sroufe

Development and Psychopathology, 11 (1999), 1–13Copyright 1999 Cambridge University PressPrinted in the United States of America

Implications of attachment theory fordevelopmental psychopathology

L. ALAN SROUFE, ELIZABETH A. CARLSON, ALISSA K. LEVY, AND

BYRON EGELANDUniversity of Minnesota

AbstractBowlby’s attachment theory is a theory of psychopathology as well as a theory of normal development. It containsclear and specific propositions regarding the role of early experience in developmental psychopathology, theimportance of ongoing context, and the nature of the developmental process underlying pathology. In particular,Bowlby argued that adaptation is always the joint product of developmental history and current circumstances (nevereither alone). Early experience does not cause later pathology in a linear way; yet, it has special significance due tothe complex, systemic, transactional nature of development. Prior history is part of current context, playing a role inselection, engagement, and interpretation of subsequent experience and in the use of available environmentalsupports. Finally, except in very extreme cases, early anxious attachment is not viewed as psychopathology itself oras a direct cause of psychopathology but as an initiator of pathways probabilistically associated with later pathology.

From its inception, attachment theory was a ory of outcome, but a theory of process. Inthe attachment trilogy and other writings hetheory of psychopathology as well as a theory

of normal development. It was concerned presented a very specific set of propositionsregarding the way in which early experienceboth with the formation and normal course of

attachment relationships and the implications contributed to psychological health or pathol-ogy. He began in the first volume of the tril-of atypical patterns of attachment. Bowlby’s

(1944) early consideration of 44 thieves re- ogy (1969) by clearly dissociating his individ-ual differences construct from causal traitvealed a consistent background of early pa-

rental privation in the lives of these young constructs. Attachment referred to a pattern oforganized behavior within a relationship, notmen. He reasoned that this was no mere coin-

cidental association but, rather, had causal im- a trait infants had in varying quantity(Sroufe & Fleeson, 1986; Sroufe & Waters,plications. Still, he doubted from the start that

the connection was simple, direct, and lin- 1977). Attachment patterns were not immuta-ble and not independent of subsequent experi-ear—an environmental malignancy with an

inevitable outcome. What, then, was the na- ence. In Volume 2 (1973) he wrote that thedevelopment of the individual “turns at eachture of this link? Attachment theory evolved

in large part to answer this question. and every stage of the journey on an interac-tion between the organism as it has developedWhat Bowlby proposed was not just a the-up to that moment and the environment inwhich it then finds itself” (p. 364, italics

Preparation of this work and the research described added). Early experience frames, but also isherein were supported by a grant from the National Insti- transformed by, later experience. In brief, hetute of Mental Health (MH 40864).

proposed what would now be called a dy-Address correspondence and reprint requests to: Alannamic systems theory of psychopathology,Sroufe, Institute of Child Development, University of

Minnesota, 51 East River Road, Minneapolis, MN 55455. based on a complex interaction of constituents

1

L. A. Sroufe et al.2

over the course of development (e.g., Samer- Cause in Attachment Theory:Organizational Construct versus Traitoff, 1989).

Still, within this theory a special role forearly experience in the initiation of important Cause is complex within attachment theory.

In accord with ecological views such as thoseprocesses was specified. Thus, attachmenttheory provides a third alternative to the of Belsky (e.g., Belsky & Isabella, 1988) and

Bronfenbrenner (1986), the child is seen asshadow box debate about whether early expe-rience causes later outcomes in the manner of nested within a network of influences operat-

ing on many levels. Some contextual influ-an immutable trait or bears only a coinciden-tal association to outcome due to its link to ences impact directly on the child, some in-

directly through their impact on parenting.later experience or third factors such as SES(Fox, 1997; Lamb, 1984; Lewis, 1997). The Developmental context is emphasized, be-

cause “changes in circumstances can lead toanswer is, neither. Early experience oftenplays a critical role in the developmental dy- changes in interaction and therefore to

changes in relationships” (Vaughn, Waters,namic that yields pathology, but this role isdependent on a surrounding context of sus- Egeland, & Sroufe, 1979, p. 974). At the same

time, the child’s history of experience is ataining environmental supports. At the sametime, processes engendered in the context of critical part of the developmental context.

There is an ongoing transaction between theearly experience may influence the nature oflater experience and the surrounding context developing child and changing circumstances

(cf. Sameroff & Chandler, 1975). The impactitself. In this paper, core propositions from at-tachment theory regarding the complex rela- of current circumstances depends on the pat-

tern of behavioral and emotional organizationtion between early experience and psycho-pathology will be elaborated. These include the child brings forward to that phase of de-

velopment. Child and context are mutuallyformulations concerning causation (and theinterplay between the individual and context), transforming. While we argue below that

early experience has special significance, stillthe role of early experience, and the nature ofearly disturbance. it “ . . . cannot be more important than later

experience, and life in a changing environ-This will be primarily a conceptual paper,tracing the theoretical implications of the ment should alter the quality of a child’s ad-

aptation” (Sroufe, 1978, p. 56). The individ-Bowlby/Ainsworth attachment theory for de-velopmental psychopathology. Data presented ual is the product of all of his or her

experiences, not early experiences alone.will be for the purpose of illustrating thesetheoretical ideas, and will be drawn primarily Many of Bowlby’s ideas regarding the

roles of prior experience and current circum-from the Minnesota longitudinal parent–childstudy. Previously published data will not be stances in adaptation and in psychopathology

are summarized within his concept of devel-presented in detail but will be accompaniedby relevant citations. Many of these data and opmental pathways (Bowlby, 1969, 1973).

Metaphorically captured by the dispersion andthose of other researchers have been reviewedelsewhere (e.g., Weinfield, Sroufe, Egeland, interconnection of tracks in a railway yard or

branchings on a tree, this model embodies& Carlson, in press). Data that have not beenpublished before will be presented more com- several key ideas (Sroufe, 1997). First, there

are more lines or branches in the broad centerpletely, along with relevant statistical infor-mation. Likewise, we will be unable to review of the larger array (there is great diversity in

normality). Second, beginning on any majorextensively topics such as representation andexperience-dependent brain development, trunk allows a large number of possible out-

comes due to subsequent branchings (multifi-which are related to the issues pursued but arenot central in this discussion. Such topics will nality). Thus, ongoing circumstances may

support pursuance of potentially deviating de-be brought in where pertinent, and citations ofrecent comprehensive reviews will be pro- velopmental pathways or deflect the individ-

ual back toward more normal adaptation. En-vided.

Implications of attachment 3

joining a pathway even early on does not infant attachment alone with the predictivepower of attachment plus cumulative assess-determine final outcome but only initiates a

set of possibilities. Cause is probabilistic not ments of quality of care in the preschoolyears. While anxious attachment in infancy bydeterministic. Third, the longer an outlying

pathways is followed, the more unlikely be- itself predicted a K-SADS psychopathologyindex (number and severity if past and presentcomes a return to centrality. In this theory

psychopathology results from a successive se- diagnoses; see Sroufe, 1997), those with a his-tory of cumulative unsupportive care showedries of adaptations. A pattern of anxious at-

tachment in infancy may initiate such a pro- significantly more problems (13% vs. 15% ofvariance accounted for). In addition, we foundcess, but only if subsequent adaptations

continue to represent deviation from positive that assessments of parent–child relationshipproblems at age 13 years (especially boundaryfunctioning does psychopathology become

likely. Change remains possible but, Bowlby dissolution and lack of emotional support orsupport for autonomy) and anxious attach-(1973) argues, becomes quite difficult by ado-

lescence if development has continued to go ment in infancy were more powerful than ei-ther alone, though each was significant sepa-awry.

Several specific, testable hypotheses may rately (see Table 1). Likewise, ongoingassessments of peer relationship problemsbe derived from these initial formulations: (a)

At any age, current quality of care will add to added to predictions of psychopathologybased on attachment alone, R2 change (2, 166)early attachment history in predicting pathol-

ogy, given that adaptation is always the joint = .04, p < .01. Similar analyses have shownthat attachment history and peer experiencesproduct of current circumstances and early

history. (b) Likewise, broader aspects of cur- in preschool or middle childhood together arebetter predictors of adolescent social compe-rent contexts, including relationships outside

of the family and stresses and challenges of tence than attachment alone (Englund,Levy, & Hyson, 1997; Sroufe, Egeland, &the period, also will increase prediction be-

yond early attachment; (c) a cumulative his- Carlson, in press).Finally, in a number of papers we havetory of maladaptation will be more pathogenic

than a single early period of poor functioning, evaluated the predictability of change. For ex-ample, in an early study we showed that qual-with pathology ever more likely the longer a

maladaptive pathway has been followed, and ity of attachment itself changed in meaningfulways between 12 and 18 months, given(d) change itself will be predictable in light of

changes in stress and/or support. changes in parental life stress (Vaughn et al.,1979). Later, Erickson, Egeland, and SroufeWe and our colleagues have examined

these hypotheses by carrying out a broad- (1985) showed that those with histories ofanxious attachment sometimes showed fewerbased prospective, longitudinal study of 180

children, assessing quality of care in infancy, behavior problems than predicted in pre-school. Such deflections in developmentalearly childhood and adolescence, along with

assessments of relationships with peers and course were best predicted by increased sta-bility of social support for the primary care-teachers, life adversities, and social support.

Based on parent, self, and teacher behavior giver.In summary, Bowlby did not say, nor doesproblem checklists at age 16 (Achenbach &

Edelbrock, 1986) and a clinical interview at attachment-oriented research suggest, thatearly anxious attachment causes later pathol-age 17.5 years (the child form of the Schedule

for Affective Disorders and Schizophrenia; K- ogy. He did claim that pathology would be ajoint product of early experience and ongoingSADS) as the sources for psychopathology

outcome, regression analyses were utilized to support or challenge, that cumulative malad-aptation would be less easily changed thanevaluate various combinations of variables.

For example, to test the hypothesis regarding early anxious attachment, and change wouldbe predictable. Each of these propositions hasthe significance of cumulative adaptational

history, we compared the predictive power of been amply supported by empirical research.


Table 1. Hierarchical regression predicting ratings of psychopathology (K-SADS)at 17.5 years from avoidant and disorganized attachment in infancy andfamily relationship quality in early adolescence (n = 120)

OverallR 2

Step Independent Variables Change β B T R 2 F df

1 Avoidant attachment score(12–18 months) .07 .27 .74 3.00** .07 8.99** 1, 119

2 Avoidant attachment score .18 .51 2.02*Disorganization rating

(12–18 months) .06 .25 .20 2.71** .13 8.41*** 2, 1183 Avoidant attachment score .18 .18 2.04*

Disorganization rating .27 .27 3.03**Relationship rating

(13 years) .05 .22 .21 2.59* .17 8.12*** 3, 117

Note: Index of psychopathology in adolescence is 7-point rating of number and severity of K-SADS diag-noses.***p < .001. **p < .01. *p < .05.

In concluding this section, it is important assessment of the coherence of discourseconcerning attachment (George, Kaplan, &to point out that the proposition regarding the

power of cumulative adaptation does not Main, 1985; Main & Goldwyn, in press). Cau-tion should be exercised in using it as a mea-mean that change cannot occur in adolescence

or adulthood. Main, Kaplan, and Cassidy sure of history. There currently exists no em-pirical evidence as to the accuracy of reported(1985) argued that the advent of formal opera-

tional thought may promote new opportunities early memories on the AAI. Second, even ifveridical, which it seems likely to be to somefor reflection upon, evaluation of, and integra-

tion of past experiences. Moreover, research extent, the AAI asks people to reflect on theirmemories from ages 5 to 12 years. At present,with the Adult Attachment Interview (AAI)

has uncovered numerous individuals who, de- there is no evidence that adults have verbalaccess to experience from infancy. It is suchspite indications of difficult life experiences,

nonetheless have autonomous “states of mind” preverbal experiences that Bowlby (1973)emphasized in his discussion of the enduringconcerning attachment (so-called “earned se-

cures”; Pearson, Cohn, Cowan, & Cowan, effects of early attachment relationships. Fi-nally, Main (personal communication) has1994; Phelps, Belsky, & Crnic, 1998).

Clearly, some individuals have integrated suggested that some number of “earned se-cures” may actually have had a positive initialvery difficult experiences, and change can oc-

cur in the years of maturity. foundation, with a secure attachment relation-ship in infancy, but adverse experience laterSuch late-occurring changes are not

counter to Bowlby’s idea that change be- in development. We will present evidence inthe next section supporting the role of earlycomes more difficult the longer the history of

maladaptation. Positive changes may, in fact, foundations as supports for positive change inadolescence.not be likely in the face of continuous adver-

sty and maladaptation from infancy throughadolescence. When late changes occur, theymay be built upon earlier foundations. Earned The Dynamic Role of Early Experiencesecurity on the AAI cannot really speak to thisissue. First, the AAI was not designed to be a Attachment theory is a structural develop-

mental theory, wherein subsequent develop-measure of early history, but rather as an


they can affect the environment. . . . (Sroufe, 1978,ment is conceived as building upon as well asp. 45)transforming what preceded (Sroufe, 1996).

In this systemic, dynamic position, psycho-pathology is viewed as a complex, organiccreation, not the simple sum of positive and Substantial research confirms the idea that

children with varying attachment historiesnegative experiences. Early experience, there-fore, has special significance because it construe the environment differently. Such

differences are revealed, for example, in theirframes the child’s subsequent transactionswith the environment. “The child not only in- completions of stories with separation themes

(Bretherton, Ridgeway, & Cassidy, 1990),terprets experience, the child creates experi-ence. As Alfred Adler has suggested, the child their social pretend play (Rosenberg, 1984),

their reactions to cartoons depicting potentialis both the artist and the painting” (Sroufe,1978, p. 57). From Freud, Bowlby drew the social conflict (Suess, Grossmann, & Sroufe,

1992), their reactions to family photographsidea of the primacy of the earliest attachmentrelationships as the first experiences of emo- (Main, 1993), their family drawings (Fury,

Carlson, & Sroufe, 1997; Main, 1993), andtional closeness. These vital relationshipsrepresent prototypes for close relationships their memories for affective-cognitive stimuli

(Belsky, Spritz, & Crnic, 1996; Rieder & Cic-throughout life, especially for intimate loverelationships and parenting. chetti, 1989). These studies show that those

with secure histories are less likely to attributeBowlby centered his own ideas on his con-cept of “internal working models.” Children hostile intent in ambiguous social situations

or to reject stimuli portraying their parents,inevitably extract from their experience ex-pectations regarding likely behavior of others and more likely to bring fantasied conflicts to

successful resolution and to see themselves asand themselves in relationships. Humans can-not keep themselves from doing this. “The connected to others, especially family mem-

bers.varied expectations of the accessibility and re-sponsiveness of attachment figures that differ- While the emphasis here has been upon

cognitive frameworks, it is fully compatibleent individuals develop during the years ofimmaturity are tolerably accurate reflections with recent writing on the formative impact

of early experience for brain system develop-of the experiences those individuals have ac-tually had” (Bowlby, 1973, p. 202). ment (e.g., Cicchetti & Tucker, 1994; Schore,

1994) and on early relationships as entrainingWhile simple and straightforward, this is aprofound idea. It means that children ap- basic patterns of emotional regulation (Sroufe,

1996, 1997). All of these levels of analysisproach new situations with certain precon-ceptions, behavioral biases, and interpretive are mutually supporting.

It is also the case that expectations and bi-tendencies. Thus, context, even new circum-stances and new arenas, are not independent ases of the child often lead to and therefore

are borne out by environmental reactions in aof the child’s history. As we have said else-where, self-perpetuating manner. For example, push-

ing others away (even if it is at some level toavoid disappointment, see below) frequently

If because of early experience the preschooler iso- leads in fact to the rejection that was ex-lates himself from the peer group, he removes him- pected. Children with histories of avoidant at-self further from positive social experiences. . . . If tachment not only expect rejection from oth-self-esteem and trust are established early, children

ers, they are, indeed, often rejected by themmay be more resilient in the face of environmental(Sroufe, 1983; Suess, Grossmann, & Sroufe,stress. They may show poor adaptation during an1992). In general, individual patterns of adap-overwhelming crisis, but when the crisis hastation elicit reactions from the environmentpassed and the environment is again positive, theythat consolidate and elaborate them. The pos-may respond more quickly. Even when flounder-

ing, some children may not lose their sense that itive feedback cycles generated by malad-


aptation are, in fact, a key part of its defini- tachment in infancy (children, in part, cre-ate their own environments).tion.

From the perspective of attachment theory,3. Similarly, reactions of others, includingthen, several mechanisms underlie continuity

those outside the family, will be predictablein adaptation (Sroufe, 1988). Continuity infrom infant attachment patterns.both environmental influences (e.g., quality of

care) and individual characteristics supports4. Even following change, early patterns of at-stability in individual functioning over time.

tachment retain a potential for reactivation.Prior history also supports continuity in adap-There is a homeorhetic tendency (Sameroff,tation, because continuity is viewed as a trans-1989).actional process involving an active self-regu-

lating organism and the environment (Sroufe &5. Certain issues and certain arenas of func-

Egeland, 1991). From this perspective, thetioning—those tapping anxiety about the

child actively participates in constructing hisavailability of others or apprehension re-

or her experience. Individuals behave in waysgarding emotional closeness—will be espe-

that elicit responses from the environmentcially likely to reveal the legacy of early

that support prior adaptation, and individualsattachment, even during periods of gener-

make choices that selectively engage aspectsally adequate functioning.

of the environment supporting a particularadaptive style. Individuals also interpret newand ambiguous situations in ways that areconsistent with earlier experience. These hypotheses are difficult to test and

will require decades of longitudinal research.Finally, as implied in many points above,early experiences may have special signifi- Often quite detailed observational data are re-

quired (especially for hypothesis 5), alongcance because of their nature. Because theyare preverbal, they are not accessible to verbal with broad assessments of ongoing support

and life stress. Still, beginnings have beenrecall and may be less readily modified bysubsequent experience. This is the essence of made and some clear directions have been set.

The regression analyses described aboveFreud’s idea of the “dynamic unconscious”(Loevinger, 1976). A basic sense of emotional already imply that early experience adds to

later assessments in predicting outcome, be-connectedness, confidence regarding theavailability of others, and feelings of self- cause attachment remains significant after

later variables are entered. This is confirmedworth may be the legacy of infancy. Whilesuch a basic orientation may, of course, be when we enter variables in reverse chronolog-

ical order with attachment last (for previouslyaltered by later experience, it might still bereactivated in certain circumstances or in cer- published data, see Sroufe, Egeland, & Kreut-

zer, 1990). For example, in predicting adoles-tain domains of functioning.Specific hypotheses that may be derived cent pathology (see above), quality of care at

age 13 is significant, accounting for 4% of thefrom these theoretical ideas are as follows:variance. Nonetheless, when infant attach-ment history (avoidant and disorganized) is

1. Early attachment history will have ongoing subsequently entered, it also is significant, ac-importance for later socioemotional adapta- counting for an additional 14% of the vari-tion, even after taking into account current ance. Early attachment also adds when the to-circumstances and intermediary experi- tal prior equation includes behavior problemsences. in grades one through six and quality of care

at age 13 years (see Carlson, 1998). Similarresults also are obtained when the outcome2. The child’s manner of engaging the envi-

ronment in subsequent developmental peri- variable taps adolescent competence withpeers.ods will be predictable from patterns of at-


Our detailed studies of a subsample ob- and friendship selections, which were concor-dant for attachment history, confirmed thatserved in nursery school, summer daycamp in

middle childhood, and weekend retreats in ad- peer experiences also were predictable fromearly attachment differences (e.g., Elicker etolescence further clarify the ongoing role of

early experience (e.g., Elicker, Englund, & al., 1992).One way we showed that early experienceSroufe, 1992; Sroufe, 1983; Sroufe, Carl-

son, & Shulman, 1993; Sroufe, Egeland, & remains available even while latent is to ex-amine groups of children who while function-Carlson, in press). Two variables that have

been clearly and consistently related to anx- ing similarly for a period are distinguished byearly history. Two such groups were estab-ious attachment history are isolation and de-

pendency (hovering near and/or being in- lished during the preschool years. For both,adaptation was consistently negative across 3volved with teachers or counselors at the

expense of peer interaction). Self-isolation ap- assessments from 42 to 54 months, yet, mem-bers of one group had earlier histories of se-pears to be more frequent in those with histo-

ries of avoidant attachment; orientation to- cure attachment. Follow-up study in elemen-tary school revealed a rebound to significantlyward adults is more frequent in those with

resistant attachment histories. better functioning (fewer behavior problems)for those with secure histories (Sroufe et al.,Reactions of peers, teachers, and counse-

lors to children with different attachment his- 1990). Thus, the resiliency of these childrenwas predictable; their rebound toward positivetories also have proven predictable. This may

be due to the obvious dependency and lack of adaptation was latent in the secure base thatpreceded the problems of the preschool pe-resourcefulness of those with resistant histo-

ries and the alienation and, at times, hostile riod.We have shown similarly that change inaggressive behavior of those with avoidant

histories. The strongest data here come from behavior problems from the late elementaryyears to adolescence is predictable from earlyour preschool study where children were

filmed daily over the course of the term. For history. To parallel and expand on the pre-school study, we created groups of childrenexample, from the films a random sample of

at least 50 interactions for each child with who were functioning comparably in terms ofbehavior problems during elementary school,each of the two head teachers was edited onto

a tape. Coders blind to attachment history and but who had differed in their attachment his-tories. A total of four groups was created, in-all other data made ratings of dimensions of

teacher treatment (Sroufe & Fleeson, 1988). cluding children with stable secure histories(at 12 and 18 months) who were functioningFor those children having secure histories, rat-

ings were consistently higher on “expecta- well or poorly in middle childhood and chil-dren with stable insecure histories who weretions for compliance” and lower on “control”

than for other groups. The resistant group was functioning well or poorly in middle child-hood (based on the CBCL in Grades 1–3).1distinguished by high ratings on “control,”

“tolerance for rule infraction,” and “nurtur-ance” and low ratings on “expectations for 1. Children were considered to be functioning well in

middle childhood if they had total scores in the norma-compliance.” The avoidant group elicited hightive range (T ≤ 55) on the Teacher Report Form of theratings on “control” and low “warmth,” “ex-Child Behavior Checklist (Achenbach & Edelbrock,

pectations,” and “nurturance” ratings, and 1986) for at least two of three assessments from firstthey were the only group that elicited anger. through third grade and also in sixth grade. Children

were considered to be functioning poorly in middleSuch reactions may be seen as perpetuatingchildhood if they had total scores in the upper rangethe immaturity of those with resistant histo-(T ≤ 62) for at least two of four assessments in firstries and the expectations for rejection of thosethrough third and sixth grade; sixth grade scores also

with avoidant histories. The same basic pic- had to be at least above the group mean. These criteriature resulted from observations in later child- were established to insure consistency in behavior or

adaptation across all of middle childhood.hood as well. In addition, peer sociometrics,


Behavior problem status in middle childhood to rebound somewhat from a difficult middlechildhood. This finding suggests some specialwas paralleled by peer competence and emo-

tional health rankings made by teachers. role for early experience, as its effects ap-peared after a long passage of time andFor the analyses of subsequent adolescent

adaptation, the critical comparisons were be- seemed to be as potent as later experience inpredicting adolescent outcomes. At the sametween the groups of children who looked the

same in middle childhood, but who differed time, children with insecure histories wereamenable to positive change, as those whoin their early attachment status. These com-

parisons revealed that the groups differed in were doing well in middle childhood werealso functioning fairly well in adolescence.later psychopathology and competence in

ways predictable from their early histories. Finally, we find that those with secure his-tories show less negative reaction to periodsAmong the groups of children who were do-

ing well in middle childhood, the ones with a of high family stress both in middle childhood(Pianta, Egeland, & Sroufe, 1990) and in ado-secure attachment history had lower scores on

present, t (44) = 2.66, p < .01, and past, t (42) lescence (Sroufe, Levy, & Carlson, 1998). Re-silience and vulnerability are best reviewed in= 2.70, p < .01, pathology indices derived

from the K-SADS at age 17.5 years. They process terms, with foundations beginning tobe laid down in the earliest years (e.g., Ege-also scored significantly higher on a global

rating of competence at age 19 years based land, Carlson, & Sroufe, 1993).Our data, and the literature in general, pro-on their functioning across work, school, and

relationship domains, t (42) = 2.91, p < .006. vide only hints regarding the hypothesis ofspecial vulnerabilities or particular domainsSimilarly, for those children who were func-

tioning poorly in middle childhood, the ones of impact that are the legacy of early experi-ence. In recent studies of adolescent groupwith a history of secure attachment scored

lower on total, t (28) = −2.99, p = .006, and functioning (including variables such as tak-ing a leadership role, the capacity for interper-past, t (29) = 2.54, p = .02, pathology on the

KSADS. Thus, as suggested earlier, individu- sonal vulnerability, and self-reported intimacyin friendships; e.g., Englund et al., 1997; Os-als showing positive change in adolescence,

following a period of maladaptation, drew toja, McCrone, Lehn, Reed, & Sroufe, 1995),we frankly were surprised at the degree ofupon a more positive foundation in the in-

fancy period. predictability of results from infant attach-ment assessments. At times these surpassedLooking across all four groups, level of

functioning in adolescence appeared to de- predictions to preschool and middle child-hood. We propose that this may be due to thepend on both early and later experience. The

children who had secure histories and who special capacities for emotional investment ofself that are called for during adolescence thatwere also functioning well in middle child-

hood were consistently significantly higher tap into early attachment experience in a waythat earlier measures of competence do not.than all other groups in their competence rat-

ings and lower in their pathology ratings. We are only now beginning to assess roman-tic relationships and parenting in the next gen-Conversely, the children with insecure histo-

ries who had behavior problems in middle eration, which will be crucial tests of attach-ment theory. Studies using the AAI (George,childhood were significantly lower than all

other groups in adolescent competence and Kaplan, & Main, 1984), which has beenlinked to infant attachment history in somehigher in pathology. Notably, the two groups

of children with mixed histories (secure at- samples but not our own, are suggestive.Autonomy (security) on the AAI has beentachment and later behavior problems or inse-

cure attachment and later positive function- linked to the capacity to use a romantic part-ner as a secure base (Crowell, 1997) and toing) were comparable to each other on both

adolescent competence and behavior prob- secure attachment in infants (van Ijzendoorn,1995).lems. A positive early foundation appeared to

be a protective factor, allowing the children More directly pertinent to psychopathology


would be differential vulnerability to major tently responsive caregivers. Even behavior ofinfants classified as disorganized (stereotypes,loss experiences. In the third volume of the

trilogy, Bowlby (1980) argued that early inad- simultaneous approach/avoidance) enables in-fants to maintain proximity in the context ofequate attachment history would leave indi-

viduals vulnerable to depression in the face of frightening caregiver behavior and internalconflict (Main & Hesse, 1990).such losses (see also Brown, Harris, & Bi-

fulco, 1986). Our sample has proven too small Distortions of early dyadic regulatory pro-cesses serve as prototypes for later dysregula-(losses of primary caregivers too few) to test

this hypothesis as yet. Frequency of unre- tion, markers of a process that leaves individ-uals vulnerable to normative stresses and thesolved losses for those who are autonomous

(secure) and nonautonomous on the AAI development of pathology. Early maladaptiverelationship patterns are internalized and car-might provide a partial test.ried forward as characteristic modes of affec-tive regulation and associated expectations,

The Nature of Early Relationshipattitudes, and beliefs. Emerging develop-

Disturbancemental capacities (cognitive, linguistic, be-havioral) become organized with respect toBowlby’s work emphasizing the quality of

early adaptation and continuity in experience patterns of restricted regulation or emotionaldysfunction (Carlson & Sroufe, 1995). Withprovides a framework for conceptualizing

early relationship disturbances and their links development, individuals construct morecomplex and elaborated types of defensive co-to psychopathology (Sroufe, 1986). From this

perspective, disturbed early attachment is ordination.Variations in early regulatory patterns pro-most profitably viewed as a marker of a be-

ginning pathological process that probabilis- vide the basis for differences in later strate-gies for coping with normative stresses, elicit-tically leads to later pathology (Sameroff &

Emde, 1989). Thus, while in the context of ing support from others and making use ofinternal signals (Carlson & Sroufe, 1995). Forextreme deprivation or maltreatment, disturb-

ance within the child may be manifested at individuals for whom the caregiver has beena source of effective emotional regulation andan early age (infancy or toddlerhood), in most

cases early disturbance lies within the dyadic comfort, relationships are valued. From theearliest relationship, the individual has experi-relationship and only gradually takes the form

of enduring disturbance within the child. In enced the rudiments of reciprocity, includinghow to receive care and respond empathicallyour view, many disturbances of childhood

may be viewed as having relational origins to others (Sroufe & Fleeson, 1986). Thechild’s expectations about self and others andwherein patterns of dyadic emotion regulation

in time are carried forward by the child and consequent behavior elicit feedback that sup-ports a particular adaptive style. Under condi-manifested in individual styles of coping with

challenge. tions of extreme stress, the child is likely toseek comfort and support with the expectationWithin this framework, anxious attachment

patterns are viewed as dyadic regulatory pat- that others will be available and will to pro-vide aid.terns that maximize to the extent possible op-

portunities for the infant to maintain closeness In contrast, individuals with histories of in-secure attachment may be more likely to form(secure base behavior) in the context of un-

available or intermittently available and unre- relationships that are not supportive and areeasily disrupted (Carlson & Sroufe, 1995).sponsive caregiving. Avoidant infants main-

tain proximity to the caregiver (in case of For the avoidant child, early experiences sup-port a view of the self as isolated, unable toextreme threat) by minimizing signals of dis-

tress and negativity that may alienate a reject- achieve emotional closeness, and unworthy ofcare. Social relationships may be viewed asing caregiver (Main, 1981). For infants classi-

fied as resistant, heightened distress signals alien and treated with hostility. With elevatedstress, the child may fail to seek comfort fromserve to maintain the attention of intermit-


others, perpetuating a view of relationships as states, may be disrupted (Carlson & Sroufe,1995). (Empirically, we have found that disor-alien or hostile. For the child with a history

of unpredictable or inconsistent caregiving ganized attachment in infancy, followed bysubsequent trauma, is the most potent combi-experiences, an anxious style, in which nega-

tive emotions disrupt rather than restore rela- nation for predicting dissociation in earlyadulthood; Ogawa, Sroufe, Weinfield, Carl-tionships, inhibits the development of stable

close relationships. Individual continuity in son, & Egeland, 1997.) Recurring trauma inthe context of inadequate or overly restrictivesuch patterns results, in part, because such

nonconscious, underlying processes are no caregiving increases the level of arousal andneed to separate or compartmentalize over-longer a part of conscious social interchange

and, as a result, no longer subject to environ- whelming affects and memories. Dissociativemechanisms that serve survival functions bymental feedback and revision.

For children with avoidant or resistant his- providing for the automatization of behaviors,escape from reality constraints, containmenttories, emotions that would have facilitated

affective communication and exchange are of overwhelming memories or feelings, and inextreme form the assignment of painful feel-defensively modified or cut off (Carlson &

Sroufe, 1995; Kobak, Ruckdeschel, & Hazan, ings to alternate personalities are strengthened(Ludwig, 1983; Putnam, 1989). Variations in1994). As a result, when experiencing distress

the child may fail to signal directly a need for outward manifestations result from differ-ences in individual psychological capacities,support, become embroiled in negative emo-

tion, and be unable to draw from potentially caregiving experiences, and social and cul-tural factors (Ludwig).supportive social relationships. Moreover, for

individuals with insecure attachment histories, In summary, early disturbances in attach-ment relations, while themselves generallyworking defensive strategies of avoidance or

resistance may themselves be vulnerable to not best construed as psychopathology, oftendo lay the foundation for disturbances in de-breaking down under stress (Kobak & Shaver,

1987). This is evident in low ratings of ego velopmental processes which can, indeed,lead to psychopathology. Understanding theresiliency, inability to cope with frustration,

and pervasive presence of negative affect of processes wherein what begins as a relation-ship disturbance can in time lead to individualinsecure children (Erickson et al., 1985;

Sroufe, 1983). In the context of stressful life disorder is one of the central tasks for the fieldof developmental psychopathology (Sam-events, these children may be faced with emo-

tional conflicts that their defensive strategy eroff & Emde, 1989).was organized to control but now are over-whelming (Carlson & Sroufe, 1995). Symp-

Conclusiontomatic forms of attachment behavior andother problems may result. Hypochondriacal Within attachment theory, psychopathology is

viewed as a developmental construction, re-reactions may serve needs for closeness orcare, or suicidal behavior may represent a de- sulting from an ongoing transactive process as

the evolving person successively interactssire to reunite with a lost attachment figure(Bowlby, 1980). Other potential links be- with the environment. Individual transforms

environment but also is transformed by it.tween anxious attachment and adult disorderare described by Dozier and her colleagues In this perspective early attachment varia-

tions generally are not viewed as pathology(Dozier, Stovall, & Albus, in press).For individuals with histories of extremely or even as directly causing pathology. Rather,

varying patterns of attachment represent “ini-harsh or particularly chaotic caregiving con-texts (disorganized attachment relationships), tiating conditions” in systems terms. In this

regard, they do play a dynamic role in patho-the process of regulation, the consolidation orintegration of self across behavioral states and logical development because of the way in

which environmental engagement is framedacquisition of control over modulation of


by established tendencies and expectations. leads inevitably to psychopathology. Changeremains possible at numerous points in devel-Moreover, patterns of infant–caregiver attach-

ment and other aspects of early experience opment, although both theory and data sug-gest that such change is more readily accom-may have a special role in the developmental

process via their impact on basic neurophysio- plished early in the process or at least whenthere is a foundation of early support.logical and affective regulation.

Still, anxious attachment by no means

References

Achenbach, T., & Edelbrock, C. (1986). Manual for The Society for Research in Child Development, Wash-ington, DC.Teachers’ Report Form and Teacher Version of the

Child Behavior Profile. Burlington, VT: Department Dozier, M., Stovall, K. C., & Albus, K. (in press). Attach-ment and psychopathology in adulthood. In J. Cas-of Psychiatry, University of Vermont.

Belsky, J., & Isabella, R. A. (1988). Maternal, infant, and sidy & P. R. Shaver (Eds.), Handbook of attachment:Theory, research, and clinical implications. Newsocial-contextual determinants of attachment security.

In J. Belsky & T. Nezworski (Eds.), Clinical implica- York: Guilford.Egeland, B., Carlson, E. A., & Sroufe, L. A. (1993). Re-tions of attachment (pp. 41–94). Hillsdale, NJ: Erl-

baum. silience as process. Development and Psychopathol-ogy, 5, 517–528.Belsky, J., Spritz, B., & Crnic, K. (1996). Infant attach-

ment security and affective-cognitive information Elicker, J., Englund, M., & Sroufe, L. A. (1992). Predictingpeer competence and peer relationships in childhoodprocessing at age 3. Psychological Science, 7, 111–

114. from early parent–child relationships. In R. Parke &G. Ladd (Eds.), Family–peer relationships: Modes ofBowlby, J. (1944). Forty-four juvenile thieves: Their

characters and home life. International Journal of linkage (pp. 77–106). Hillsdale, NJ: Erlbaum.Englund, M., Levy, A., & Hyson, D. (1997). Develop-Psychoanalysis, 25, 19–52 and 107–127.

Bowlby, J. (1973). Attachment and loss: Vol. 2. Separa- ment of adolescent social competence: A prospectivelongitudinal study of family and peer contributions.tion. New York: Basic Books.

Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss. Poster session presented at the biennial meeting of theSociety for Research in Child Development, Wash-New York: Basic Books.

Bowlby, J. (1982). Attachment and loss: Vol. 1. Attach- ington, DC.Erickson, M., Egeland, B., & Sroufe, L. A. (1985). Thement (2nd ed.). New York: Basic Books.

Bretherton, I., Ridgeway, D., & Cassidy, J. (1990). As- relationship between quality of attachment and behav-ior problems in preschool in a high risk sample. Insessing internal working models of the attachment re-

lationship: An attachment story completion task for I. Bretherton & E. Waters (Eds.), Growing points inattachment theory and research. Monographs of the3-year-olds. In M. T. Greenberg, D. Cicchetti, & E.

M. Cummings (Eds.), Attachment in the preschool Society for Research in Child Development, 50(1–2,Serial No. 209), 147–186.years: Theory, research, and intervention (pp. 273–

308). Chicago: University of Chicago Press. Fox, N. (1997). Attachment in infants and adults: A linkbetwen the two?. Paper presented at the biennialBronfenbrenner, U. (1986). Ecology of the family as a

context for human development: Research perspec- meeting of the Society for Research in Child Devel-opment, Washington, DC.tive. Developmental Psychology, 22, 723–742.

Brown, G., Harris, T., & Bifulco, A. (1986). Long-term Fury, G., Carlson, E., & Sroufe, L. A. (1997). Children’srepresentations of attachment relationships in familyeffects of early loss of parent. In M. Rutter, C.

Izard, & P. Read (Eds.), Depression in young people drawings. Child Development, 68, 1154–1164.George, C., Kaplan, N., & Main, M. (1984). Attachment(pp. 251–296). New York: Guilford Press.

Carlson, E. A. (1998). A prospective longitudinal study interview for adults. Unpublished manuscript, Uni-versity of California, Berkeley.of attachment disorganization/disorientation. Child

Development, 69, 1107–1128. Kobak, R. R., Ruckdeschel, K., & Hazan, C. (1994).From symptom to signal: An attachment view ofCarlson, E. A., & Sroufe, L. A. (1995). The contribution

of attachment theory to developmental psychopathol- emotion in marital therapy. In S. M. Johson & L. S.Greenberg (Eds.), The heart of the matter (pp. 46–ogy. In D. Cicchetti & D. Cohen (Eds.), Develop-

mental processes and psychopathology: Vol. 1. Theo- 71). New York: Brunner/Mazel.Kobak, R. R., & Shaver, P. (1987, June). Strategies forretical perspectives and methodological approaches

(pp. 581–617). New York: Cambridge University maintaining felt security: A theoretical analysis ofcontinuity and change in styles of social adaptation.Press.

Cicchetti, D., & Tucker, D. (1994). Development and self Paper presented at the Conference in honor of JohnBowlby’s 80th Birthday, London, England.regulatory structures of the mind. Development and

Psychopathology, 6(4), 533–549. Lamb, M. (1984). Fathers, mothers and childcare in the1980s: Family influences on child development. In K.Crowell J. (1997). Current relationship representations:

Relations to secure base behavior in marital interac- Borman, D. Quarm, & S. Gideouse (Eds.), Woman inthe workplace (pp. 61–88). Norwood, NJ: Ablex.tions. Paper presented at the biennial meeting of the


Lewis, M. (1997) Altering fate. New York: Guilford. Rosenberg, D. (1984). The quality and content of pre-school fantasy play: Correlates in concurrent social-Loevinger, J. (1976). Ego development. San Francisco:

Jossey–Bass. personality function and early mother–child attach-ment relationships. Unpublished doctoral dissertation,Ludwig, A. M. (1983). The psychobiological functions of

dissociation. American Journal of Clinical Hypnosis, University of Minnesota, Minneapolis.Sameroff, A. (1989). General systems and the regulation26, 93–99.

Main, M. (1981). Avoidance in the service of attachment: of development. In M. Gunnar & E. Thelen (Eds.),Minnesota Symposia in Child Psychology (Vol. 22,A working paper. In K. Immelmann, G. Barlow,

L. Petrinovich, & M. Main (Eds.), Behavioral devel- pp. 219–235). Hillsdale, NJ: Erlbaum.Sameroff, A., & Chandler, M. (1975). Reproductive riskopment: The bielefeld interdisciplinary project

(pp. 651–693). New York: Cambridge University and the continuum of caretaking casualty. In F. D.Horowitz (Ed.), Child development research (Vol. 4).Press.

Main, M. (1993). Discourse, prediction, and recent stud- Chicago: University of Chicago Press.Sameroff, A., & Emde, R. (1989). Relationship disturb-ies in attachment: Implications for psychoanalysis.

Journal of the American Psychoanalytic Association, ances in early childhood. New York: Basic Books.Schore, A. (1994). Affect regulation and the origin of the41, 245–260.

Main, M., & Goldwyn, R. (in press). Adult attachment self: The neurobiology of emotional development.Hillsdale, NJ: Erlbaum.scoring and classification system. In M. Main (Ed.),

Behavior and the development of representational Sroufe, L. A. (1978, October). Attachment and the rootsof competence. Human nature, 1, 50–57.models of attachment: Five methods of assessment.

New York: Cambridge University Press. Sroufe, L. A. (1983). Infant–caregiver attachment andpatterns of adaptation in preschool: The roots of mal-Main, M., & Hesse, E. (1990). Parents’ unresolved trau-

matic experiences are related to infant disorganized adaptation and competence. In M. Perlmutter (Ed.),Minnesota Symposia in Child Psychology (Vol. 16,attachment status: Is frightened and/or frightening pa-

rental behavior the linking mechanism? In M. T. pp. 41–83). Hillsdale, NJ: Erlbaum.Sroufe, L. A. (1986). Bowlby’s contribution to psychoan-Greenberg, D. Cicchetti, & E. M. Cummings (Eds.),

Attachment in the preschool years (pp. 161–182). alytic theory and developmental psychology. Journalof Child Psychology and Psychiatry, 27, 841–849.Chicago: University of Chicago Press.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in Sroufe, L. A. (1988). The role of infant–caregiver attach-ment in development. In J. Belsky & T. Nezworskiinfancy, childhood, and adulthood: A move to the

level of representation. In I. Bretherton & E. Waters (Eds.), Clinical implications of attachment (pp. 18–38). Hillsdale, NJ: Erlbaum.(Eds.), Growing points of attachment theory and re-

search. Monographs of the Society for Research Sroufe, L. A. (1996). Emotional development: The organ-ization of emotional life in the early years. New York:in Child Development, 50(1–2, Serial No. 209), 66–

104. Cambridge University Press.Sroufe, L. A. (1997). Psychopathology as outcome of de-Ogawa, J., Sroufe, L. A., Weinfield, N. S., Carlson, E., &

Egeland, B. (1997). Development and the fragmented velopment. Development and Psychopathology, 9,251–268.self: A longitudinal sudy of dissociative symptom-

atology in a non-clinical sample. Development and Sroufe, L. A., Carlson, E., & Shulman, S. (1993). Thedevelopment of individuals in relationships: From in-Psychopathology, 9, 855–1164.

Ostoja, E., McCrone, E., Lehn, L., Reed, T., & Sroufe, fancy through adolescence. In D. C. Funder, R. Parke,C. Tomlinson–Keesey, & K. Widaman (Eds.), Study-L. A. (1995, March). Representations of close rela-

tionships in adolescence: Longitudinal antecedents ing lives through time: Approaches to personality anddevelopment (pp. 315–342). Washington, DC: Ameri-from infancy through childhood. Poster presented at

the biennial meeting of the Society for Research in can Psychological Association.Sroufe, L. A., & Egeland, B. (1991). Illustrations of per-Child Development, Indianapolis, IN.

Pearson, J. L., Cohn, D. A., Cowan, P. A., & Cowan, son and environment interaction from a longitudinalstudy. In T. Wachs & R. Plomin (Eds.), Conceptual-C. P. (1994). Earned- and continuous-security in adult

attachment: Relation to depressive symptomatology ization and measurement of organism–environmentinteraction (pp. 68–84). Washington, DC: Americanand parenting style. Development and Psychopathol-

ogy, 6, 359–373. Psychological Association.Sroufe, L. A., Egeland, B., & Carlson, E. A. (in press).Phelps, J. L., Belsky, J., & Crnic, K. (1998). Earned secu-

rity, daily stress, and parenting: A comparison of five One social world: The integrated development of par-ent–child and peer relationships. In W. A. Collins &alternative models. Development and Psychopathol-

ogy, 10, 21–38. B. Laursen (Eds.), Relationships as developmentalcontext: The 29th Minnesota symposium on child psy-Pianta, R., Egeland, B., & Sroufe, L. A. (1990). Maternal

stress in children’s development: Predictions of chology. Hillsdale, NJ: Erlbaum.Sroufe, L. A., Egeland, B., & Kreutzer, T. (1990). Theschool outcomes and identification of protective fac-

tors. In J. E. Rolf, A. Masten, D. Cicchetti, K. Neuch- fate of early experience following developmentalchange: Longitudinal approaches to individual adap-terlen, & S. Weintraub (Eds.), Risk and protective fac-

tors in the development of psychopathology (pp. tation in childhood. Child Development, 61, 1363–1373.215–235). New York: Cambridge University Press.

Putnam, F. W. (1989). Diagnosis and treatment of multi- Sroufe, L. A., & Fleeson, J. (1986). Attachment and theconstruction of relationships. In W. Hartup & Z. Ru-ple personality disorder. New York: Guilford.

Rieder, C., & Cicchetti, D. (1989). Organizational per- bin (Eds.), Relationships and development. Hillsdale,NJ: Erlbaum.spective on cognitive functioning and cognitive-affec-

tive balance in maltreated children. Developmental Sroufe, L. A., & Fleeson, J. (1988). The coherence offamily relationships. In R. A. Hinde & J. Stevenson–Psychology, 25(3), 382–393.


Hinde (Eds.), Relationships within families: Mutual tations, parental responsiveness, and infant attach-ment: A meta-analysis on the predictive validty of theinfluences. Oxford: Oxford University Press, 27–47.

Sroufe, L. A., Levy, A. K., & Carlson, E. (1998, March). adult attachment interview. Psychological Bulletin,117(3), 387–403.Resilience in adolescence: Changing the odds. Paper

presented at the biennial meeting of the Society for Vaughn, B., Waters, E., Egeland, B., & Sroufe, L. A.(1979). Individual differences in infant–mother at-Research on Adolescence, San Diego, CA.

Sroufe, L. A., & Waters, E. (1977). Attachment as an tachment at 12 and 18 months: Stability and changein families under stress. Child Development, 50(4),organizational construct. Child Development, 48,

1184–1199. 971–975.Weinfield, N. S., Sroufe, L. A., Egeland, B., & Carlson,Suess, G. J., Grossmann, K. E., & Sroufe, L. A. (1992).

Effects of infant attachment to mother and father on E. (in press). The nature of individual differences ininfant–caregiver attachment. In P. Shaver & J. Cas-quality of adaptation in preschool: From dyadic to in-

dividual organization of self. International Journal of sidy (Eds.), Handbook of attachment: Theory, re-search, and clinical implications. New York: Guil-Behavioral Development, 15(1), 43–66.

van Ijzendoorn, M. H. (1995). Adult attachment represen- ford.

sroufe

Documents