Reprinted from: Handbook of Developmental Psychopathology (2nd Ed.. Arnold J. Sameroff, Michael Lewis, and
Suzanne M. Miller (Eds). Kluwer Academic/ Plenum Publishers, New York, 2000.

5 Relationships, Development,
and Psychopathology

L. Alan Sroufe, Sunita Duggal, Nancy Weinfield, and Elizabeth Carlson

Interpersonal relationships are pivotal for studying psychopathology in general and devel-
opmental psychopathology in particular. This is so at multiple levels of analysis, from defining
psychopathology, to describing preconditions and contexts, and to understanding its origins
and nature.
For example, relationship problems often are markers of disturbance, and the diagnosis of
disorder often centers on relationship considerations. From social phobias to conduct problems
to psychotic disorders, across the whole range of problems in childhood and adulthood, distur-
bances in interpersonal relationships are prominent criteria for classification in psychopa-
thology. Thus, when there is psychological disturbance, interpersonal relationships also are
likely to be disturbed. Given the critical importance of relationships in human adaptation, this
is not surprising. This role of relationship problems as markers of pathology would, in and of
itself, be sufficient grounds for emphasizing the developmental study of relationships for the
field of psychopathology. But this is only the beginning.
Social relationships also are viewed by many theorists as important contexts within which
psychopathology emerges and persists or desists. Psychogenic positions on pathology all focus
on relationships, whether this be social learning experiences, the isolation and anomie empha-
sized by sociological models, or the emphasis on vital close relationships in psychodynamic
and evolutionary positions (Lazare,1973). Research on risk and protective factors in Psychopa-
thology, as well as process-oriented research involving moderator and mediator variables,
commonly grants a prominent role for relationship variables. For some problems, such as con-
duct disorders (Dodge, Chapter 24, this volume), relationship experiences clearly play a domi-
nant role. But all disorders develop in context (e.g., Lewis, 1984; Sameroff,1997; Sroufe,
1997), and relationships with caregivers, peers, and others are a critical part of the child's de-
velopmental context.
Finally, a more thoroughgoing point of view has been proposed by some theorists (see, e.g.,
Bowlby,1973; Sameroff & Emde,1989). In this perspective, vital early relationships are seen
as the progenitors of disorders; psychopathology is the outgrowth of relationship disturbances.
Relationship disturbances themselves may constitute the roots of pathological processes that
only later are manifest in individual behavior in broader contexts. A pathway to pathology is
initiated and maintained by critical relationships in which the child participates. This viewpoint
has some kinship with family systems perspectives, in which disorder is seen in the relation-
ship system and not the individuals (e.g., Jackson, 1977). But in this relationship perspective,
the reality of individual disorder is granted. However, the prototype for this disorder may lie in
the patterns of relationships previously experienced.
In summary, relationship issues are not only important for defining pathology but also for
understanding the origins and course of disorder. From a wide array of theoretical vantage
points, social relationships have a key role in the etiology, maintenance, and remediation of
disturbed behavior. In the following sections, we discuss relationships in terms of markers of
disorder and as risk factors, protective factors, and contexts with regard to pathology. We end
with a discussion of relationship disturbances as initiating pathways to psychopathology.

---

Sroufe et al. Attachment and Psychopathology
RELATIONSHIP PROBLEMS AS CRITERIA FOR DISORDER
Interpersonal relationships may be defined as patterns of interaction with specific partners,
such as parents or peers, that are carried out over time and entail some degree of investment by
participants (Hinde, 1979). Our definition of relationship problems is more inclusive, including
failures to form relationships, incompetent social behavior, social withdrawal, social anxiety;
and behavior that is noxious to others.
Even causal perusal of the current psychiatric classification system for disorders (American
Psychiatric Association, 1994) reveals the centrality of interpersonal relationship problems in
major disorders. While social relationship criteria commonly are more extensive and more
clearly delineated for disorders first diagnosed in childhood, they are also quite prevalent in
major adult disorders. Moreover, for all major child disorders and many adult disorders
(including, for example, Major Depressive Disorder and Bipolar Disorder), one criterion for
diagnosis is "significant impairment" in social functioning.
Many major childhood and adult disorders have relationship disturbance criteria (see Table
5.1). The very first criterion for Autistic Disorder, for example, is "qualitative impairment in
social interaction." Failure to develop peer relationships, lack of emotional sharing with others,
lack of social or emotional reciprocity, and communication deficits are specifically cited.

The Attention Deficit, Disruptive Behavior Disorders all have social features. While per-
haps not obvious criteria of Attention Deficit/Hyperactivity Disorder, relationship features are
nonetheless germane. As with many childhood problems, it is the impact of the child's behav-
ior on others that leads to referral and diagnosis. Specific symptoms include "interrupting,"
DSM-IV Diagnostic Criteria with Implications for Relationships

DSM-IV disorder Examples of relevant DSM-IV diagnostic criteria

Autistic Disorder Qualitative impairment in social interaction.
Delays or abnormal functioning in (1) social interaction, (2) language as used in
social communication, or (3) symbolic or imaginative play.
Attention-Deficit/ Often does not seem to listen when spoken to directly. Often interrupts or intrudes
Hyperactivity Disorder on others (e.g., butts into conversations or games)
Conduct Disorder Often bullies, threatens, or intimidates others; often initiates physical fights; has
been physically cruel to people.
Oppositional Defiant Disorder A pattern of negativistic, hostile, and defiant behavior lasting at least 6 months.
Separation Anxiety Disorder Developmentally inappropriate and excessive anxiety concerning separation from
home or from those who whom the individual is attached.
Reactive Attachment Disorder Markedly disturbed and developmentally inappropriate social relatedness in most con-
texts.
of Infancy Early Childhood
Substance Abuse Continued substance use despite having persistent or recurrent social or inter
personal problems caused or exacerbated by the effects of the substance.
Schizophrenia Social/occupational dysfunction.
Social Phobia A marked and persistent fear of one or more social or performance situations in which
the person is exposed to unfamiliar poeple or to possible scrutiny by others.
The avoidance, anxious anticipation, or distress in the feared social or performance
situation(s) interferes significantly with the person's normal routine.
Posttraurnatic Stress Disorder Feelings of detachment or estrangement from others.
2

---

Sroufe et al. Attachment and Psychopathology
"intruding," or "not listening" to others. In the case of Conduct Disorders, the child's bullying,
threatening, cruel, or aggressive behavior toward others is often central. The severity specifica-
tions for this disorder explicitly refer to effects (especially amount of harm) caused to others.
Oppositional Defiant Disorder, of course, is defined by arguing with, annoying, defying, and
refusing to comply with parents, teachers, or other adults.
Separation Anxiety Disorder and Reactive Attachment Disorder were included in the DSM
system specifically to capture explicit forms of relationship problems. The former entails ex-
cessive distress in the face of separation from an attachment figure or excessive worry with
regard to possible or upcoming separations that may be manifest in a variety of ways. Reactive
Attachment Disorder is defined by inappropriate social relatedness manifest either in (1) fail-
ure to appropriately initiate or respond to social encounters or (2) indiscriminate sociability or
diffuse attachment. It is noteworthy that presumed pathogenic care also is a defining criterion
for this disorder.
An array of adult disorders likewise have relationship problems as central features. From
Social Phobias and Generalized Anxiety Disorders to psychosis, impairments in social rela-
tionships are prevalent. For example, one increasingly prominent anxiety disorder, Posttrau-
matic Stress Disorder PTSD is characterized by feelings of detachment or estrangement from
others. The social withdrawal and inappropriate social behavior associated with many forms of
schizophrenia are well known. Relationship problems are especially prominent in the personal-
ity disorders. All personality disorders, from Schizoid to Multiple Personality Disorder, are
characterized by markedly deviating functioning in interpersonal relationships and/ or affectiv-
ity (dependency, antisocial behavior, etc.). Borderline Personality Disorder is characterized by
profound abandonment worries and extreme lability in relationships, in which partners are al-
ternately idealized and devalued. Those with Narcissistic Personality Disorder have superficial
relationships and demand to be idealized.
Even disorders that on the surface are defined outside of the interpersonal domain often
entail relationship criteria. Substance abuse, for example, requires for diagnosis continued use
of the substance despite persistent or recurrent "social or interpersonal problems" caused or
exacerbated by the effects of the substance (e.g., physical fights or arguments with spouse
about the consequences of substance use).
In summary, throughout the DSM system, relationship problems play a key role in both
determining that there is a problem warranting diagnosis and in determining the specific classi-
fication. This is testimony both to the centrality of social relationships in human functioning
and to the merit of research in developmental psychopathology focusing on relationship issues.
(A more complete tabular summary of relationship criteria for disorders is available from the
authors.)
RELATIONSHIPS AS CONTEXTS FOR PSYCHOPATHOLOGY
When child problems and relationship problems co-occur, it is often difficult to establish
causality. Clearly, child disturbance would have an impact on relationships with parents and
peers, as implied by the preceding discussion of relationship criteria. Moreover, there is docu-
mentation of such child effects in the literature; for example, changes in parental behavior fol-
lowing reduction in child symptomatology (Hinshaw & McHale,1991; Sroufe,1997). Many
models of child problems entail concepts of ongoing, mutual influence of parents and child (e.
g., the work of Patterson, discussed later; see also Dodge, Chapter 24, this volume). Still, a
persuasive case may be made for the role of relationships in the onset and course of psychopa-
thology. Relationship disturbances often precede the manifestation of individual pathology,
and relationship strengths predict differential resistance to adversity (e.g., Masten, 1994).
Moreover, relationship change has been shown to precede change in individual disturbance
and to influence the effect of other variables on psychopathology (e.g., Erickson, Sroufe, &
Egeland, 1985). All of this is reflected in the literature on risk factors, protective factors, mod-
erators, and mediators. Cause is complex in psychopathology. Rarely can one say that a certain
3

---

Sroufe et al. Attachment and Psychopathology
pattern of parenting (or a certain relationship experience) directly led to a pathological out-
come in a linear manner, yet it is certain that relationship experiences often are a crucial con-
text for the emergence, waxing, and waning of pathology.
Relationships as Risk Factors for Disorder
Risk is a population concept. To say that an individual is "at risk" for pathology is to indi-
cate that he or she is a member of a group that has an increased likelihood of later manifesting
the disorder in question. A causal role is not necessarily implied, but risk factors are often seen
as part of a causal network. Within this framework, both aspects of children's relationships
with others and the broader relationship context in which they are developing have been identi-
fied as risk factors for psychopathology. From examining certain relationship variables it is
possible to increment predictions of later pathology, sometimes dramatically.
Parent-Child Relationships as Risk Factors.
Dimensions of Parenting. More than three decades of research have established two basic di-
mensions of parenting as risk factors for psychopathology: (1) harsh treatment (hostility, criti-
cality, rejection); and (2) lack of clear, firm discipline or supervision (e.g., Farrington et
al.,1990; Maccoby & Martin, 1983; Patterson, Debaryshe, & Ramsey, 1989). These factors
together, and in interaction with other variables, are often especially predictive and at times
capable of differentiating various pathological outcomes.
Countless studies have underscored the predictive power of harsh treatment or rejection,
with findings especially consistent for externalizing problems in boys (e.g., Campbell, 1997;
Earls, 1994; Eron & Huesmann, 1990; Farrington et al., 1990; Harrington, 1994; Jenkins &
Smith, 1990; see also Dodge, Chapter 24, this volume; Fiese, Wilder, & Bickham, Chapter 7,
this volume). Rejection, lack of support, and hostility also have been consistently related to
depression (e.g., Asarnow, Tompson, Hamilton, Goldstein, & Guthrie, 1994). Many of these
studies are prospective, for example, predicting conduct problems throughout childhood and
even into adulthood. Feldman and Weinberger (1994) found that parental rejection and power
assertive discipline predicted delinquent behavior of sixth-grade boys 4 years later. Ge, Best,
Conger, and Simons (1996) found that parental hostility predicted 10th graders' behavior prob-
lems, even after controlling for 7th-grade symptom levels, and distinguished between those
with conduct disorders and those with depression. Using a behavior genetic design, Reiss et al.
(1995) found that the specific level of parental negativity directed to one member of a sibling
pair predicted that child's level of conduct problems, thus showing this effect above and be-
yond any genetic contribution. Likewise, Patterson and Dishion (1988) reported that aggres-
sive treatment of children was more predictive of conduct problems than parent trait measures
of aggressiveness (a genetic surrogate). In our own research, we have found that low parental
warmth predicted childhood depression, even after controlling for maternal depression (Duggal
et al., in press).
Many of the studies cited here also demonstrated the impact of inconsistent discipline. One
of the most powerful variables to be delineated in the last 15 years is the degree of parental
"monitoring" (supervision and oversight; e.g., Dishion, Patterson, Stoolmiller, & Skinner,
1991). While some report only concurrent correlations, numerous prospective, longitudinal
studies confirm the relation of lax discipline to later pathology, especially conduct disorders (e.
g., Feldman & Weinberger, 1994; Ge et al., 1996; see also Fiese et al., Chapter 7, this volume)
and association with deviant peers (e.g., Dishion et al.,1991). We discuss the role of monitor-
ing as a moderator/mediator variable in the next major section.
A variable somewhat related to caregiver inconsistency has emerged from our own re-
search: parent-child "boundary violation." This refers to an abdication by'the adult of the pa-
rental role, especially when firm guidance is needed, and treating the child in a peer-like or
spousal-like way (role reversal). Assessment of this variable at age 42 months was found to be
a consistent predictor of attention deficit/hyperactivity symptoms in elementary school, and to
predict above and beyond measures of temperament, perinatal difficulties, or other early child
4

---

Sroufe et al. Attachment and Psychopathology
measures (Carlson, Jacobvitz, & Sroufe, 1995). Likewise, a comparable measure at age 13
years predicted subsequent conduct problems in boys (Nelson, 1994) and dating and sexuality
problems in girls (Hennighausen, Collins, Anderson, & Hyson, 1998). Early pregnancy was
predicted by the 42-month measure, and early impregnation (the comparable measure for boys)
was predicted by the 13-year variable (Levy, 1998). A more general measure of parental
boundary difficulties ("intrusiveness") obtained in infancy has been found to predict behavior
problems throughout childhood and adolescence, being strikingly more powerful than infant
temperament variables (Carlson et al., 1995; Egeland, Pianta, & Ogawa, 1996).
Child Maltreatment. The substantial literature on child maltreatment (e.g., Cicchetti, Toth,
& Maughan, Chapter 37, this volume) confirms the role for parental hostility and harshness
outlined earlier. As Toth, Manly, and Cicchetti (1992) have suggested, maltreatment reflects
"an extreme on the continuum of caretaking casualty" (p. 98). Prospective studies show that
maltreatment (including physical abuse and emotional unavailability) is associated with con-
duct problems, disruptive behavior disorders, attention problems, anxiety disorders (including
PTSD and mood disorders (Cicchetti et al., Chapter 37, this volume; Cicchetti & Lynch,
1995). Egeland (1997) found that 9096 of children with an observed history of childhood mal-
treatment showed at least one diagnosable disorder at age 17'% years, compared to 3096 of the
poverty control subjects who were not maltreated.
Sexual abuse, the extreme of boundary violation, appears to be especially pathogenic, being
related to a variety of problems (Kendall-Tackett, William., & Finkelhor,1993; Toth & Cic-
chetti, 1996). Even in comparison to other maltreatment groups, those who are sexually abused
manifest more forms of pathology and more extreme pathology (Egeland,1997; Toth & Cic-
chetti, 1996). Sexual abuse is strongly and specifically associated with PTSD (Putnam, Chap-
ter 39, this volume) and with depression. In our research, it accounted for depression in both
childhood and adolescence, even after taking into account maternal depression and other po-
tentially confounding factors (Duggal et al., in press).
Interpersonal Conflict. Divorce, parental disharmony, and family violence all have been
consistently associated with child behavioral and emotional problems (e.g., Amato & Keith,
1991; Emery & Kitzmann,1995; Fiese et al., Chapter 7, this volume). Such conditions are
overlapping and also co-occur with misnt or neglect of children, making causal conclusions
difficult. Numerous studies have shown children of divorce to have more problems than those
in intact families (see Amato & Keith, 1991, for a mete-analysis). Researchers believe this is
largely due to the 'conflict preceding and surrounding the marital breakup (e.g., Wallerstein &
Kelly, 1982). It is the case that behavior problems often precede the divorce (Cherlin et
al.,1991), and that parental conflict is consistently found to be a stronger predictor of child
maladjustment than marital status (Emery & Kitzmann, 1995). Across eight studies reviewed,
Amato and Keith (1991) found that children from high-conflict, intact families showed more
problems (including depression and anxiety) than children from divorced families in general.
They also reported more problems for children of divorce (where there was often conflict) than
for those who lost a parent through death. Still, even if research to date shows little impact of
divorce above and beyond the role of conflict, it remains an important marker variable and is a
risk factor in the descriptive, population sense defined earlier.
Family violence has also been found to be associated with child pathology (e.g., Sternberg
et al., 1993). Here, a major problem is distinguishing the impact upon the child of witnessing
violence from the consequences of direct maltreatment, which often co-occurs, or from the
general life stress and chaos in which family violence is nested However, in a recent analysis
of prospective, longitudinal data, Dodds (1995) was, able to control for these potential con-
founds. Presence of spousal abuse in early childhood predicted externalizing behavior prob-
lems in boys (but not girls, a common result), even with child maltreatment, socioeconomic
status (SES), and life stress statistically controlled.
Peer Relationships as Risk Factors
5

---

Sroufe et al. Attachment and Psychopathology
One reason for the power of family factors in predicting later pathology may be their im-
pact upon peer relationships. Maltreatment, for example, is consistently associated with lack of
competence with, and rejection by, peers (e.g., Cicchetti et al., Chapter 37, this volume; Cic-
chetti & Lynch, 1995), as have patterns of anxious attachment, especially the avoidant subtype.
(e.g., Sroufe, Egeland, & Carlson, 1999). Ample research shows that poor peer relationships
and association with deviant peers themselves are risk factors for psychopathology (Rudolph
& Asher, Chapter 9, this volume). Given the strong concurrent association between behavior
problems and peer problems, our own review refers only to prospective studies in which peer
measures precede later measures of psychopathology.
Numerous studies have found that general problems with peers, lack of social competence,
or unpopularity (based on observation, teacher ratings, or peer sociometrics) are related to later
behavioral and emotional problems (e.g., Masten 8t Coatsworth,1995). For example, in one
early study, a single item rated by teachers ("Fails to get along with other children") predicted
psychiatric problems, including hospitalizations, 12 years later in adulthood (Janes, Hessel-
brock, Myers, & Penniman, 1979). In our own research, we have found that teacher rankings
of peer competence, beginning in early elementary school, predict behavior problems and psy-
chopathology throughout childhood and adolescence (Sroufe et al., 1999).
Established patterns of sociometric status (e.g., rejected vs. neglected children) have proven
to be very useful (see Rudolph & Asher, Chapter 9, this volume), predicting somewhat differ-
ent problems later. Peer rejection is especially powerful, even in comparison to peer neglect (e.
g., Ollendick, Weist, Borden, & Greene, 1992). Numerous studies have documented a relation
between a history of peer rejection and later maladjustment, both externalizing and internaliz-
ing problems, sometimes even with earlier behavior problems controlled (e.g., Rudolph &
Asher, Chapter 9, this volume; Burks, Dodge, & Price, 1995; Coie, Terry, Lenox, Lochman, &
Hyman, 1995; Dodge, Chapter 24, this volume; Ollendick et al., 1992).
Finally, a great deal of recent research has emphasized the negative impact of deviant peer
group membership (e.g., Cairns, Cairns, & Neckerman, 1989; Keenan, Loeber, Zhang,
Stouthamer-Loeber, & Van Kammen,1995; Patterson et al.,1989). Such a relationship experi-
ence is especially implicated in delinquency and school dropout.
Relationships as Protective Factors, Moderators, and Mediators
Technically, a protective factor, when present, moderates the impact of a risk variable; that
is, protection is always particular to specific risks (Rutter, 1990). Thus, when factors are gener-
ally associated with positive outcomes (often simply being the other end of a risk dimension),
they are best described as assets or, to use a term coined by Sameroff (1997), "promotive" fac-
tors. Of course, this distinction is not always easy to make, and the same variable may be
viewed as an asset or protective factor depending on the context. This is certainly the case for
certain relationship variables.
Relationship experiences may also moderate the impact of other risks or alter their impact
(e.g., combine to lead to a distinctive outcome). In other circumstances, relationships may be
the mechanism through which a certain risk factor has its impact. This is referred to as a me-
diator variable. While there is less information on relationships as moderators and mediators,
compared to risks and assets, such research is of clear importance. within a developmental per-
spective.
Family Relationships as Assets, Moderators, and Mediators
The most widely studied assets and protective factors in the parent-child relationship are
parental warmth and emotional support, and the security of the attachment between infant and
caregiver. Numerous studies have documented the link between parental warmth and psycho-
logical well-being and emotional health of the child (e.g., Campbell, 1997; Fiese et al., Chapter
2; Hetherington & Clingempeel,1992; Sroufe,1997). Infant attachment security has been linked
with later self-esteem, social competence, prosocial behavior, ego resiliency, and overall ad-
6

---

Sroufe et al. Attachment and Psychopathology
justment (Sroufe, 1997; see also below).
Attachment security also is associated with recovery from behavioral problems (Sroufe,
Egeland, & Kreutzer,1990) and is a protective factor with regard to family life stress; that is,
children with histories of secure attachment show fewer problems in the face of family stress
than do children with histories of anxious attachment (Pianta, Egeland, & Sroufe, 1990).
Much of the research on family risk factors also contains evidence of other family relation-
ship factors as moderators and mediators of such risk. Davies and Cummings (1994) propose
that secure relationships with parents moderate the impact of marital conflict. Indeed, Miller,
Cowan, Cowan, Hetherington, and Clingempeel (1993) found just that to be true for preschool-
ers and early adolescents. Other research suggests that the course or trajectory of problem be-
havior may be altered by parent-child relationship qualities. Campbell (1997), for example,
reports that "authoritative parenting" accounts for desistance of behavior problems between
preschool and elementary school. Finally, the research of Patterson and colleagues contains
exemplary process analyses. They find, for example, that the impact of parental conflict and
aggressiveness is mediated by (leads to) lax parental monitoring, which is then the more pow-
erful influence on adolescent conduct problems (Capaldi & Patterson, 1991). They also de-
scribe the transactive nature of the developmental process between parents and peers (see be-
low).
Peer Relationships as Assets, Moderators, and Mediators
As with family relationships, peer relationships may represent assets as well as risks. Peer
competence is associated with low behavior problem scores or absence of pathology just as
much as peer problems are associated with disorder. Moreover, peer competence measures
have been associated with academic achievement and school completion, which themselves
may be viewed as assets (Teo, Carlson, Mathieu, Egeland, & Sroufe,1996), although reversed
statements would be equally true and we could speak of risk.
Patterson and associates' work specifically points to a mediating role for peer experiences
in the perpetuation of conduct problems. In their model (e.g., Dishion at al.,1991; Patterson et
al.,1989), poor parental discipline and monitoring lead to conduct problems, which in turn are
associated with peer rejection and academic failure. These factors converge to promote com-
mitment to a deviant peer group, leading to consolidation of antisocial behavior. Problem be-
haviors (and peer competence) are best viewed as drawing upon the convergence of previous
family and peer experiences.
Other Relationships and General Social Support
While not so widely studied, relationships with grandparents, other adults, and siblings
have been suggested to serve protective or moderating roles in the face of stress or other risk
factors (Lewis, 1984). For example, Jenkins and Smith (1990) reported that a close relation-
ship with an adult outside of the family (usually a grandmother) moderated the effect of dishar-
monious marriages on child psychopathology. In our work, we found that an "alternative"
close relationship with an adult (again, often a grandmother, and sometimes a therapist) pre-
dicted breaking the cycle of abuse; those parents who themselves were abused but did not mis-
treat their own children much more often had such a factor present (Egeland, Jacobvitz, &
Sroufe, 1988).
Sibling relationships have been the subject of considerable interest to those studying peer
relationships, parental conflict, and psychopathology. It is clear that there is an association be-
tween quality of sibling relationships and adjustment or behavior problems (e.g., Dunn, Slom-
kowski, Beardsall, 8t Rende, 1994; Stormshak, Bellanti, & Bierman, 1996). However, these
findings may be interpreted in terms of troubled sibling relationships simply marking child dis-
turbance. Patterson (1986) has argued that sibling conflict may play a role in "training for
fighting" within coercive families. And there are some hints that siblings may play a protective
role. East and Rook (1992), for example, reported that peer-isolated children were less anxious
7

---

Sroufe et al. Attachment and Psychopathology
if they had a supportive sibling relationship, though they were still more anxious than average
children. Jenkins (1992) found that in disharmonious homes, children with a close sibling rela-
tionship had less symptomatology than children without such a relationship. But this is not a
widely reported finding; more often, the sibling relationship reflects the degree of parental
conflict (Hetherington, 1988).
Finally, there is substantial literature concerning the importance of social support for indi-
viduals at risk for or already experiencing problems, and for caregivers during the child's de-
velopment (e.g., Cohen & Wills, 1985; Nuechterlein et al.,1992; Robinson & Garber, 1995).
For example, Windle (1992) found that (lower) perceived social support from the family pre-
dicted both internalizing and externalizing behaviors for adolescent girls. Using an indirect
model of support, Goodyer, Herbert, Tamplin, Secher, and Pearson (1997) reported that moth-
ers' lack of confiding relationships with partners was related to the maintenance of disorder in
a clinical sample of 8- to 16-year-olds over a 36-week period.
RELATIONSHIP DISTURBANCES AND PATHWAYS TO DISORDER
In the preceding discussion, relationship experiences have been viewed as contributors to
psychopathology because of their role as risk factors, protective factors, mediators, or modera-
tors. However, a more thoroughgoing and revolutionary view of the role of relationships in
disturbance may be proposed. More than simply being risk factors, relationship disturbances
may be the precursors of individual psychopathology, through their role in establishing funda-
mental patterns of emotional regulation. They may represent the initiation of developmental
pathways probabilistically leading to disorder (Sroufe, 1997). Individual disturbance, in this
view, begins as relationship disturbance. This view is in contrast to the DSM framework, in
which the reality of relationship disturbances is allowed but sequestered into a few isolated
categories (Attachment Disorders). Here, relationship disturbances are hypothesized to be the
forerunners or prototypes of many major childhood disorders and adult personality disorders as
well.
The Relationship Perspective on Psychopathology: Rationale
Problems in emotional regulation, like relationship disturbances, are pervasive markers of
psychopathology. Such problems underlie most disorders of children and adults (Cole, Michel,
& O'Donnell-Teti, 1994). Indeed, "emotional disturbance" often is used as a synonym for psy-
chopathology. Moreover, difficulties in emotional regulation and relationship difficulties are
intertwined. This is the starting point for our developmental-relationship perspective on psy-
chopathology. Emotional regulation is the defining feature of all close relationships and the
central goal of early primary relationships (Sroufe, 1996). Particular relationship experiences
may be argued to be the progenitors of psychopathology precisely because of their role in early
regulation.
Human infants are not very able to regulate their own arousal or emotional states. To be
well regulated, they require ample assistance from caregivers. To be sure, they can express dis-
tress and contentment in the first weeks, and within a few months a greater range of feelings
and needs. By the end of the first year, infants can signal many wishes with intention (raising
their arms to be picked up, calling for caregivers when frightened, offering a toy for inspec-
tion). But throughout this time, they rely on caregivers to read these "signals," whether in-
tended or not. Infants are equipped to play only a primitive role in their own regulation. They
are not capable of self-regulation, but only "co-regulation" (Fogel, 1993). To be well regulat-
ed-to be competent as infants, they require sensitive, responsive caregivers (Ainsworth 8t Bell,
1974). In Sander's (1975) terms, there is an affective-behavioral organization early in life, but
this organization lies in the infant-caregiver system, not the infant alone.
Thus, what will become functional self-regulation, or various forms of dysregulation, be-
gins as caregiver-infant regulation (Lyons-Ruth & Zeanah, 1993). Researchers have now de-
scribed this initial dyadic regulation process in great detail, including its changing form over
time, as well as variations between particular infant-caregiver pairs (e.g., Brazelton,
8

---

Sroufe et al. Attachment and Psychopathology
Koslowski, & Main, 1974; Fogel, 1993; Stern, 1985). Caregivers maintain smooth regulation
by attending to the infant's changes in alertness or discomfort and signs of need, imbuing
primitive infant behaviors with meaning. They quickly learn to "read" infant signals and to
provide care that keeps distress and arousal within reasonable limits. By effectively engaging
the infant and encouraging ever longer bouts of emotionally charged but organized behavior,
they provide the infant with critical training in regulation. Within the secure "holding" frame-
work of the relationship, infants learn something vital about "holding" themselves, containing
behavior, and focusing attention (Brazelton et al., 1974).
In time, routine patterns of interchange are established. As the infant's capacities for en-
gagement and repertoire of behaviors increase, a semblance of reciprocity-of back-and-forth
communication-emerges. Caregiver and infant may, for example, engage in a series of mutual
exchanges characterized by increasingly positive emotion expressed by both partners and a
waxing and waning of engagement that helps the infant stay organized. In the early months, it
is the caregiver that is adjusting behavior purposefully, always accommodating to the infant
and creating space for the infant to fit in as well (Hayes, 1984). Such patterns of caregiver-or-
chestrated regulation set the stage for more truly dyadic regulation as new infant capacities
emerge.
By the second half-year, the infant exhibits purposeful, goal-directed behavior (Sroufe,
1996). Infants at this age behave in order to elicit a particular response from the caregiver, for
example, calling to the parent and raising their arms to indicate a desire to be picked up. They
now actively participate in the regulation process. If the caregiver misreads a signal, the older
infant will adjust the behavior, often until the desired response is received (e.g., crawling to
parents if they do not come to the infant). Thus, dyadic regulation follows inevitably upon the
heels of caregiver-orchestrated regulation. It requires only growth of intentional capacity,
which occurs in all normal infants during this period. The form and structure of dyadic regula-
tion is in place from the preceding period. What changes is the role of the infant, from reflex-
ive or automatic signaling, to active, intentional signaling; the patterning is based on what was
established earlier. In time, this patterning is carried forward, becoming the core of self-regula-
tion.
Early relationship experiences are vital because they are the first models or prototypes for
patterns of self-regulation. Infants have no choice but to generalize from what they experience.
If they have experienced within their caregiving relationships that distress is routinely fol-
lowed by recovery, that behavior can stay organized in the face of strong emotion, that positive
k experiences are shared, and that caregivers are central to all of these experiences, they will
come to expect such contingencies (Lewis & Goldberg, 1969). One can turn to others when in
need, and they will respond. At the same time, in a complementary manner, infants will come
to believe in their own effectiveness in maintaining regulation and, because their needs are
routinely met, in their own self-worth. Bowlby (1973) argues that this is inevitable. A sense of
personal effectiveness follows automatically from routinely having one's actions achie