 1990; Marvin, 1977). Moreover, it may be
expected to continue to diminish until the activation of attachment behavior becomes a less
frequent occurrence in adulthood and one which, among the ordinary events of daily life, rarely
spells the difference between life and death.
The function of attachment, however, is not entirely realized by promoting the survival of
infants to adulthood. It also is necessary that individuals, in their own adulthood, foster the
survival of the next generation. Thus, we should expect that, as the relationship of children to
their attachment figure matures, children not only will assume more responsibility for their own
safety, but also will develop skills and/or propensities which will ultimately prepare them to
assume the role of attachment figure for their own progeny. Empirical evidence of the behavior
of preschool-aged children suggests that both processes may be occurring.
Protection. Quality of attachment in infancy refers to children's expectations regarding
maternal availability (Ainsworth, Waters, Blehar, & Wall, 1978). By the preschool years, these
expectations are largely formed as are children's strategies for managing their desire for
proximity. The new developmental task of the preschool period involves the goal-corrected
partnership (Bowlby, 1969/1982, 1973; Marvin, 1977). Whereas infants approaching one year of
age can be expected to monitor their attachment figure's whereabouts, to signal their desires, and
to follow their attachment figures, thus, helping to maintain the dyadic set-goal, preschool
children, whose cognitive and linguistic skills are substantially more advanced, seek
opportunities to communicate with their attachment figures regarding their mutual access to one
another (Marvin, 1977). Doing so gives children some control over their own situation and, thus,
leads to feelings of competence and security. Consequently, preschool-aged children may be
expected to tolerate, without great anxiety, separations which are substantially more extensive
than those tolerable to infants. Without such communication, even children who expect that their
attachment figure will be available if needed may feel anxiety, self-doubt, and anger.
With the development of more sophisticated patterns of communication and negotiation, dyads
are able to establish shared plans (Bowlby, 1969/1982, Marvin, 1977). These plans can include
conditional set-goals for maintaining children's feeling of security during the absence of their
mothers. Unlike set-goals in infancy which were constrained to the distance of a look or a call
and the time of relatively immediate access, conditional set-goals in the preschool period can
vary considerably. Moreover, simple self-protection and self-calming procedures, to be used by
children, can be incorporated into the plan for meeting the set-goal as can the use of alternate
caregivers (Cicchetti et al., 1990). Even time can be managed with references to the predictable
landmarks of children's days, e.g., after naptime, when Daddy gets home. Furthermore, plans for
changing the set-goal in response to changed conditions can be built into the plan, e.g., if Mrs. X
leaves, come home; if you get scared, call me. This flexibility not only protects children, it also
provides children with increased opportunities for exploration and affiliation. Thus, more
sophisticated means of managing the attachment relationship not only promote children's safety
but also permit, and even foster, increased opportunity for development in other areas.
Preparation for becoming an attachment figure. The transition from being an attached person
with the benefits of having an attachment figure to being an attachment figure to a child is
essential to the fulfillment of the ultimate function of attachment. It is hypothesized here that, in
order to function successfully as an attachment figure, adult humans should to be able to: 1)
communicate effectively using reciprocal patterns of non-verbal signals; 2) take the perspective
of others; 3) accept responsibility for regulating relationships; and 4) empathize with others who
are in distress. Although it may be presumed that there is a genetic bias favoring the
development of these skills, there is no reason to assume that they spring forth full-blown in
adulthood. Rather, it is probable that they develop over the course of childhood and that their
final form is the result of genetic heritage interacting with the unique history of the organism in a
context favoring their display. Moreover, it is probably not coincidental that, as threats to
survival becomes less immediate, developmental shifts occur which promote in preschool-aged
children precisely the forerunners of these behaviors. Quality of attachment in the preschool
years should, therefore, include aspects of child reciprocity, perspective-taking, management of
relationships, and empathy.
In conclusion, the development of attachment in the preschool years may be conceived of as a
two-fold process which promotes the survival of children and, concurrently, prepares them for
the role of attachment figure to the next generation. Quality of attachment in the preschool years,
therefore, refers both to children's expectations regarding the goal-corrected partnership and also
to the manner in which the relationship enables children to develop certain skills in the context of
Internal Representational Models
Development reflects the dynamic interaction of the changing individual with the environment
(Bronfenbrenner, 1979). To be fully understood, attachment relationships should be viewed from
the perspective of the changing competencies, needs, and environments of growing children. One
approach to integrating these influences on development is through internal representational
models. Such models reflect both aspects of human ontogeny and also the influence of unique
experiences on unique organisms, i.e., individual differences.
In his third volume, Bowlby (1980) focussed in detail on internal representational models
and their role in pathological outcomes. Bowlby conceived of these models as representing, in an
efficient shorthand, organisms' learned experience with attachment figures. The function of these
models was to reduce the time and effort needed to scan perceptually the state of the
environment and to formulate responses.
Internal representational models and working models. Bowlby considered a model to be a
"working" model when it enabled the individual: 1) to construct and evaluate several alternative
explanations for perceptions; 2) to explore mentally the probable effects of alternate responses;
and 3) to select and implement a logically coherent response. When experience had been
exceptionally painful, however, Bowlby believed that models often excluded important
information from perception (and thus interpretation); under such conditions models could
become rigid (i.e., "non-working"), thus precluding the mental exploration of alternatives.
Most current usage refers to models in general as "working models." It appears, however,
that this usage is not intended to make the differentiation inherent in Bowlby's selection of the
phrases "internal representational models", i.e., the inclusive, generic term, and "working
models", the narrower, more specific term. Because rigidity of models is critical to
understanding pathological behavior, the usage here will conform to Bowlby's.
Procedural, episodic, and semantic models. Similarly, many investigators write as though
each person could be characterized by a single representational model. Bowlby, however,
referred to studies of memory systems indicating that humans have several separate memory
systems, each with an internal representational model (Bowlby, 1980). Based on Tulving's
(1972) work, Bowlby addressed specifically only the two best differentiated systems, semantic
and episodic, but pointed out that other memory systems would probably be identified in the
Tulving's more recent work has identified "procedural" memory as encoding knowledge
in the form of patterns of behavior (Tulving, 1979, 1985). This memory system is believed to
operate preconsciously throughout the life-span and to be the primary guide to everyday
behavior. That is, until an individual perceives a discrepancy between behavior and expectations,
behavior will occur preconsciously based on information in procedural memory. Crittenden
(1990) has theorized that procedural internal representational models are associated with this
memory system and that inferences regarding the nature of procedural internal representational
models can be drawn from patterns of behavior.
Although almost all behavior results from procedural memory, when asked verbally
about their behavior, individuals respond in terms of information encoded (verbally) in semantic
or episodic memory. When the models associated with these three memory systems are relatively
similar, explanations from one system will fit behavior from another. On the other hand, when
the systems contain different information, only concerted effort, and the distress resulting from
disparities between experience and expectation, enables individuals to become conscious of the
functioning of their procedural models.
These distinctions are very relevant to exploration of the transition from infancy to the
preschool years. With the development of additional memory systems during the preschool
years, the possibility for coherence or lack of coherence in functioning becomes greater. Thus,
the preschool years hold the potential for opening new pathways to growth or, conversely, for
augmenting the distortions associated with anxious infant attachment.
The development of internal representational models. The Ainsworth system for
classifying pattern of attachment in infancy relied primarily on observed pattern of behavior
(Ainsworth, Blehar, Waters, & Wall, 1978). Because infants encode, store, and retrieve
information through sensorimotor means, their internal representational models can only consist
of patterns of behavior, i.e., procedural models. Given the limited cognitive structure of infants,
assessing procedural memory is appropriate. In the preschool years, however, the increased
sophistication of children adds to the complexity both of observed behavior and also of the
underlying cognitive structures. In order to capture this complexity, a wider range of information
and more fully articulated theory are needed (Cicchetti et al., 1990; Crittenden, 1990).
Preschool-aged children's advances in cognition and language create the possibility of
abstract models encoded through language and available to children for conscious review and
modification. Because both the semantic and episodic memory systems depend upon language as
a means of representing information, internal representational models based on these memory
systems may be presumed to first develop during the preschool years. Although the nature of
these models has not been explored widely, there is a wide basis in theory and empirical research
for speculations regarding the development of procedural models in the preschool years
(Cicchetti et al., 1990). Consequently, they will be emphasized here. Future work, however,
should focus more fully on the development and nature of models accessible through more
abstract means (Bretherton, 1990, 1991).
Developmental Shifts in the Salient Aspects
of Quality of Attachment
Ainsworth identified three patterns of behavior among infants; a fourth combination
pattern was identified by both Crittenden (1985ab) and Radke-Yarrow and her colleagues
(Radke-Yarrow, Cummings, Kuczynski, & Chapman, 1985)
. Assessment of these patterns of
behavior, i.e., secure (B), avoidant (A), ambivalent (C), and avoidant/ambivalent (A/C), in the
Strange Situation in infancy is the basis for inferring infant quality of attachment. The behavior
of preschool children, however, appears to be sufficiently subtle and complex that additional
information is needed to determine quality of attachment. In this section, the nature of that
information will be considered.
Observation and interpretation of the behavior of samples of disturbed children played a
critical role in the recognition of this subtlety and complexity. Two major transformations in
quality of attachment were observed in samples of maltreated and emotionally disturbed
Main's disorganized/disoriented category is discussed below in the section titled
"Organization, reorganization, and disorganization."
children. The first was the observation that some children used patterns of defensive behavior
other than avoidance. This required reconceptualizing the "avoidant" strategy as a "defended"
strategy (Ainsworth, personal communication, December 31, 1991; Crittenden, 1990). Second,
the organization of preschoolers' C-type behavior around a coercive strategy was observed in the
behavior of maltreated children. Their more extreme behavior highlighted the functional
outcome of resistant and coy behavior more clearly than did the behavior of adequately reared
children. Moreover, knowledge of the nature of these children's daily experience, challenges, and
limitations (Crittenden, 1988ab) greatly facilitated the recognition of functionally similar
patterns of behavior in the laboratory among the myriad of unique behavioral variations
Recognition of these patterns depended upon aspects of quality of attachment other than
pattern of behavior. Although these aspects are parts of a whole which cannot be dissected
without distortion, specification of them provides a means of clarifying and focusing meaning.
None of these aspects of attachment is sufficient to define quality of attachment; neither are they
independent of each other. Nevertheless, together, they provide a context within which children's
behavior may be interpreted in terms of overall strategy. Consequently, developmental change in
six aspects of quality of attachment will be considered: strategy, pattern of behavior, pattern of
negotiation, regulation of affect, secure base phenomena, and the attachment figure's behavior.
Strategy refers to the notion that repeated patterns of behavior serve a consistent function for
children in the attachment relationship (Main, 1981). These patterns indicate how children have
resolved the dilemma of maintaining access to an attachment figure during periods of perceived
danger while retaining the opportunity to explore the environment and affiliate with others when
not threatened. Although it will not be asserted here that the three patterns of attachment
identified by Ainsworth are necessarily the only relevant patterns for the preschool years,
extensive viewing of a widely diverse group of Strange Situations with preschool-aged children,
based on the opposite assumption, has yielded additional patterns of behavior which,
nevertheless, resolve functionally to the same three strategies. It is particularly important to note
that this is true even for samples including very high risk dyads, dyads with clear pathology in
interpersonal relationships, and dyads of varied ethnicity and social class. Consequently, for
preschool children, as for infants, three strategies and one combination strategy are proposed:
I. The strategy used by secure (B) children consists of engaging the attachment figure in a
goal-corrected partnership that allows them to maintain proximity during stressful conditions and
to explore widely during periods of felt security. In such a partnership, children share with
attachment figures responsibility for appraising potential danger, communicating with the
attachment figure about plans for protecting children and about feelings regarding those plans,
and behaving in ways which increase children's feeling of security.
Preschool-aged children's ability to take another's perspective enables them to recognize
that their attachment figure may have different plans or desires, but that they will not be
abandoned when their attachment figure attends to these other concerns (Marvin, 1977). Instead
they either trust their attachment figure's return or negotiate its conditions. Upon reunion, even if
they were upset, they quickly resolve their feelings by finding comfort and satisfaction in close
and intimate contact with the attachment figure and, using their attachment figures as a secure
base, move on to other activities.
II. The strategy used by defended (A) children consists of maintaining protection during
periods of perceived danger without alerting the attachment figure to children's true feelings or
desires for proximity, both of which might arouse the attachment figure's anger. To do this,
children avoid focusing attention on the relationship or on feelings and accept full responsibility
for monitoring the environment, maintaining access to the attachment figure, and regulating
emotions. By monitoring attachment figures' behavior and affect to infer the attachment figures'
plans , they are able to organize their own behavior so as to achieve a satisfactory balance
between physical availability and emotional distance.
A balance is sought between the threat provided by the attachment figure's anger and the
threat of not having access in times of danger. Both because inhibition of the display of such
feelings is most difficult in the context of intimate communication and because anger is more
dangerous when the attachment figure is very close, defended children regulate proximity to the
attachment figure so as to avoid intense closeness. Concurrently, they avoid too much distancing
by trying to prevent their attachment figure from withdrawing either physically or
psychologically. In other words, defended children seek to be close, but not too close.
Ironically, this pattern of behavior implies well developed perspective-taking skills which, in
the context of inhibition or distortion of affect, fail to culminate in joint communication and
planning. To the contrary, they leave the child with the added burden of thinking for a second
person. Among defended children, awareness of the attachment figure's perspective often
appears to be clearer than awareness of their own.
III. The strategy used by coercive (C) children consists of attempting to coerce a reluctant
attachment figure to meet children's set-goal of constant availability. The coercion involves the
use of threats and bribes combined with refusal to enter into negotiations which could result in
compromise. Coercive children accomplish this by affectively weighting their communications
so as to avoid intimacy with its potential for reciprocal perspective-taking, compromise, and
empathy. Threats are communicated by displays of anger to alert attachment figures to the
presence of a problem and to their responsibility for resolving it. These displays imply (falsely)
children's feeling of powerfulness. Bribes are communicated by displays of coy babyishness.
These focus the attachment figure's attention on the desired outcome (i.e., nurturance) by
displaying (falsely) children as incompetent.
The ambivalence of coercive children reflects both their uncertainty about how to repair
the situation, i.e., threaten or entice, and also their conflicting feelings with regard to the
attachment figure, i.e., anger and desire. (Such conflicting feelings are common in relationships.
Secure dyads, however, are clear about how to express and resolve them, i.e., by intimate
exchange of information accompanied by perspective-taking and compromise which respects
both partners' positions. Defended dyads choose not to deal with feelings at all.) The intensity of
coercive children's feeling is apparent from their inability to engage in alternate activities, e.g.,
exploration and affiliation, until access to their attachment figures is restored.
IV. Children classified as defended/coercive (A/C) (Crittenden, 1985) show both a defended
strategy and a coercive strategy. The strategies may be merged or may alternate. In the latter
case, children who begin using one strategy in the first reunion switch to the other strategy in the
second reunion. In other cases, children begin a reunion using one strategy, but after a short time,
switch. When changes in the attachment figure's behavior can account for changes in child
behavior, the A/C pattern should be considered organized. When, however, the child switches
from a defensive to a coercive strategy apparently as a function of increasingly intolerable stress,
evidence of disorganization (see below) should be sought. In addition, many A/C children show
stereotypies; whether these indicate disorganization or merely stress has not yet been determined.
Patterns of behavior
Ainsworth has made the case that the specific behaviors used by infants do not predict quality
of attachment (Ainsworth et al., 1978). Her notion, common to system theories but quite foreign
to other approaches, is that many different behaviors can be used to reach a goal. Some may be
positive in that they are aimed at reaching the goal whereas others are negative in that they are
aimed at precluding conditions which are incompatible with the goal. The specific behaviors
used are less important than how they are organized to achieve the goal (Sroufe & Waters, 1977).
This perspective, which is essential to understanding quality of attachment in infancy,
becomes even more important and complex in the preschool years. Because preschool children
recognize subtler aspects of behavior and behavior sequences, their potential for accommodation
to external conditions is greater than that of infants. Their repertoire of behaviors and patterns of
behavior may be expected to become correspondingly complex. Whereas in infancy a single
pattern of behavior was associated with each pattern, the patterning of behavior appears
substantially more complex in the preschool years.
Strategies and patterns of behavior. Indeed, there appear to be multiple patterns, each tied to
different circumstances, associated with each of the three strategies in the preschool years. These
patterns represent alternate ways of implementing the strategy. For example, defended children
who fear that displaying attachment behavior will result in rejection may enact the defended
strategy by inhibiting displays of fear and anger at the attachment figure's departure. Instead of
displaying attachment behavior, they may appear unconcerned regarding the whereabouts or
behavior of the attachment figure.
On the other hand, the same strategy may be enacted by other defended children who fear loss
of availability of the attachment figure through withdrawal (rather than through rejection). In this
case, children may attempt to maintain availability by initiating interaction with the withdrawn
attachment figure using false, but attractive, displays of cheerfulness instead of accurate displays
of fear, anxiety, and anger (Lieberman & Pawl, 1988; Main & Cassidy, 1988). In this case, the
false cheerfulness serves to keep the attachment figure at least minimally aware of, and involved
with, children (and, thus, available to protect children).
Still other defended children whose attachment figures are easily angered may find that
compliance with the attachment figures' demands combined with inhibition of displays of
negative affect best serve to maintain accessibility for protection without arousing anger toward
the children (Crittenden & DiLalla, 1988). Such anger, of course, would reduce the attachment
figure's protective function.
Patterns and behaviors. Similarly, because the same behavior or behavioral unit can be used to
implement more than one pattern, it is essential to consider the function of the behavior as it is
used by each child. For example, a child might appear softly sweet toward the attachment figure.
In the context of accepting offered nurturance following distress, this might be a secure child
relaxing in the comfort of maternal warmth. In the context of adult depression and psychological
withdrawal, this behavior might function so as to enable a defended child to attract their
attachment figure's attention. On the other hand, in the context of adult unpredictability, this
might be coy behavior aimed at enticing the attachment figure to respond with nurturance. In the
latter two cases, the charming quality of the behavior also serves to appease any anger the adult
might feel toward the child. Morphologically similar (or even identical) behaviors are, thus,
labeled with terms which imply the function of the behavior.
Anger also can be used in service of more than one strategy. A securely attached child might
show anger at the attachment figure for leaving. The anger, however, will be accompanied by
information regarding what the attachment figure did (leave or not explain her departure) and
how children felt (frightened or hurt). Moreover, the anger will be resolvable. When displayed
by a defended child, the anger will probably not be apparent to the attachment figure, i.e., it will
be displayed only when the attachment figure is not present or when the child's back is turned.
Furthermore, because there is no acknowledgement of the anger, resolution is unlikely. A
coercive child using anger is likely to be very clear, even punitive, about feeling anger but
accompanying feelings of hurt and vulnerability are unlikely to be revealed. Moreover, the true
source of the anger, i.e., the nature of the relationship, is not revealed. Without recognition of
these feelings and their source, the possibility of reconciliation is limited.
Patterns of behavior. Assessment of patterns of behavior is complicated further because
individuals tend, on occasion, to incorporate patterns typical of other strategies, into their
predominant strategy. For example, securely attached children will, in appropriate circumstances,
use defensive or coercive behavior patterns. Such flexibility is, of course, adaptive. Nevertheless,
the result is that the simple display of a behavior or pattern of behavior becomes insufficient to
differentiate definitively among qualities of attachment.
Instead, the differentiation of the secure, defended, and coercive patterns of behavior depends
upon how the behavior functions with regard to the overall strategy. In order to identify the
function, it is important that patterns be described in terms of the conditions which activate and
terminate behavioral systems. The particular systems under consideration here are the
attachment, exploratory, and affiliative systems. The relevant conditions include both the
observation procedure and attachment figures' behavior in that procedure.
Securely attached preschool-aged children, like securely attached infants, enjoy proximity to
the attachment figure. As older children, however, they explore more widely and need less
contact to reassure them (Crittenden, 1988c). Unless the children are very distressed, distal
interaction and subtle gravitating or drifting toward the attachment figure with engagement in
play and only brief touching are more common than direct and immediate approaches with
requests for body contact (Cassidy & Marvin with the Working Group of the John D. and
Catherine T. MacArthur Foundation on the Transition from Infancy to Early Childhood, 1989).
The latter behavior, in fact, becomes indicative of anxiety with increasing child age. Three
subgroups, i.e., reserved (B1-2)
, comfortable (B3), and reactive (B4), are proposed.
Defended children commonly display (at least) three patterns of behavior, i.e., inhibition (A1-
2), compulsively caregiving (A3), and compulsively compliant (A4). Inhibited children use
subtle forms of avoidance. Caregiving children's behavior ranges from overbright cheerfulness to
non-reciprocal monologues directed toward withdrawn attachment figures to role reversing
caretaking of the attachment figure. Compliant children vigilantly attend to the attachment
figure, inhibit negative affect in the context of unpleasant interference from the attachment
The analogous subgroup in the infant classificatory system is indicated. If none exists, a new
subgroup letter is provided. Wherever the meanings are similar, the names selected are those
used by other investigators; new names are provided when the function of the behavior seems
different from that described by others or where analogous subgroups do not exist in other
figure, and comply rapidly with the attachment figure's suggestions or demands. Although the
outward morphology of the behaviors differs greatly, in the context of the attachment figure's
behavior, each pattern is congruent with the defended strategy of maintaining access without
It is not clear whether most defended children select only one of these patterns of regulating
affect or whether the same children use different styles in different situations. It seems probable,
however, that some children use only the inhibited pattern of behavior whereas other children
who use the caregiving or compliant patterns also use the inhibited pattern.
Among coercive children, the raw display of anger or helplessness which characterizes
ambivalent infants becomes a pattern of behavior used to coerce the attachment figure to do as
the child wishes. There is often evidence that such threatening and disarming behavior is
instrumentally coercive. For example, a tantrumming child may pause, listening for the effect of
the tantrum, before continuing to scream. Even though in a given instance a child may appear
primarily, or even entirely, angry or coy, it appears that there is considerable flexibility in
children's use of threatening and bribing behavior to fulfill the coercive strategy.
Although the behavior of coercive children has often been labeled dependent or immature, it
should be remembered that it is not typical of children of any age (Bowlby, 1980). Instead, it is
indicative of a path of development which is responsive, as are other paths, to developmental
change. Its similarity to the behavior of young children may reflect the protective function of
"babyish" characteristics, such as wide eyes and pre-cry expressions, which tend to evoke
caregiving responses from parents. Used in conjunction with threatening behavior, they may
systems. Full descriptions of the subgroups is provided in Appendix 1.
function to appease the otherwise angry attachment figure.
Four subgroups are proposed: threatening (C1), disarming (C2), punitive (C3), and
helpless (C4). Threatening children are openly angry, resistent, and/or pouty. Punitive children
extend this pattern to open aggression, shaming, and/or embarrassment of the attachment figure.
Disarming children use infantile and seductive behavior, i.e., coyness, flirting, bashfulness. At
the extreme, helplessly coercive children appear so immaturely incompetent as to require the
attachment figure's assistance.
Infants live in the present. Consequently, it is not possible for attachment figures to negotiate
meaningfully with them about future conditions. Moreover, they lack the communicative skills to
carry out such exchanges. Finally, they have no understanding of the motivation for others'
Preschool-aged children, on the other hand, have conscious memory of familiar and frequent
past experiences and a rudimentary understanding of the concept of future. They also have a
wide repertoire of non-verbal shared meanings and an increasing vocabulary of words. In
addition, they have an understanding, albeit limited, that others feel differently from themselves
and want things that they do not. In sum, they are aware that others have different perspectives
even if they do not understand or share those perspectives (Marvin, 1977).
These skills make it possible for preschool children to negotiate with their attachment figures
regarding their access to one another, i.e., to establish goal-corrected partnerships (Bowlby,
1969/1982; Marvin, 1977). The nature and content of this negotiation is an important indication
of the strategy used by children to maintain attachment figures' protective function.
Secure children feel confident that they and their attachment figures share the basic goal of
maintaining psychological closeness and comfort. Consequently, they use open and direct
negotiation to exchange information about their own and their attachment figure's whereabouts
and about plans for future access as well as feelings regarding the plans. Because sharing a set-
goal of mutually planned accessibility with their attachment figure leads to feelings of security,
they are able to negotiate with their attachment figure to find a plan which meets both that goal
and the desire of both partners to engage in other activities. Although such a plan does not
necessarily require proximity, it does require reciprocal communication which is sensitive to
children's feelings. This negotiation may be very abbreviated when, for a given dyad, much can
be assumed in the context of an on-going intimate and supportive relationship. The point is that
anger and concern are not left unresolved.
Defended children use unilateral refusal to negotiate in order to avoid rejection by their
attachment figures. Because exchanges regarding true feelings highlight children's desires for
support from the attachment figure, they increase the probability of rejection. Consequently,
defended children find it difficult, if not impossible, to negotiate the terms of separations and
reunions without increasing the risk of separation.
Rather than risk either rejection or loss of emotional control (with its risk of eliciting
rejection), defended children do not seek to negotiate with their attachment figure regarding
mutual plans. Even if their attachment figure attempts to open a discussion of the child's feelings,
perhaps with a comment such as "Did you miss mama?", most defended children do not reveal
their feelings and refuse to enter discussions which could resolve, but in fact will probably
exacerbate, their feelings.
Coercive children, who are constantly fearful of their attachment figure's unavailability, but
who are not afraid of being rejected, can be expected to use leveraged negotiation, i.e.,
demandingly coercive behavior without reciprocal negotiation. Some offer threats of disruptive
misbehavior or pouty sullenness for disregard of their wishes whereas others entice their
attachment figure with the reward of coy sweetness for compliance by the attachment figure.
The underlying demand is, however, unlikely to be as simple as some form of
compliance, e.g., not leaving the room. Instead, calming coercive children may be expected to
require that the attachment figure acknowledge and accept children's feelings together with
including children in negotiations over mutual access. Coercive children do not, however, take
responsibility for either explaining the nature of the problem or negotiating possible alternative
solutions. Because they cannot count on the attentiveness and responsiveness of their attachment
figures, compromises which permit increased unavailability cannot be accepted. Instead
attachment figures are left to both identify the problem and meet their demands. Occasional
direct discussions of children's resentment of the attachment figures' absence rarely to result in
reciprocal communication with displays of true feelings and compromise. Without these, the
process of negotiation may be expected to fail.
Affect regulation refers to two distinct universal human processes. The first is the regulation
of internal feeling states. The second is regulation of the display of affect. (To avoid confusion,
the term feelings will be used to refer to actual internal states; displays of affect will denote the
type of feeling that is displayed.)
Regulating feelings. All children experience feelings of comfort, fear, anxiety, anger, and
relief in response to their attachment figures. These feelings both serve as indicators to children
of the nature of the situation and also motivate children to take appropriate action (Bowlby,
1969/1982). From an evolutionary perspective, if children, in response to these feelings, act in
ways which reduce negative feelings and maintain positive feelings, they will tend to increase
the probability of experiencing safe and growth enhancing situations.
Children who trust their attachment figures recognize their own feelings and use them to
appraise the degree of risk in their situation. They also communicate these feelings accurately to
their attachment figure, and, with the help of adults, develop means of managing them. Doing so
involves a challenge to the developing skills of preschoolers, especially their ability to accept
that others have independent needs and plans that prevent them from acting as children wish.
This understanding is essential to children's willingness to compromise, to their development of
self-initiated coping skills, and to resolution of feelings when children are forced to tolerate
unpleasant situations. If children cannot recognize their feelings or refuse to act upon them, a
situation of greater risk is created. Conversely, if feelings are appraised but not managed,
children risk becoming overwhelmed by the feelings themselves and, consequently, at risk for
failing to correct potentially dangerous situations (see Cicchetti, Ganiban & Barnett, 1991).
The issue, then, is twofold: 1) to attend to feelings as one means of appraising situations, and
2) to use that information to behave in ways which promote desirable conditions. One important
form of behavior which modifies situations is affective communication, i.e., displays of affect.
Others include direct action to change the situation, using distractions to wait out the natural end
of the situation, self-given "pep talks," and seeking help in resolving problems (Cicchetti et al.,
Securely attached children regulate their feelings by seeking and obtaining information
from attachment figures regarding their whereabouts and plans. When their attachment figure is
not available, they seek help from other adults to sooth their feelings and to provide information
regarding their attachment figure's whereabouts and return. They both actively seek their
attachment figures and also engage in self-soothing or self-distracting behaviors. Their range of
coping skills usually enables them to keep their feelings of anxiety under control such that they
remain able to solve problems creatively.
Like avoidant infants, defended preschool children attempt to regulate their emotions
solely through their own efforts. Most use regulation of the display of affect as an important
means of controlling both their own feelings and their attachment figure's behavior. Some do this
by displaying only non-threatening emotions. Past experience has taught these children that their
attachment figures will not help them to resolve uncomfortable feelings of anxiety or anger and
that displays of such feelings may actually exacerbate both the problem and their feelings.
Consequently, defended children tend to inhibit displays of affect which might either lead to
rejection by their attachment figure or overwhelm their own self-control. This is often facilitated
by focusing on alternate activities (Main, 1981).
Coercive children accept little or no responsibility for regulating their own feelings.
Instead they substitute demanding and/or helpless behavior for more productive efforts to resolve
their feelings of anxiety. Although such behavior encourages greater involvement of attachment
figures with children, the use of such limited means of resolving problems combined with the
children's experience of their attachment figure's insensitivity tends to increase both their
feelings of anxiety and anger and also their actual vulnerability.
Display of affect. Displayed affect signals to others what individuals feel and suggests,
therefore, how they may be expected to act. By alerting others to the nature of children's feelings,
negative affective communication can be helpful in changing the situations that aroused feelings.
Once alerted, adults (either attachment figures or others) can help children both to resolve the
problem and also to learn appropriate ways to regulate feelings. Moreover, because children
learn to recognize the effects of their displays of affect, displayed affect also has an instrumental
function in influencing the behavior of others (Campos, Campos, & Barrett, 1990). Failure to
share true feelings, i.e., inhibition of affect and false affect, makes it difficult for others to
identify how children feel or what type of action would be helpful. In such situations, children
may be left to resolve problems entirely with their own resources.
Securely attached children are likely to display accurately both positive and negative feelings.
For example, when, after leaving them against their wishes, their mothers have returned, many
securely attached children first act pouty and then quickly make up and show clear pleasure at
Regulation of the display of affect among defended children varies from inhibiting displays of
affect, common to children who select an inhibited pattern of behavior, to falsely displaying an
affect which does not match the felt emotion, common to children who select the compulsively
caregiving or compliant patterns.
Inhibition is particularly evident during moments of heightened emotion such as when the
attachment figure is leaving; most defended children refuse to look at the attachment figure at
this moment, choosing instead to look down, engage another person, or play with toys. Others
make an oblique bid to keep their attachment figures nearby by highlighting a toy or asking a
question just as the attachment figure leaves. In these cases, children focus the request entirely
upon the toy without indicating that they feel uncomfortable or want the attachment figure.
Displays of false affect include overbright, solicitous affect which is neither consistent with
the internal state of the child nor with the circumstances. Such false displays are usually sudden
in both appearance and disappearance; moreover, they are often unusually intense, e.g., a sudden
flashing smile, a loud, high-pitched laugh or squeal. Both inhibition of affect and falsely
displayed affect are often accompanied by "leakage," displays of affect which reflect true
feelings, particularly feelings of anger, fear, distress, and despair. Children who display false
affect with the attachment figure may show genuine feelings when she is absent.
Exaggerated displays of affect are the predominant means used by coercive children to force
attachment figures into compliance with their desires. Such displays reflect accurately some, but
not all, of children's feelings. For example, threatening/punitive children display anger, but not
anxiety, hurt, or fearfulness, which would reflect the vulnerability of children. Their revelation to
the attachment figure would create the opportunity for alternative solutions whereas their
absence tends to prevent intimately sensitive responses from others. Conversely,
disarming/helpless children display only their vulnerability and inhibit displays of anger. The
absence of anger tends not to alert others to the intensity of their feelings or to motivate others to
respond. Sole reliance upon this interactive process may rob children of the opportunity to learn
to regulate for themselves the underlying feelings.
Regulation of displays of threatening or disarming affect is dependent upon children's
assessment of the attachment figure's affective state and its likely outcome. Threatening and
punitive behavior that makes attachment figures feel guilty can be expected to result in
compliance with child wishes. Behavior which results in anger or extreme frustration will likely
lead to aggression or withdrawal, respectively, by the attachment figure. Similarly, disarming
and helpless behavior that makes the attachment figure feel responsible will tend to result in
nurturance. When, however, dependence is maintained too long, it may result in attachment
figures' punishing children for babyishness or giving up trying to help them. Consequently, it is
essential that coercive children monitor attachment figures' affect and switch behavior patterns
when attachment figures' feelings of guilt begin to shift to feelings of anger or despair. Failure to
do so, leaves the child at risk for abusive displays of anger or neglectful failure to nurture.
Unfortunately, when displays of affect are so exaggerated that they are not easily terminated,
children may become preoccupied with the display. In such cases, children may become too
distraught to communicate anything specific regarding the cause of distress or may so frustrate
caregivers that attachment figures lash out in anger or withdraw assistance. In such cases,
children not only create risk to themselves, they also fail to learn to monitor others' affect.
Secure base phenomena
Among Ainsworth's contributions to attachment theory is the recognition that attachment
contributes to development when the attachment system is not activated by facilitating
exploration of the environment (Ainsworth, 1979). This promotes survival by enabling infants to
maximize their potential for adaptation through cognitive development.
It is proposed here that, in the preschool years, the attachment figure also becomes a secure
base for exploration of the social world. Children's confidence in themselves, derived from their
relationship with their attachment figure, affects greatly the extent and nature of their exploration
of interpersonal relationships (Elicker & Sroufe, in press). The quality of their attachment also
affects the manner in which they handle their desires for intimacy and feelings of vulnerability as
well as how they interpret and respond to others' behavior.
Securely attached children tend to explore the environment and affiliative relationships
freely as long as there is no signal of danger. Even in play they display their growing ability to
take the perspective of others and to empathize with vulnerable persons. Play is less mature if
they are upset by their attachment figure's absence.
Although most securely attached children accept overtures from unfamiliar people with ease,
they usually initiate interaction with them quite tentatively and cautiously suggesting an
awareness and respect for the barriers separating unfamiliar people. Rather than being indicative
of fearfulness or wariness, this respect suggests their understanding that relationships involve a
potential for intimacy and result from a mutually accepted, reciprocal process between
individuals. Their behavior is more attuned to that of other people and the context than the facile
or superficial behavior of defended children.
Defended children frequently employ the exploratory behavior system to regulate their
feelings of anxiety. They use toys as a distraction during times of stress. Their play, however,
often lacks the interactive or cognitive complexity and maturity of the play of securely attached
children. Especially missing are indications of the development of empathy; themes of
aggression are, in fact, more common.
Defended children often appear comfortable with unfamiliar people, but unable to seek or use
their assistance to resolve problems or to engage in close, interpersonal exchanges. They may
become promiscuously intimate, i.e., intimacy without sharing of true feelings and without
perspective-taking or mutuality. In this case, they remain brightly impersonal without revealing
their true feelings and treat people more as objects than as responsive beings. Such behavior
mitigates against the development of genuinely and mutually intimate interpersonal
The attachment figures of coercive children tend not to provide a secure base from which
children can confidently explore the environment and relationships. On the contrary, most
coercive children are preoccupied with their attachment figures. This appears both as
involvement with the attachment figure and as inability to engage with people or explore the
environment productively. Some coercive children are very active, even hyperactive, whereas
others aimlessly drift from object to object. Frequently, they seem unable to settle on an activity
and demand constant adult involvement in immature play, i.e., "What's this?" sequences. Other
coercive children are subdued and pouty or helpless. Many are unable to present even the
pretense of play when alone; such children are totally preoccupied with regaining the attachment
figure to the exclusion of all other activities or alternatives.
In some cases, however, coercive children have separated their own competence from their
relationship with their attachment figure. In such cases, the children show resentfulness, anger,
and preoccupation primarily when they are with the attachment figure. When the attachment
figure is fully absent (and they are safe), however, they relax, seeming almost grateful to be able
to set the struggle aside. Such children appear able to use the absence of the attachment figure to
free them for exploration and affiliation. These coercive children often behave more maturely
with other adults than with the attachment figure. The prevalence, causes, and outcomes of this
pattern need further research.
The attachment figure's behavior
Attachment figures' past and current behavior are both essential for understanding children's
quality of attachment in the preschool years.
Past behavior. Previously, this has been judged the more important type of behavior and
consists of the cumulative experience of the dyad, as children perceive it, encoded as internal
representational models of the attachment figure and the self (Bowlby 1980; Crittenden, 1990;
Crittenden & Ainsworth, 1990; Main, Kaplan, & Cassidy, 1985). These models summarize, in
both behavioral and abstract ways, what children know about the meaning of attachment figures'
behavior in terms of causal conditions, identifying features, and consequences as well its relation
to child behavior and feelings. It is on the basis of these models that children 1) develop
strategies for managing the relationship with the attachment figure, and 2) interpret ongoing
maternal behavior (Bretherton, 1990). A substantial body of research and theory exists
identifying for infants the relation among patterns of maternal behavior at home, hypothesized
internal representational models formed by infants, and infant behavior at home and in the
Strange Situation (Ainsworth, 1985; Belsky, Rovine, & Taylor, 1984; Crittenden, 1981, 1985ab;
DiLalla & Crittenden, 1990; Grossmann, Fremmer-Bombik, Rudolph, & Grossmann, 1987;
Sroufe & Fleeson, 1986). Similar work is needed for the preschool years.
Current behavior. The second type of behavior necessary for understanding child quality of
attachment consists of attachment figures' behavior during observational procedures for
assessing quality of attachment. Such behavior is evaluated by children in terms of both the
attachment figure's previous pattern of behavior and also the unique characteristics of the
immediate setting. On the basis of this evaluation, children select responses from their repertoire
of possible responses. In other words, children adjust their behavior to fit the actual behavior of
the attachment figure in the immediate circumstances. Because behaviors can be used in service
of different strategies and patterns of behavior, knowledge of both the attachment figure's
behavior and the context is necessary for interpretation of the meaning of child responses.
Although maternal behavior in the infant Strange Situation is constrained to be as uniform
across attachment figures as possible, it appears that, particularly in the preschool years, the
attachment figure's behavior during the Strange Situation does influence child behavior in the
Strange Situation. This should not be taken to mean that children's quality of attachment varies
with differences in attachment figures' behavior during the Strange Situation (or other
observational procedure). On the contrary, children's representational model, i.e., quality of
attachment, is both stable and also the means by which children interpret variations in maternal
behavior and select specific behavioral responses. Consequently, without information regarding
maternal behavior (as well as experimentally induced constraints upon maternal behavior), the
observer will find it difficult to accurately identify the strategy (and internal representational
model) underlying children's behavior.
Attachment figures of secure children are typically interested in their children's activities and
show affect which matches or meshes with the child's, e.g., matched happy expressions, maternal
concern meshed with child distress. Moreover, they often take an active role in helping their
children learn to resolve difficult situations and manage their feelings.
Some attachment figures of defended children are overbright and out of synchrony with the
child; they behave as though the child were warmly engaged with them (Crittenden, 1981,
1985ab). In these cases, both the attachment figure and the child cover negative feelings with
neutral or positive affect (i.e., mismatch of felt and displayed affect.) Attachment figures
showing this behavior typically have children who are inhibited. Other attachment figures of
defended children (particularly of those who use compulsively caregiving or compliant patterns
of behavior) are unresponsive, withdrawn, depressed, and/or overtly angry. Neither group is
competent at helping their children resolve difficult situations, e.g, entering a new room, meeting
a strange person, separating from the attachment figure.
Attachment figures of coercive children are unresponsive to all but the most intense of
their children's signals. They often vacillate between being angrily frustrated by the children's
demands and also empathic to the extent of experiencing their children's feeling states. In either
case, attention to children's feelings results in such intense feeling on the attachment figures' part
that they are unable to be supportive. Often such attachment figures also seem very dependent
upon the approval of their children. They themselves typically cajole or threaten their children to
obtain compliance. In what appear to be efforts to obtain attention and affection from their
children, they frequently extend leave-taking unnecessarily, ask about their child's love for them,
and express frustration at their inability to please their children. The children capitalize on this
evidence of adult vulnerability by being pointedly rejecting or openly aggressive when they are
displeased and charmingly tender when trying to cajole the attachment figure into compliance.
In play with their children, attachment figures of coercive children seem unable to yield
"ownership" of the play to children; instead the appearance is often of two children squabbling.
A few seem seriously disturbed themselves, appearing passive, depressed, dangerously angry,
and/or unpredictable. Many of these make derogating remarks to their children or actively tease
and frustrate their children. Overall, there is a substantial match in the way that the attachment
figure and child treat one another and clear dissonance between the partners, i.e., each appears to
be in power at the expense of the other.
Organization, reorganization, and disorganization
The notion of organization of behavioral systems, reflected in internal representational
models, is central to attachment theory (Bowlby, 1980). However, neither the models nor the
patterns of organization are thought to be static. To the contrary, at any given moment, models
are hypothesized to represent the most complex representation of reality of which the individual
is capable. That is, using Piaget's notions of accommodation and assimilation, infants are
presumed to construct models which fit some aspects of reality but ignore many subtleties. These
are beyond the abilities of infants to perceive, interpret, and/or respond to.
As a function of developing cognitive, linguistic, social, and emotional competencies, children
become capable of accommodating more of the information in their environment. Thus,
development results in new information to be incorporated into representational models and new
processes for integrating that knowledge. From this perspective, internal representational models
are viewed as dynamic, but simplified, versions of experience which accommodate some aspects
of reality and assimilate the remainder. That is, they are constantly in a state of reorganization as
new experiences and maturation that create new possibilities for perception, integration, and
response. When this process involves substantial or abrupt shifts, both behavior and internal
representational models may become disturbed. Indeed, the signs of disturbance may be
indicative of periods of rapid developmental change.
Such signs have been associated with the notion of "disorganization" when they have been
observed in the Strange Situation (and other separation procedures). These signs of
"disorganization" are behavior which does not fully match the descriptions in the Ainsworth
classificatory system (Main & Weston, 1981). Specifically, this includes behavior that is a) out
of the typical order, b) a mixture of incompatible behaviors, or c) odd, e.g., stilling, dazed
expressions, frozen gestures. Children who show these behaviors and who are classified as
"disorganized" (D) are frequently found to have histories which include unresolved maternal loss
or trauma, maltreatment, and maternal depression (Main & Hesse, 1990; Main & Solomon,
It is suggested here that questions regarding disorganization might better to be framed in
terms of internal representational models: Are children's models of attachment figures and of
themselves organized? That is, do the models allow children to perceive and interpret relevant
information and to respond in ways which may be expected to yield the predicted outcome of the
From the perspective of attachment theory as it has evolved since 1985 and as it is being
developed here, many children who might otherwise be considered disorganized can now be seen
to have an underlying logic, i.e., organization, to their behavior. Understanding this logic
depends upon recognizing critical aspects of situations presented to children, identifying the
function or predictable outcome of children's behavior, and/or postulating the process by which
information is perceived, interpreted, and acted upon.
Organization. Some "disorganized" behavior can be reconceptualized as a) fitting one of the
insecure strategies, b) allowing extra time for information gathering or processing, or c) enabling
the child to make transitions from one behavior pattern to another. On the surface, these suggest
organization, in some cases very sophisticated organization.
a. Behavior indicative of organized subclassifications.
Some indicators of "disorganization" appear to be organized subcategories of the three major
categories. For example, children who are extremely passive may be considered coercively
helpless (C4) from the perspective offered here. Their incompetence is seen as functioning to
shift the burden of maintaining the relationship from the dyad to the mother, with the
helplessness communicating to the mother that the child can never be left alone. Children who
show mixtures of strategies are assigned to the Defended/Coercive (A/C) classification. As with
all qualities of attachment, it is postulated here that A/C's can be either organized or
disorganized. Information other than the mixture of strategies or patterns of behavior is,
however, necessary for the determination of disorganization.
In the approach offered here, two- to six-year old children identified as "controlling" (D) using
the Cassidy and Marvin preschool system (Cassidy & Marvin with the Working Group of the
John D. and Catherine T. MacArthur Foundation on the Transition from Infancy to Early
Childhood, 1991) or the Main and Cassidy six-year reunion system (Main & Cassidy, 1988) are
seen as having adaptive and coherent patterns of behavior that fulfill the defended and coercive
strategies. Specifically, the caregiving and punitive patterns, respectively, are placed as
subclassifications of those qualities of attachment (see Appendix 1 for a fuller description of
b. Behavior indicative of information processing.
Many of the behaviors used to identify disorganization, e.g., stilling, dazed expressions, and
displacement behavior, appear to function to give children additional time for information
processing. This may involve seeking more information or further consideration of existing
information (or both). When situations are sufficiently complex, e.g., an attachment figure who is
alternately unresponsive and hostile, children may need information beyond the simple presence
of the attachment figure in order to decide what to do. In these cases, stilling may represent an
active process of organizing a response to the situation at hand; this process may or may not
imply that the underlying representational model of the self or the other is disorganized. Again, it
becomes important to remember that, to have meaning, the behavior must be viewed in terms of
the history of its use, the context in which it is displayed, and its function within that context.
c. Behavior indicative of changes in pattern of behavior.
In other situations, behaviors which have been associated with disorganization may function
to give children the opportunity to change behavior pattern (but not strategy) when changes in
the situation make the original pattern less functional. For example, in the Strange Situation, the
compliant child of a hostile mother was observed actively approaching the door to seek her
absent mother. As she did so, the door suddenly opened and the mother appeared. The child
stopped very abruptly, froze, recovered her balance, did a cute little "dance," then turned to pick
up a toy and engaged the stranger.
Using guidelines for coding disorganization, this child's freezing, dance, and change of
direction would probably be considered evidence of disorganization. From the perspective
offered here, the child's behavior would be evidence of the highly organized and flexible use of
two different patterns of behavior to implement the defended strategy of maintaining access
without closeness. In approaching the door through which the mother had disappeared, the child
(who had talked to the stranger about regaining her mother but who had not sought help) was
taking upon herself the responsibility for maintaining access to the mother. When, however, her
mother entered, it became important that she avoid the imminent possibility that they be thrown
together in very close proximity. On the other hand, simply turning away, when she was so
obviously approaching her mother, might be perceived by the mother as a blatant rebuff, thereby,
eliciting the very anger the child was seeking to avoid. Hence the "appeasement dance" before
the substitution of an acceptable alternate engagement with a toy and the stranger.
In this case, the child used a defended strategy of "close, but not too close" throughout. Her
behavior, however, changed because her mother changed from being unavailable to being
suddenly very close. The appearance of "disorganization" occurred because the child needed
both to change her behavior and also to cover that change in order to avoid appearing to snub her
mother. The appeasement dance was distracting, appealing, and protective; as such, it appeared
to reflect a high degree of organization operating in the context of very difficult circumstances.
Seeking more information, taking time to process unexpected information with care, and
creating transitions between behavior patterns do not seem inherently indicative of
disorganization of internal representational models. Rather, they suggest that the child is actively
using the model as a working model with which to analyze information before acting. In cases in
which children have reason to fear the consequences of a misstep, this analysis may be both well
worth the effort and also evidence of the extent of children's organization.
Reorganization. When developmental changes are under way and new patterns of behavior are
being incorporated into children's repertoires, extra time may be needed for children either to
consider using the new response or to initiate it in the context of competing older responses.
Similarly, when new organizations of behavior patterns are being created, behavior may "stutter"
until the new organization is refined and familiar.
The history of some of children labeled "disorganized" suggests such developmental
processes through which children learn to resolve mixed feelings and complex circumstances. As
noted above, all children experience mixed feelings some of the time; in infancy, such mixed
feelings are often expressed as ambivalent behavior, i.e., crying, fussing, resisting, and passively
Children with sensitive mothers quickly learn that communicating these feelings enlists
their attachment figure's aid in resolving them. Children with consistently insensitive mothers
learn to end their emotional discomfort through denial, i.e., avoidance, of their need for contact
with their mothers. Others, whose attachment figures are comfortable with close bodily contact
but who often unresponsive to infant signals, combine evidence of anger with bids for proximity
to the mother (Tracy & Ainsworth, 1981). This group of ambivalent (C) infants appears to have
no strategy for coping with their feelings; they cannot enlist the mother's support and yet she
remains sufficiently responsive that they do not block her out with a defensive strategy.
Two paths seem open for resolving this situation. One is to await the cognitive maturity
associated with pre-operational cognition. With the ability to recognize more complex relations,
C-type preschool-aged children may be expected to notice the systematic effects of their protest
and helpless behavior on their attachment figure and to use that information to organize the
behavior into a goal-corrected coercive strategy.
Other ambivalent infants, presumably those with relatively less sensitive mothers, may
resort to an avoidant strategy as their only means of reducing the discomfort of conflicting
feelings. The change of some 12 month old ambivalently attached infants to avoidance at 18
months may be indicative of this pattern (Egeland & Farber, 1984; Egeland & Sroufe, 1981).
Later some of these avoidant toddlers may come to recognize the power of affective
communication to influence adult behavior and begin to reorganize their procedural internal
representational models on the basis of their new understanding. As preschool-aged children,
they may then use an organized coercive strategy. Signs of confusion and disorientation are
likely to accompany all of these shifts.
A second type of reorganization is possible; in this case, children develop new behavior
patterns to fit existing strategies. In the process, they may look "disorganized". Some avoidant
(A1-2) infants may look "disorganized" during the process of developing compulsively
caregiving (A3) and compulsively compliant (A4) patterns of behavior. Others who were
resistant (C1) or passive (C2) are likely to look "disorganized" as they develop punitive (C3) and
helpless (C4) patterns of behavior. In both cases, the underlying strategy remains the same
whereas the pattern of behavior used to implement the strategy becomes more finely attuned to
the specific features of children's situations and to their increasingly mature cognitive and
These cases suggest that behavioral signs of "disorganization" should sometimes be
viewed as evidence of maturation. Testing this hypothesis would require classifying (or
reclassifying) a longitudinal set of infant and preschool tapes using the approach offered here.
Disorganization. In some cases, however, long-term disorganization of internal
representational models may be a more probable explanation of children's behavior. When
children cannot control their behavior, when extra time does not lead to more appropriate
behavior, when children seem caught between inconsistent alternative models which cannot be
resolved into a single coherent model, and when they appear to have no model through which to
interpret their present situation, disorganization may be an appropriate explanation.
Children who experience truly unpredictable environments may be expected to show true
disorganization as might some children who experience massive and disruptive changes of
conditions. Inability to reorganize, however, would be the critical condition for the designation
of "disorganization", rather than "reorganization".
Even when environments are consistent, children with neurological, mental, sensory, or
perceptual handicaps may be less able to draw meaning from experience and to organize
responses on the basis of that meaning. Such children may be delayed in their ability to process
information or may process information differently than other individuals. Where such
impairment exists, the internal representational models of relationships which children construct
may be expected to be more frequently disorganized or even entirely absent.
Implications of organization for development. Using this perspective, disorganization would
appear not to be defined solely by behavior. Nor would mixed or conflicting feelings be
sufficient to determine the presence of disorganization. As indicated above, conflicting feelings
are universal; it is the means of resolving them which is of interest to attachment theorists. The
critical question appears to be whether individuals' internal representational models are coherent,
consistent, and responsive to changes in child competencies and environmental conditions.
Carried even further, the notion of organization becomes less a state, i.e., the state of being
organized, than a process, i.e., the process of perceiving and organizing information in ways that
From this perspective, the indicators of "disorganization" identified by Main and her
colleagues can be viewed as evidence of the accommodation of children to their environments. If
the child is intact and open to new information, if assistance is given by someone trusted as a
model and supporter, i.e., an attachment figure, and if the risks attendant to inappropriate
responses are small, minor adjustments to the model will probably be made almost continuously
leaving little or no behavioral evidence of the process of change. However, even under such
favorable conditions, children may experience conditions, e.g., major separations, death, so
removed from prior experience that their models offer no interpretation of them. In such cases, it
is hypothesized here that even the models of securely attached children may be disrupted
(Cicchetti & Schneider-Rosen, 1986; Cummings & Cicchetti, 1990; Greenberg & Speltz, 1988).
In these situations, the process of reorganization may become visible for a period of time.
On the other hand, not all children experience supportive environments. For some, external
conditions are highly unstable. Such children experience frequent and disruptive changes, e.g.,
changes of residence, family membership, etc. which disrupt the fit of models to experience
(Lyons-Ruth, Zoll, Connell, & Grunebaum, 1989). For others, the insensitive behavior of their
attachment figure leads to defensive patterns of processing information (Crittenden, 1990). In
such cases, subtleties of experience may be blocked from processing until the lack of fit between
child and environment becomes so great that the strategy no longer functions. Other children
cannot effectively use their attachment figure for guidance in how to interpret novel situations; in
such cases, attachment figures provide few affective cues for children to reference in social
situations and/or offer few suggestions to children regarding how to cope with experiences. In
other cases, the risk associated with trying new responses is so great that exploration of
interpersonal relationships becomes almost impossible (Crittenden, 1990).
It seems plausible that, given such difficult circumstances, children whose models are
relatively more open will engage in the process of reorganization more frequently than will more
fully defended children. Compared to children in more favorable circumstances, the process of
accommodation may be expected to be carried out by anxiously attached children with greater
effort and disruption and to be more readily visible.
Turning to aspects of children which may impede the process of accommodation, it is
apparent that not all children are equally intact. The most difficult situations, as all clinicians
know, occur when an individual who lacks adequate coping resources must face the challenges
of a complex environment (Cicchetti, 1990). Disorganization, in such cases, may be unavoidable.
In conclusion, it appears that the concept of disorganization might best be used to refer to the
process of organizing experience into internal representational models. Behavior considered by
Main and her colleagues as indicative of "disorganization" would then be reinterpreted as a
signal that models were actively under construction or revision.
The outcome of that process requires interpretation in terms of the function of behavior in
the context of children's development, environment, and biological potential. Even when models
are highly complex or the behaviors to be interpreted using models very unusual, models can still
be organized. On the other hand, very complex models may be expected to be derived primarily
from distorted realities. These tend to be destructive to development and to limit children's
adjustment to current experience and ability to organize future experience.
Assessing Quality of Attachment Through Situations That
Activate Attachment Behavior in the Preschool Years
An ideal situation for the assessment of procedural models of quality of attachment in the
preschool years should 1) present children with situations that activate the attachment system, 2)
permit attachment figures to function as attachment figures, and 3) provide opportunities for
children to display strategies for gaining protection, coping with stress, and for exploring the
environment. Situations such as the Strange Situation that present children with an unexpected
and impending separation and that focus adults' attention on their role as attachment figure
(rather than teacher, playmate, disciplinarian, etc.) meet these criteria.
Separations used in observational procedures should be unexpected because dyads will have
developed private patterns for handling familiar separations, e.g., preschool, daycare. Activation
of these familiar patterns may deprive the observer of information regarding critical aspects of
quality of attachment. Moreover, attachment figures should not be told what to say, or not to say,
when departing so that dyads' patterns of negotiation can be observed. Separations should be
impending in order to give children time to respond before the attachment figure leaves and
becomes unavailable for negotiation. Finally, attachment figures should not be given directions
that encourage teaching, playing, or disciplining children because these activities will divert
adults' (and observers') attention from the attachment role.
The extent of the separation is probably relatively unimportant as long as children perceive
that the attachment figure will be unavailable for a time. This alone should be sufficient for
attachment behavior to be activated. Attachment figures, however, should not be gone so long as
to cause most children to become extremely upset. (Under intense stress, all children lose access
to strategic behavior). In situations meeting these criteria, children's strategy for negotiating a
set-goal can be observed as can strategies for managing feelings of distress.
The need for a stranger in an observational procedure involving preschool-aged children has
been debated. Observations of Strange Situations with and without strangers suggest strongly
that attachment behavior is not generally activated by leaving preschool-aged children with a
stranger (Greenberg & Marvin, 1982; Marvin, 1977). Being left alone, however, tends to activate
the attachment system in almost all preschool-aged children. Thus, it is probably quite possible
to observe the attachment system in a procedure which contains reunions following "child alone"
On the other hand, the presence of the stranger provides information regarding the ways in
which children manage feelings. Children who are distressed have the option of using the
stranger for distraction, comfort, information, and/or instrumental access to the attachment
figure. In order to observe these reactions, however, it is important that the stranger not initiate
interaction with children unless comfort is clearly needed. Furthermore, the stranger should not
provide information regarding the attachment figure. Instead, the stranger can confirm the child's
observation (e.g., I know, your mommy isn't here now), reassure the child without discussing the
mother (e.g., I am here now. Can I take care of you?), or query regarding the child's feelings
(e.g., Are you worried?). The presence of the stranger also permits observation of children's
exploration of social relationships.
Securely attached children may be expected to use the stranger in one or more of the ways
listed above. Defended children, by refusing to acknowledge their feelings, are limited to using
the stranger as a distraction. Coercive children, when mildly distressed, tend to be coercive with
the stranger and, when more strongly distressed, to be unable to accept the stranger at all. Some
coercive children, however, become more relaxed than they were in the attachment figure's
presence and play quite comfortably with the stranger. This suggests the possibility of multiple,
person-or role-specific models of how to interact with others (see Crittenden 1990 and Lynch &
Cicchetti, 1991 for more extensive discussion of multiple models).
The theory presented here has direct implications for immediate adaptation, for long-term
adaptation under various and changing circumstances, and for the mental health of individuals
differing in quality of attachment (Ainsworth, 1984). In addition, it has implications for
intervention and treatment in cases of maladaptation.
Immediate adaptation. If the behavior of attachment figures associated with each quality of
attachment in the preschool years is similar to that identified for infants, then it appears that each
strategy outlined here maximizes children's probability of being protected by their attachment
figures (Crittenden, in press, a). That is, all of the strategies function to promote survival and
each represents successful accommodation to the constraints of children's actual situations.
Both theory and preliminary data suggest that the strategy of securely attached children is
highly flexible and organized so as to maximize the potential of children to a) learn to care for
themselves, b) gain support from their attachment figures, and c) enlist the protection of others.
Open, direct, and reciprocal patterns of communication, combined with the ability to take the
attachment figure's perspective, form the basis of this flexibility. The theory does not, however,
necessarily imply that Secure Comfortable (B3) children either are the best protected or feel the
most secure as all securely attached children appear to enjoy these advantages. Hypotheses
regarding any benefit associated with being a B3 would require testing using longitudinal
designs and subjects both who experience the sorts of change in circumstances predicted to call
for flexibility of behavior and strategy and also who vary in culture and innate, i.e.,
Because their attachment figures are less sensitively responsive, both defended and coercive
children make compromises in how they manage protection and feelings. Defended children
appear to bear, with little or no help from others, the primary responsibility for managing their
protection and their feelings. Because attachment figures are not included in defended children's
plans and, based on prior experience, cannot be expected to monitor conditions for their children,
all conditions carry some risk and evoke feelings of insecurity. In this situation, defended
children select a strategy that compromises their protection for the benefit of reducing their
otherwise constant feelings of distress.
The strategy of coercive children makes the opposite compromise. Coercing their attachment
figures, through threats and bribes, into responding to them focuses on the evolutionary problem
of maintaining the attachment figure's protectiveness without resolving the accompanying
emotional turmoil. On the contrary, the intensity of the display of affect is both so emotionally
arousing and so cognitively distracting that coercive children often lose sight of the true problem,
i.e., being safe and feeling secure. This preoccupation can interfere with their generating, or even
noticing, alternative solutions. In such circumstances, the appraisal function of feelings, which
initially led to the activation of the attachment system, is disconnected from outcomes, which
remain tied to the child's coercive strategy. Refusing help from a comforting stranger, even to the
point of retreating from access to the attachment figure, is an example of interference of the
strategy with resolution of the problem.
Long-term adaptation. Long-term adaptation refers to the ability of children to respond
appropriately to changed circumstances. Such circumstances include loss of attachment figures
and availability of additional attachment figures as well as normative developmental change in
children. Clearly, the three strategies and associated patterns of behavior have very different
probabilities of fostering adaptive coping in the event of changed circumstances (Crittenden, in
press, a). Long-term adaptation also refers to the preparation of children to assume in adulthood
the role of attachment figure to a child (Crittenden, 1984).
Once again, the advantage is to secure children. With communication and negotiation,
securely attached children are able to co-ordinate with others to develop new coping skills to
meet the demands of changed conditions and to establish new relationships (Bretherton, 1990;
Elicker & Sroufe, in press; Hazan & Shaver, 1987; Kobak & Sceery, 1988; Lynch & Cicchetti,
1991; Sroufe & Fleeson, 1986). Their experience, with a loved person, of an enduring shared
state of mind and a mutual concern for one another that transcend physical proximity facilitates
their recognition of the potential for creating similar relationships with other people. Assuming a
continued favorable course of development, these achievements, together with the development
of empathy, appear to bode well for the secure child's preparation to become a competent
Defended children have little access to their feelings as a means of appraising situations and
little willingness to communicate those feelings to others. These jeopardize their ability to
recognize changed conditions or to engage the assistance of others in adapting to those
conditions. Their increasing language skills and emerging ability to construct semantic models of
relationships may lead to schisms among representational models. For example, a procedural
model of maternal intrusiveness and child defensive behavior may dominate their behavior in
spite of their describing verbally the relationship as loving. As a consequence, they are unlikely
to modify appropriately their own behavior or to select future attachment figures who interact
On the other hand, their ability to take the perspective of others may enable them to fulfill the
wishes of adults, thus, providing a modicum of protection. The limited development of empathy
among defended children, the repeated experience of being rejected by attachment figures, and
the failure to acknowledge feelings suggest a developmental trajectory which is not conducive to
the development of the sensitive responsiveness needed to become an attachment figure to a
securely attached child.
Coercive children, on the other hand, are exquisitely aware of any feelings indicative of
increased risk and communicate the urgency of those feelings most effectively. They are limited,
however, in their ability to take the perspective of others and to develop joint plans which
contain workable compromises between their desires and those of adults. Their highly charged
emotional behavior and refusal to compromise reduces the help from others which they will
accept, prevents their learning the skills necessary for self-reliance, and often drives help-givers,
including attachment figures, away. The result is that coercive children struggle to maintain the
status quo rather than learning to adapt to new situations. The unwillingness of coercive children
to accept responsibility for themselves or to empathize with others, if extrapolated into
adulthood, suggest limitations to their ability to become responsive and responsible attachment
figures to children. It also suggests the probability of their selecting mates who will be equally
unable to function as a competent and protective attachment figure (Crittenden, Partridge, &
Claussen, in press; Daly & Wilson, 1981).
It should not be assumed, however, that past experience determines future development. As
others have pointed out, events themselves can lead to changes in quality of attachment
(Cicchetti & Schneider-Rosen, 1986; Main & Hesse, 1990). More importantly, inclusion of
aspects of developmental change into the theory presented here implies that changes in quality of
attachment may occur as a result of how either children or attachment figures manage
developmental change. For example, a mother who was sensitively responsive to her securely
attached infant may resent her preschool-aged child's attempt to establish a goal-corrected
partnership. By becoming unwilling to communicate, negotiate, and compromise, such mothers
are insensitive to their older child's needs and competencies. This insensitivity may lead to a shift
in their children's quality of attachment from secure to anxious.
The reverse may also occur. For example, an infant in daycare whose mother is
sensitively responsive may, nevertheless, be anxiously attached because of the mother's
inexplicable (to the infant) unavailability. With the cognitive advances of the preschool period
and his mother's willingness to establish a goal-corrected partnership, this child may shift from
anxious to secure. Conditions favoring shifts in quality of attachment need as much consideration
as do the conditions favoring continuity.
Mental health. Mental health can be defined in many ways. This discussion will be limited to
considering how quality of attachment is related to 1) feelings of security and competence, 2) the
ability of individuals to establish and maintain intimate and satisfying relationships, 3) the ability
to resolve internal conflict, and 4) the portrayal and valuation of the self.
Although a full discussion of these issues is beyond the scope of this paper, a few points are
relevant. First, only securely attached children have the advantage of feeling secure most of the
time. In addition, they can feel competent to care for themselves, either through their own efforts
or through enlisting the help of others. More than anxiously attached children, securely attached
children are able to apply all of their faculties, i.e., feelings, attention, perceptions, cognitions, to
the challenges of life, whereas defended and coercive children block out and/or distort feelings,
perceptions, and cognitions and/or focus attention in ways which limit their development.
Moreover, defended and coercive children cope with underlying feelings of anxiety and doubts
about their competence.
Second, intimate relationships are particularly difficult for anxiously attached children because
intimacy depends upon a balance among self-disclosure, perspective-taking, and empathy and
because, in revealing their true feelings, children risk experiencing further emotional pain.
Ironically, the very adaptations (i.e., distorted perceptions, lack of emotional availability, and
defensive and angry behavior) used by children to protect themselves from the pain of disturbed
relationships may be expected to increase the probability of future unsatisfying relationships.
Third, where internal emotional conflict exists, these same conditions are likely to prevent
children from being able to acknowledge and respond to all aspects of the situation, thereby
reducing the probability of satisfactory resolution of conflicted feelings. On the contrary, it is
likely that the mixed feelings both will continue to disturb children and result in inconsistent
behavior which will confuse, mislead, and ultimately dismay others.
Finally, and most important, anxiously attached children have learned that a full presentation
of themselves to their attachment figure results in rejection and/or disregard. Consequently, they
find their "real" selves to be ineffective at meeting both basic life needs, i.e., protection, and also
deeply felt desires to be loved, nurtured, and appreciated. Consequently, both defended and
coercive children construct, through distorted displays of affect, a false self which is more
acceptable to their attachment figure. In ways which clinicians have documented extensively,
children who present only a false self, and who distort feeling, perception, and response in order
to maintain a false self, feel pain and anger at the denial of their real self and, often, shame at its
existence. Even with recognition of the situation (i.e., awareness of distortion in semantic
models), their distortions in information processing and reliance upon procedural models to
inform behavior frequently prevent change. A fuller understanding of the process through which
these distortions occur and through which they may be changed is needed.
Treatment. There are three important implications of the theory on quality of attachment
in the preschool years for treatment of anxious relationships. First, it is suggested that, because
similar overt behavior can serve different strategies, overt behavior alone is insufficient to
determine the origin, meaning, or appropriate response to children's behavior. Instead, the
behavior should be interpreted in the context of its display and function. This will lead to a range
of interpretations and corresponding interventions. For a discussion of an application of this
point to preschool classrooms, see Crittenden (1989).
Second, because individuals' strategies are likely to be used in many of their
relationships, it is expected that families with one seriously disturbed relationship will
experience many such relationships. Certainly, this appears to be true for maltreating families
(Crittenden, 1985b, 1988a, 1987; Jean-Gilles & Crittenden, 1990; Lynch & Cicchetti, 1991).
Consequently, assessment of all family relationships following identification of one distorted
relationship as well as family-based interventions are necessary (Bretherton, in press; Marvin &
Finally, the development in the preschool years of semantic and episodic memory
systems suggests the need for multiple types of intervention. Teaching mothers more sensitive
ways to interact (Crittenden, 1985a; Crittenden & Snell, 1983) can be expected to modify
mothers' and children's procedural models. In addition, effort should be made to guide the
development of emerging semantic and episodic models in ways that reduce the potential for
perceptual exclusion of information, misattribution of the meaning of information, or
strategically limited responses. In other words, intervention strategies should be developed for
each of the internal representational models. Elsewhere, a theory for integrating behavioral,
family systems, cognitive, and psychodynamic approaches around this goal is presented
(Crittenden, in press, b).
Implications of Quality of Attachment
in the Preschool Years for Theory
The study of attachment in the preschool years expands attachment theory beyond the firm
theoretical and empirical base in infancy toward the more uncharted older ages. The preschool
years are a particularly advantageous age upon which to focus because they bridge the transition
from sensorimotor cognition to the rudiments of symbolic and conscious thought. In addition, the
greater flexibility of preschool-aged children in employing a wide range of behaviors and
patterns of behavior in service of different strategies is representative of the behavioral
complexity of individuals at later ages. Aspects of preschooler's behavior, such as using
morphologically similar behaviors for different functions and regulating affective displays, are
readily applied to older individuals. In this way the preschool years provide an unusually wide
window on later development.
Of particular importance are the hypotheses regarding the processing of information.
Defended behavior has been explored quite extensively with reference to avoidant infants and
from a psychoanalytic perspective in terms of defense mechanisms. A contribution of this paper
may be to expand that strategy to include compulsively caregiving and compulsively compliant
patterns of behavior. A second contribution may be to conceptualize the ambivalent infant group
in terms of the development in the preschool years of a coercive strategy. Its complexity and
substantial emotional demands suggest that, for many children, it may require more cognitive
sophistication than the defended inhibited pattern of behavior.
The theory presented here is also important because it suggests the nature of the process by
which attachment becomes related to the wide variety of correlates and consequences already
identified by follow-up studies of infants. In particular, the processes of affect regulation,
negotiation, and interaction of behavior systems around protection and exploration suggest how
widely disparate areas of development may come to be related. The development of styles of
information processing and the impact of patterns of behavior on the interpersonal environment
that children create for themselves have been suggested but need far more elaboration. This
perspective is, however, limited by its almost exclusive focus on procedural representational
Finally, this view of attachment in the preschool years can help to bridge the gap between
research on attachment and clinical practice. Although many researchers working in the area of
infant attachment have felt that infant quality of attachment was relevant to clinical work, most
clinicians have been substantially less convinced. Specifically, clinicians often find it difficult to
map attachment classifications onto standard diagnostic categories. The approach outlined here
offers direct relations to clinical work through specification of the processes of management of
feelings, regulation of display of affect, negotiation, and exploration. Such processes constitute
aspects of behavior upon which clinicians can focus intervention.
Implications of the Method Used for Developmental Psychopathology
Historical trends. The historical roots of developmental psychology lie primarily in
questions about early development raised by pathological outcomes. In other words, a central
impetus for the study of development, as well as the origin of several theories of development,
was the need to treat disorders of development. The early methods of study consisted primarily
of 1) reconstruction of developmental processes derived from clinical case histories and 2)
philosophical reasoning about the nature of development.
In the increasingly empirical climate of twentieth century science, these approaches are
seen as far too subjective. Moreover, in spite of elegant theory and constructs, they resist
reduction to hypotheses that can be tested empirically. In addition, they often defy replication
and elimination of alternative explanations for the phenomena. Consequently, these approaches
have largely been abandoned by the research community.
They have been replaced by a more empirical approach involving two shifts in emphasis.
First, methodology has become a dominant aspect of research. The use of controlled group
designs, implemented with as few biases as possible, and tests of statistical significance have
resulted in better specified developmental theory and more reliable and valid results. Second,
partly because such controlled studies benefit from homogeneous samples, developmental
enquiry has been directed primarily toward normative, non-pathological samples. Studies of
pathology, on the other hand, tend to be carried out by clinicians with little emphasis on
developmental processes (Gelfand & Peterson, 1985; McMahon & Peters, 1985).
As a consequence of these changes, some of the richness and scope of the early theories
has been lost. In the extreme, the emphasis on limited types of designs and subject populations,
together with the use of ever more sophisticated statistical procedures, can result in sterile
findings generated by "blind" investigators who depend on statistics to give meaning to their
findings. In the words of Bowlby, excessive emphasis on methodological issues has sometimes
resulted in our "knowing more and more about less and less" (Bowlby, 1988, p.65).
An integration is needed which includes both a focus on disorders and the use of
subjective and intuitive methods and also study of basic, normative processes and scientific
rigor. Such an integration will capitalize on the dual abilities of the human mind to construct
meaning from experience and to test empirically the validity of that construction.
Participant observers. Developmental psychopathology provides this opportunity by
combining a tradition of objective and rigorous research with the more subjective experience of
clinically treating disorders of development. In addition, new theories, many of which employ
systemic principles derived from treatment, have the potential to integrate meaningfully the
(subjective) experience of the investigator with maintenance of research standards. As a result,
our knowledge of both normal and atypical development is enhanced (Cicchetti & Toth, 1990;
Garmezy & Masten, 1986; Lieberman, 1991).
In particular, the role of the investigator in determining hypotheses, creating methods of
obtaining relevant data, and interpreting results has been overlooked in the emphasis on objective
empiricism. Recent trends toward incorporation of systemic perspectives and emphases on affect
provide a basis, however, for fresh consideration of the role of investigators.
One of the advantages of studying development in the context of pathology is that
investigators have the opportunity to experience their data. Clinical studies often compel
investigators to be involved both because the complexity of the phenomena require very close
and extensive observation to be comprehended and because the affective experience of subjects
draws forth an affective response from observers. Consequently, clinical investigators often
know their subjects by name, by life history, and by their own affective responses.
Instead of trying to eliminate, or deny, that response, systems theories provide a basis for
using it productively to generate new conceptualizations of the interpersonal processes affecting
development. Because both subjects' and investigators' expectations, i.e., internal
representational models, affect their perceptions and responses, clinical investigators working
with disturbed samples constantly find themselves forced to accommodate the skewed models of
their subjects. Doing so involves investigators affectively as well as cognitively and provides
unique opportunities for investigators to discover and share subjects' experience. Although this
process is possible with all subjects, the greater dissonance between disturbed subjects' models
and investigators' increases the probability that investigators will become aware of, and explore,
the differences. Moreover, exaggerated dissonance can be expected to facilitate discovery of
processes which are displayed more subtly in normative samples.
Although such contact and involvement is routinely used in treatment (as transference
and countertransference), it has been treated methodologically in research as contamination.
Nevertheless, it is the stuff from which insights are drawn. Sharing an emotion with another
person or stretching to understand the unexpected and inexplicable aspects of a distressed
person's behavior throws light on both their functioning and one's own. In addition, it creates a
reservoir of experience that extends the limits of one's understanding.
The accepted means of generating hypotheses is consideration of the body of literature
that precedes one's own work. In a field based solely on logic and empiricism (for example,
mathematics), this may be sufficient. It is not sufficient, however, for studies having to do with
affect, particularly interpersonal affect. These are enhanced greatly when the investigator
experiences the phenomena of interest and then consciously reconstructs that process both
through the lens of one's own representational models and, vicariously, through the lens of
subjects' models. If the self is known well enough, subjects' unusual behavior can be recalibrated
in terms of it. In this context, the "intuitive" leap of affective and cognitive understanding
The reward is that the normal is used to experience the deviant and expands our
understanding of basic developmental processes. Clinically, this process is both very rich and
very personal. As such, it falls in the realm of subjective speculation. Tested as specified
hypotheses using sound research methods, however, it possesses the potential to enlighten the
interpretation of data and, thus, to enhance greatly our knowledge base and theories of
An application. The theory presented here reflects precisely that process. In this case, I
intervened for three years with at risk families with young children. Later, I carried out an
extensive research project on family relationships with similar families. In the research, I
arranged to interact with almost every family by being the stranger in the Strange Situation and
by closing our contact by thanking and paying the parents. These interactions permitted me to
experience something of the context in which each family member interacted with the others.
This was done purposefully so that I could feel the effect of their personalities in
interaction with my own. In other words, I used the only lens any of us has on experience, my
own personality, to "test" the nature of interpersonal relationships with the subjects. To protect
the data from my biases, I chose to do this after the data to be analyzed were collected and by
being the stranger. Doing so, facilitated my integration of experience with these subjects with my
broader knowledge of other families.
Many years later, the videotapes of the Strange Situations of the preschool-aged children
in this sample were viewed with other investigators, each of whom brought unique personal
perspectives, as well as additional videotapes, to the viewing. This, together with case history
material and clinical knowledge, created a rich environment for generating and testing
hypotheses regarding developmental pathways. Behavior observed in the Strange Situation could
be compared with other parent-child behavior as well as with my experiences interacting with the
child and with the mother. Case history material, particularly information abuse or neglect,
provided external validation of the meaning of identified patterns of behavior. Had we tried to
interpret the meaning of the behavior in the Strange Situation using only procedures specified a
priori and without case history material or experiential and clinical knowledge, it is unlikely that
the theory offered here would have been generated.
Although the test of this thinking can, and must, be carried out using rigorous empirical
standards on new samples, its basis was in a broad range of experience with families. A
particularly salient aspect of that experience was affective response to the videotapes and, in my
case, to the individuals when I interacted directly with them. In other words, by studying our
experience of their experience through a systemic theory focussed on process, a set of
hypotheses regarding pathways of development was derived.
Very little research refers to investigators' own experience with subject populations. Yet
in many, possibly most, studies of pathological or risk samples, such experience exerts a
profound impact on the questions framed by investigators and the means chosen to explore them.
Not only should this be acknowledged, it should be explored purposefully with the intent of
maximizing its potential to focus research in productive directions.
The ideas presented here represent an expansion into the preschool years of theory developed
from the study of infants. Although that theory is broadly supported by empirical studies, the
theory has yet to be empirically tested. It has been developed, however, through a process of
extensive clinical and empirical work with preschool-aged children on whom extensive
concurrent and/or longitudinal data were available. Observation of a wide range of children (and
families) who differed in terms of race, ethnicity, nationality, and adequacy of functioning has
been especially valuable in constructing theory relevant to the development of most children. It
is hoped that this approach will help to enhance understanding of the developmental process in
normal and atypical preschoolers.
Ainsworth, M.D.S. (1990). Epilogue: Some considerations relevant to theory and
assessment regarding attachment beyond infancy. In M. Greenberg, D. Cicchetti, and
E.M. Cummings (Eds.), Attachment in the preschool years: Theory, research, and
intervention (pp. 463-488). Chicago: University of Chicago Press.
Ainsworth, M.D.S. (1967). Infancy in Uganda: Infant care and the growth of love. Baltimore:
Johns Hopkins University Press.
Ainsworth, M.D.S. (1984, April). Attachment, adaptation, and continuity. Paper presented at
the International Conference on Infant Studies, New York City.
Ainsworth, M.D.S. (1979). Infant-mother attachment. American Psychologist, 34, 932-937.
Ainsworth, M.D.S. (1985). Patterns of infant-mother attachment: antecedents and effects on
development. Bulletin of the New York Academy of Medicine, 61, 771-91.
Ainsworth, M.D.S., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment: A
psychological study of the Strange Situation. Hillsdale, NJ: Erlbaum Assoc.
Ainsworth, M.D.S., & Bowlby, J. (1991). An ethological approach to personality
development. American Psychologist, 46, 333-341.
Belsky, J., Rovine, M., & Taylor, D.G. (1984). The Pennsylvania Development Project, III:
The origins of individual differences in mother-infant attachment: Maternal and infant
contributions. Child Development, 55, 719-728.
Bowlby, J. (1969/1982). Attachment and loss. Vol. I: Attachment. New York: Basic Books.
Bowlby, J. (1973). Attachment and loss. Vol. II: Separation. New York: Basic Books.
Bowlby, J. (1980). Attachment and loss. Vol. III: Loss. New York: Basic Books.
Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human development.
New York: Basic Books.
Bretherton, I. (1991). Pouring new wine into old bottles: The social self as internal working
model. In M. Gunnar and L.A. Sroufe (Eds.), Self processes and development: The
Minnesota Symposia on Child Development (pp. 1-41). Hillsdale, NJ: Erlbaum.
Bretherton, I. (1990). Open communication and internal working models: Their role in the
development of attachment relationships. In R.A. Thompson (Ed.), Nebraska Symposium
on motivation: Socio-emotional development (pp. 57-113). Lincoln, NE: University of
Bretherton, I. (in press). Theoretical contributions from developmental psychology to the
development of family theory. In P. Boss, W. Doherty, R. LaRossa, W. Schumm, & S.
Steinmetz (eds.) Sourcebook of family theories and methods: A contextual approach.
New York: Plenum Press.
Bronfenbrenner, U. (1979). The experimental ecology of human development. Cambridge,
MA: Harvard University Press.
Campos, J.J., Campos, R.G.., & Barrett, K.C. (1989). Emergent themes in the study of
emotional development and emotion regulation. Developmental Psychology, 25, 394-
Cassidy, J., & Marvin, R.S., with the Working Group of the John D. and Catherine T.
MacArthur Foundation on the Transition from Infancy to Early Childhood (1991).
Attachment organization in three- and four-year olds: Coding guidelines. Unpublished
manuscript, University of Virginia.
Cicchetti, D. (1990). The organization and coherence of socioemotional, cognitive, and
representational development: Illustrations through a developmental psychopathology
perspective on Down syndrome and child maltreatment. In R. Thompson (Ed.), Nebraska
Symposium on Motivation, Vol. 36. Socioemotional development (pp. 266-375).
Lincoln, NB: University of Nebraska Press.
Cicchetti, D., Cummings, E.M., Greenberg, M., & Marvin, R. (1990). An organizational
perspective on attachment beyond infancy: Implications for theory, measurement, and
research. In M. Greenberg, D. Cicchetti, and E.M. Cummings (Eds.), Attachment in the
preschool years: Theory, research, and intervention (pp. 3-49). Chicago: University of
Cicchetti, D., Ganiban, J., & Barnett, D. (1991). Contributions from the study of high risk
populations to understanding the development of emotion regulation. In J. Garber and K.
Dodge (Eds.), The development of emotion regulation and dysregulation (pp. 15-48).
New York: Cambridge University Press.
Cicchetti, D., & Greenberg, M. (1991). (Eds.) Attachment and developmental
psychopathology. Special issue of Development and Psychopathology, 3 (4).
Cicchetti, D., & Schneider-Rosen, K. (1986). An organizational approach to childhood
depression. In M. Rutter, C. Izard, & P. Read (Eds.), Depression in young people (pp. 71-
134). New York: Guilford Press.
Cicchetti, D., & Toth, S. (1991). The making of a developmental psychopathologist. In J.
Cantor, C. Spieker, and L. Lipsitt (Eds.), Child behavior and development: Training for
diversity, (pp 34-72). Norwood, NJ: Ablex.
Crittenden, P.M. (in press, a). Children's strategies for coping with adverse home
environments: An interpretation using attachment theory. Child Abuse and Neglect.
Crittenden, P.M. (1988a). Distorted patterns of relationship in maltreating families: The role
of internal representational models. Journal of Reproductive and Infant Psychology, 6,
Crittenden, P.M. (1988b). Family and dyadic patterns of functioning in maltreating
families. In K. Browne, C. Davies, & P. Stratton (Eds.), Early prediction and prevention
of child abuse (pp. 161- 189), London: Wiley.
Crittenden, P.M. (1990). Internal representational models of attachment relationships.
Infant Mental Health Journal, 11, 259-277.
Crittenden, P.M. (1985a). Maltreated infants: Vulnerability and resilience. Journal of Child
Psychology and Psychiatry, 26, 85-96.
Crittenden, P.M. (1987). Non-organic failure-to-thrive: Deprivation or distortion? Infant
Mental Health Journal, 8, 56-64.
Crittenden, P.M. (1988c). Relationships at risk. In J. Belsky & T. Nezworski (Eds.), The
clinical implications of attachment (pp. 136-174). Hillsdale, N.J.: Lawrence Erlbaum.
Crittenden, P.M. (1989). Teaching maltreated children in the preschool. Topics in Early
Childhood Education, 9, 16-32.
Crittenden, P.M. (1984). Sibling interaction: Evidence of a generational effect in
maltreating families. International Journal of Child Abuse and Neglect, 8, 433-438.
Crittenden, P.M. (1985b). Social networks, quality of parenting, and child development. Child
Development, 56, 1299-1313.
Crittenden, P.M. (in press, b). Treatment of child abuse and neglect. Human systems.
Crittenden, P.M., & Ainsworth, M.D.S. (1989). Child maltreatment and attachment
theory. In D. Cicchetti and V. Carlson (Eds.), Child maltreatment: Theory and research
on the causes and consequences of child abuse and neglect (pp. 432-463). New York:
Cambridge University Press.
Crittenden, P.M., & DiLalla, D. (1988). Compulsive compliance: The development of an
inhibitory coping strategy in infancy. Journal of Abnormal Child Psychology, 16, 585-
Crittenden, P.M., Partridge, M.F., & Claussen, A.H. (1991). Family patterns of
relationship in normative and dysfunctional families. Development and Psychopathology,
Crittenden, P.M., & Snell, M.E. (1983). Intervention to improve mother-infant interaction.
Infant Mental Health Journal, 4, 23-41.
Cummings, E.M., & Cicchetti, D. (1990). Attachment, depression, and the transmission of
depression. In M.T. Greenberg, D. Cicchetti, & E.M. Cummings (Eds.), Attachment
during the preschool years (pp. 339-372). Chicago: University of Chicago Press.
Daly, M., & Wilson, M. (1981). Child maltreatment from a sociobiological perspective. New
Directions for Child Development, 11, 93-112.
DiLalla, D.L., & Crittenden, P.M. (1990). Dimensions in maltreated children's behavior: A
factor analytic approach. Infant Behavior and Development, 13, 439-460.
Egeland, B., & Farber, E.A. (1984). Infant-mother attachment: Factors related to its
development and changes over time. Child Development, 55, 735-771.
Egeland, B., & Sroufe, A. (1981). Attachment and early maltreatment. Child Development, 52,
Elicker, J., & Sroufe, L.A. (in press). Predicting peer competence and peer relationships in
childhood from early parent-child relationships. In R. Parke & G. Ladd (Eds.), Family-
peer relationships: Modes of linkage. Hillsdale, NJ: Earlbaum.
Garmezy, N., & Masten, A.S. (1986). Stress, competence, and resilience: Common frontiers
for therapist and psychopathologist. Behavior Therapy, 17, 500-521.
Gelfand, D.M., & Peterson, L. (1985). Child Development and psychopathology. Beverly
Hills, CA: Sage.
Greenberg, M.P. (1982). Reactions of preschool children to an adult stranger: A behavioral
systems approach. Child Development, 53, 481-490).
Greenberg, M., & Speltz, M. (1988). Attachment and the ontogeny of conduct problems. In
J. Belsky and T. Nezworski (Eds.), Clinical implications of attachment (pp. 177-218).
Hillsdale, NJ: Erlbaum Assoc.
Grossmann, K., Fremmer-Bombik, E., Rudolph, J., & Grossmann, K.E. (January, 1987).
Maternal attachment representations as related to patterns of infant-mother attachment
and maternal care during the first year. Paper presented at the Conference on Intrafamilial
Relationships, Cambridge, Great Britain.
Hazen, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process.
Journal of Personality and Social Psychology, 52, 511-524.
Jean-Gilles, M., & Crittenden, P.M. (1990). Maltreating families: A look at siblings. Family
Kobak, R., & Sceery, A. (1988). Attachment in late adolescence: Working models, affect
regulation, and representations of self and others. Child Development, 59, 135-146.
Lieberman, A.F. (1991). Attachment theory and infant-parent psychotherapy: Some
conceptual, clinical, and research considerations. In D. Cicchetti & S.L. Toth (Eds.),
Rochester symposium on developmental psychopathology, Vol. 3: Models and
integration (pp. 261-287). Rochester: University of Rochester Press.
Lieberman, A.F., & Pawl, J.H. (1988). Infant mental health: A model for service delivery.
Journal of Clinical Child Psychiatry, 14, 196-201.
Lynch, M., & Cicchetti, D. (1991). Patterns of relatedness in maltreated and non-
maltreated children: Connections among multiple representational models. Development
and Psychopathology, 3, 207-226.
Lyons-Ruth, K., Zoll, D., Connell, & Grunebaum, H. (1989). Family deviance and family
disruption in childhood: Associations with maternal behavior and infant maltreatment
during the first two years of life. Development and Psychopathology, 1, 219-236.
McMahon, R.J., & Peters, R. (Eds.) (1985). Childhood disorders: Behavioral developmental
approaches. New York: Bruner/Mazel.
Main, M. (1981). Avoidance in the service of attachment: A working paper. In K.
Immelmann, G. Barlow, L. Petrinovich, and M. Main (Eds.), Behavioral development:
The Bielfeld interdisciplinary project (pp. 651-693). New York: Cambridge University
Main, M, & Cassidy, J. (1988). Categories of response to reunion with the parent at age six:
predictability from infant attachment classifications and stable across a one-month
period. Developmental Psychology, 24, 415-426.
Main., M., & Hesse, P. (1990). Lack of resolution of mourning in adulthood and its relationship
to infant disorganization: Some speculations regarding causal mechanisms. In M.
Greenberg, D. Cicchetti, and E.M. Cummings (Eds.), Attachment in the preschool years
(pp. 161-182). Chicago: University of Chicago Press.
Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood and adulthood: A
move to the level of representation. In I. Bretherton and E. Waters (Eds.), Growing points
of attachment theory and research: Monographs of the Society for Research in Child
Development. Serial N. 209, Vol. 50, Nos. 1-2, 66-104.
Main, M., & Weston, D.R. (1981). The quality of toddlers' relationship to mother and father:
Related to conflict behavior and readiness to establish new relationships. Child
Development, 52, 932-940.
Main, M., & Solomon, J. (1986). Discovery of an insecure disorganized/disoriented attachment
pattern: Procedures, findings, and implications for the classification of behavior. In M.
Yogman & T.B. Brazelton (Eds.), Affective development in infancy (pp. 121-160).
Norwood, NJ: Ablex.
Main, M., & Solomon, J. (1990). Procedures for identifying infants as
disorganized/disoriented during the Ainsworth Strange Situation. In M. Greenberg, D.
Cicchetti, and E.M. Cummings (Eds.), Attachment in the preschool years (pp. 161-182).
Chicago: University of Chicago Press.
Marvin, R.M. (1977). An ethological-cognitive model for the attenuation of mother-child
attachment behavior. In T.M. Alloway, L. Krames, and P. Piner (Eds.), Advances in the
study of communication and affect. Vol. 3. The development of social attachments (pp.
25-60). New York: Plenum.
Marvin, R.S., Marvin, C.N., & Abramovitch, L.A. (1973, April). The development of coy
expressions in young children. Paper presented at the biennial meeting of the Society for
Research in Child Development, Philadelphia, PA.
Marvin, R.S., & Pianta, R.C. A relationship-based approach to self- reliance in young
children with motor impairments. Infants and young children, in press.
Marvin, R.S., & Stewart, R.B. (1990). A family systems framework for the study of
attachment. In M. Greenberg, D. Cicchetti, and E.M. Cummings (Eds.), Attachment in
the preschool years (pp. ). Chicago: University of Chicago Press.
Radke-Yarrow, M., Cummings, 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.
Sroufe, L.A. (1983). Infant-caregiver attachment and patterns of adaptation in preschool: The
roots of maladaptation and competence. In M. Perlmutter (Ed.), Minnesota Symposia on
Child Psychology (Vol. 16, pp. 41-81). Hillsdale, NJ: Erlbaum Assoc.
Sroufe, L.A., & Fleeson, J. (1986). Attachment and the construction of relationships. In W.
Hartup & Z. Rubin (Eds.), Relationships and development (pp. 51-71). Hillsdale, NJ:
Sroufe, L.A., & Waters, E. (1977). Attachment as an organizational construct. Child
Development, 48, 1184-1199.
Tracy, R., & Ainsworth, M. (1981). Maternal affectionate behavior and infant-mother
attachment patterns. Child Development, 52, 1341-1343.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.),
Organization of memory. New York: Academic Press.
Tulving, E. (1985). How many memory systems are there? American Psychologist, 40, 385-
Tulving, E. (1979). Memory research: What kind of progress? In L.G. Nilsson (Ed.),
Perspectives on memory research: Essays in honor of Uppsala University's 500th
anniversary (pp. 19-34). Hillsdale, NJ: Erlbaum.
Patterns of Behavior
The labeling of categories is complex. The major categories in the original infant system
are identified by verbal descriptors, i.e., secure, avoidant, and ambivalent whereas the
subcategories are identified only by letters and numbers, e.g., A1, B3. Later systems have
retained the alpha-numeric subclassifications and added verbal descriptors, e.g., secure reserved.
Where prior systems have used a label applicable to the theory presented here, that label has
been retained and referenced to its source. Where another label seems to describe better the
function of the behavior, new labels have been provided. The system of category letters used by
Ainsworth has been retained as a means of quick identification and continuity across systems and
ages. In four cases, however, additional subclassifications have been added; these classifications,
i.e., A3, A4, C3, and C4, are unique to this classificatory system.
Secure Patterns of Behavior
Children classified as Secure Reserved (B1-2) (Cassidy & Marvin, 1991; Main & Cassidy,
1989) tend to be verbally direct and clear and to take the initiative in involving their attachment
figures in play or conversation but to engage in little close proximity with their attachment
figures. They seem confident and comfortable resolving issues verbally.
Children classified as Secure Comfortable (B3) are open with their attachment figures
regarding separations and/or reunions, but do not require extensive plan-making or reassurance.
They provide evidence of an assumed trust in their attachment figure's whereabouts and
willingness to communicate with them regarding their mutual set-goal. They display their
feelings without distortion and are particularly competent at regulating these feelings and
resolving discomfort. They are likely to resolve any discomfort upon reunion by sharing some
form of intimacy with the attachment figure, e.g., long eye contact, emotionally intimate
dialogue, co-ordinated and close body positioning. They are the most relaxed, calm, and
comfortable of the groups of children; in fact, their easy assurance with the relationship can be
overlooked in expectation of something more demonstrative.
Children classified as Secure Reactive (B4) tend to want more reassurance and closeness than
children classified as Secure Comfortable. They present their feelings openly, express some
doubts regarding their own competence, and request more help with affect regulation than
children in other subclassifications of the secure pattern. Like other secure children, however,
they trust their attachment figure's willingness to communicate with them, take responsibility for
regulating their own affect, explore the environment and affiliative relationships, and use both to
assist with affect regulation. Behaviorially, they often seek physical proximity even in
preseparation episodes, are quite likely to cry when left alone, and tend to engage in less
exploration than other secure children. They are unlikely, however, to be persistently pouty or
Themes of anger, fear, and helplessness may be present. In spite of these concerns, however,
secure reactive children present their fears directly to their attachment figures and either resolve
the issue themselves or seek and receive reassurance from their attachment figures. In contrast to
children classified as anxious, any fear expressed is fear of what will happen if the child is
unprotected rather than fear of the attachment figure. It is significant that these children are not
separated from their attachment figure by these concerns, but instead openly seek their
attachment figure's support in resolving them.
Defended Patterns of Behavior
Children classified as Inhibited (A1-2) avoid close contact or bids for close contact with an
attachment figure who generally behaves in an interfering and rejecting manner (i.e., physically
distancing). Because the goal is to maintain proximity without causing the attachment figure to
become angry, avoidant children do not make demands, signal that the relationship poses a
problem, or deny maternal bids for attention. Because preschool-aged children recognize that
pointed avoidance would signal the presence of a problem as well as offend the attachment
figure, their avoidance may be expected to become more covert, i.e., a maintenance of neutrality
often displayed as engagement in appropriate alternate activities (Cassidy & Marvin, 1991; Main
& Cassidy, 1989). Thus, inhibited preschool-aged children usually appear focussed on other
acceptable activities, such as toy play, which allow them both to remain in proximity to their
attachment figure and also to be so sufficiently and justifiably occupied that their attachment
figure could not expect them to interact closely.
Occasionally, however, they resist imposed physical intimacy by squirming or pushing away;
such behavior is not, however, offered gratuitously or as a threat. Rather it is a last ditch effort to
avoid intimacy. When resistance occurs in an inhibited child, it is likely, upon achievement of
greater distance, to be followed by appeasing behavior.
Other defended children, classified as Caregiving (see Bowlby, 1980) for a discussion of the
defended caregiving pattern), employ a different pattern of behavior in response to attachment
figures who behave in a withdrawn and unresponsive manner (i.e., psychologically distant). They
fear becoming unable to arouse their attachment figure more than they fear the attachment
figure's rejection. Moreover, because there are no bids from the attachment figure for them to
avoid, they need to create at least a minimal level of availability. To do this, many develop a
caregiving pattern of behavior (Cassidy & Marvin, 1991; Main & Cassidy, 1989; Main &
Solomon, 1986, 1990).
This pattern includes attempting to cheer the attachment figure with overbrightness or
nurturance, simulating involvement by carrying on a monologue directed rhetorically toward the
attachment figure, and keeping the attachment figure busy. In extreme cases, caregiving children
appear brightly frantic and compulsively active. Unlike inhibited children, they may initiate or
tolerate closeness in an attempt to please the attachment figure. In all cases, however, the
positive behavior appears brittle. Underlying all of these behaviors is the anxiety and anger
associated with not having an available and sensitively responsive attachment figure.
Like the inhibited pattern of behavior, the caregiving pattern both provides a defense against a
direct confrontation with, and loss of access to, the attachment figure and concurrently protects
children from the pain of having the attachment system aroused without satisfaction.
The caregiving pattern may be mixed with the inhibited pattern. Particularly when attachment
figures of caregiving children take the initiative, formerly caregiving children may shift to
distancing behavior. In either case, the strategy is one of maintaining access without the threat of
rejection associated with emotional intimacy.
A third group of defended children, classified as Compliant, defend against the attachment
figure's hostile displays of anger by becoming compliant (Crittenden, 1988c; Crittenden &
DiLalla, 1989). These children tend to be vigilant and over-responsive to any suggestion of
demands from the attachment figure. Because of the children's extreme readiness to comply,
their attachment figures rarely appear demanding or controlling. The children's limited range of
affect (most often guarded but also including fear, overbrightness, and sadness, together with the
inhibition of expectable negative affect) and their inability to engage in productive play are
partial signals of this pattern. Like caregiving children, compliant children will often tolerate
extended episodes of closeness rather than offend their attachment figure. When they fear having
offended the attachment figure, compliant children usually display appeasing behavior.
Coercive Patterns of Behavior
Coercively threatening (C1) children use angry behavior (e.g., ranging from poutiness,
whining, and openly displayed anger) in order to threaten the attachment figure into compliance
with children's wishes. Such behavior serves as a powerful deterrent to non-compliance by
Coercively disarming (C2) children use coy and winsome behavior (see Marvin, Marvin, &
Abramovitch, 1973) to bribe the attachment figure into rescuing the child. This behavior ranges
from shy (e.g., standing with head down, thumb in mouth, fiddling with cloths) to sweetly
flirtatious (e.g., coy looks, whispered entreaties, high babyish voice tone), to seductively
disarming (e.g., sudden glorious smiles, proffered gifts, lispy tendernesses). In all cases,
however, there is evidence of an underlying struggle to force the attachment figure to meet
children's wishes on the child's terms. Such behavior serves as a powerful incentive to
compliance for the attachment figure.
Coercively punitive (C3). Some children carry threatening behavior to an extreme by
creating situations in which the attachment figure is openly punished (e.g., painful hitting,
tantrumming, furious screaming, impenetrable isolation to an entreating attachment figure) or
shamed or embarrassed (e.g., refusal of affection, demands that the attachment figure say or do
babyish, futile, or foolish things before the child will respond). Such behavior appears hostile
and retaliatory in nature.
Coercively helpless (C4). At other times, coercive children display extreme helplessness.
This behavior (e.g., sitting frozen and motionless with lowered head, to pitifully helpless,
whimpered calls to attachment figure using "baby talk") functions as both as a reproach and a
demand that the parent change before the child will make any attempt to resolve the situation.
Helpless children often appear to be resolutely helpless as if hoping that their intense
vulnerability will finally penetrate their attachment figures' awareness and evoke an empathic