Developmental in Psychiatrically
Drew
This
study
explored
History and Object Relations Disturbed Adolescent Girls
Westen, Ph.D., Pamela Ludoiph, Ph.D., Jean Wixom, Ph.D., and F. Charles
empirically
the
relationship
be-
tween developmental history variables and several dimensions of object relations in a sample of 36 female adolescent inpatients. The results document the importance of preoedipal experience, the relationship with the mother, and continuity of attachments in shaping object relations. In addition, the data point to the importance of distinguishing different dimensions of object relations, such as the affective quality of the object world and the logic and accuracy of attributions, which may have different developmental correlates. The findings also suggest the impact of sexual abuse, typically a postoedipal experience, on enduring objectrelational processes. (AmJ Psychiatry 1990; 147:1061-1068)
C
linicians and researchers (1) have speculated for years on the mole of developmental history factors and traumatic events in the genesis of severe psychopathology. Freud initially implicated sexual trauma in the etiology of hysteria but later came to emphasize the role of fantasy. Bolstered by observation of children with disrupted attachment histories (2) and by relevant primate evidence, object relations theorists came to focus again on the impact of meal deprivation in infancy and childhood. Winnicott (3) and Kohut (4) focused on failures in maternal responsiveness; Bowlby (5) implicated disruptions in the attachment relationship. Research on patients with severe personality disondens, particularly borderline patients, has begun to document the mole of experiences such as early loss, multiple and changing caretakers, violence, alcoholism, family chaos, and neglect in the genesis of these disorders (6-9). Several studies have established a link
Received Aug. 1 1, 1989; revision March 2, 1990. From the Department Michigan. Address reprint requests Psychology, University of Michigan,
receivedJan. 30, of Psychology, to Dr. Westen, 580 Union Dr.,
1990; accepted University of Department of Ann Arbor, MI
48 109-1346. Supported by grants from the Department of Psychiatry, the University Center for the Child and the Family, and the Faculty Fund, Department of Psychology, University of Michigan. Copyright © 1990 American Psychiatric Association.
Am
J
Psychiatry
147:8, August
1990
M. Judith Block, Wiss, M.A.
M.A.,
between borderline personality disorder and sexual abuse in adult and adolescent samples (8, 10-12). Presumably, chronic pathogenic experiences and traumatic events in childhood should have an enduring and systematic impact on later patterns of object nelations, although this has received little empirical attention. Studies to date on the influence of parental psychopathology and maltreatment on social cognition, which is theoretically of relevance to dimensions of object relations (13), have not, by and large, yielded positive findings (14). Research on adult attachment, particularly by Main and colleagues (15), has begun documenting the relationship between a mother’s attachment status (the security of hen attachment to her own mother) and the attachment status of hem child and have found that securely attached mothers are more likely to have securely attached children. This is one of the few areas of empirical research that has tested the relationship between developmental expenience and ways of experiencing relationships. The present study attempted to explore empirically the relationship between developmental history vaniables and object relations in a sample of psychiatrically disturbed adolescents. “Object relations” refers, most broadly, to enduring patterns of interpersonal behavion and to the cognitive and affective processes mediating functioning in close relationships. The present study focused on the internal (cognitive and affective) dimensions of object relations, rather than on their behavioral expression. We hypothesized, first, that the presence of developmental history variables such as maternal separations, neglect, physical and sexual abuse, and parental psychopathology should predict more pathological object relations in adolescence and second, that pneoedipal risk factors and disrupted attachments should be particularly predictive. We also predicted that different experiences should affect different dimensions of object relations. For example, one might suspect that sexual abuse would lead to a more malevolent object world, since caretakers would be likely to be experienced as hostile and unprotective, and a sense of basic trust (16) would be difficult to maintain. Grossly inconsistent parental behavior, as might occur with alcoholic parents, might be expected to lead to difficulties making logical and accurate attributions of the causes of people’s behavior. Because this is a pre-
1061
DEVELOPMENTAL
HISTORY
liminary study, we along these lines.
did
AND
not
OBJECT
make
RELATIONS
specific
hypotheses
METHOD Subjects were 36 female adolescents who were psychiatnically hospitalized at a major medical center and originally selected as part of a study comparing adolescents with borderline personality disorder to other psychiatrically disturbed adolescents on a number of dimensions (9, 1 1, 17). Patients with chronic psychosis, evidence of gross neuropathology, IQ below 70, or medical problems that would complicate diagnosis or psychological testing were excluded from the study. Subjects received a number of DSM-III and DSM-IIIR discharge diagnoses, including borderline personality disorder, mood disorder, anorexia nenvosa, bulimia, and other personality disorders. Subjects ranged in age from 14 to 18 years (mean±SD= 15.60±0.99). A number of developmental history variables were scored by chant review by a trained mater. The mater was blind to the psychological testing data from which object-relational variables were assessed. Patients were typically hospitalized for several months, so chart information was reasonably thorough. For each chart the rater read the intake summary, discharge summary, individual therapist’s process notes, case conference report, first 2 weeks of nursing notes, neuropsychological testing reports (if available), information from previous therapists and hospitalizations, school reports, and court reports (if available). Charts were mated for variables assessing the genetic family history; childhood symptoms of the patient; neurological history of the patient; and a variety of traumatic events, including documented neglect, physical and sexual abuse (fondling or penetration), extrusion from the home (being removed or “kicked out” of the residence of the primary caretaker), grossly inappropriate panental behavior (such as a parent double-dating with the child and being sexually provocative with the date), and significant separations, including death, divorce, adoption, prolonged separations from primary canetakers, and foster care history. The variables were coded as present only if there was conclusive evidence. It is therefore very likely that these data contain few false positive codes and relatively more false negatives. Although the data were largely factual and required little inference, a check was undertaken to ensure meliable coding. A second coder blindly rescored all vanables on a small sample of subjects; the two coders achieved perfect agreement on 89% of the scores. Object relations were assessed from Thematic Apperception Test (TAT) responses, using a procedure developed by Westen and colleagues (18). The TAT is a projective test in which subjects make up stories about characters depicted in ambiguous social scenes. It is a useful source of data for assessing object relations because, in describing characters and social episodes, subjects apply their understanding and expeni-
1062
ence TAT
of relationships and people. Administration of the to this sample is described more fully elsewhere (17). Testers asked subjects to tell a stony, including what was happening in the picture, what led up to it, the outcome, and what the characters were thinking and feeling. Testers were not familiar with the hypotheses or measures used in this study. The present study examined responses to six cards: cards 1, 2, 3, 4, 13MF, and 15. Responses to the six TAT cards were coded by using a multidimensional measure of object relations for use with the TAT. This measure assesses four aspects of object relations: complexity of representations of people (the extent to which the subject ascribes complex dispositions to characters whose perspectives are cleanly differentiated), affect-tone of relationship paradigms (the affective quality of the object world, from malevolent to benevolent), capacity for emotional investment in relationships and moral standards (the extent to which the person transcends a need-gratifying interpersonal orientation), and understanding of social causality (the extent to which attributions of causality in the social realm are accurate, complex, and psychologically minded). As depicted in table 1, each scale has five levels; level 1 represents the lowest level response and level 5, the highest. Several recent studies have attempted to validate these scales on a number of samples (19, 20). In nonmal and clinical samples the measures have predicted both self-reported and clinician-reported social adjustment, as well as scores on relevant subscales (such as hostility, paranoia, and interpersonal sensitivity) of the SCL-90-R (21). Research with a normal sample documented significant correlations between these measures and analogous measures developed for use with interview data, such as psychiatric interviews and psychothenapy transcripts (e.g., the extent to which subjects manifest a malevolent object world in describing TAT characters correlates with the malevolence of their descriptions of actual relationship episodes; see reference 20). This research also found predicted conrelations between these measures and theoretically related validated instruments such as Loevinger’s test of ego development (22) and Blatt et al.’s measure of qualitative aspects of parental representations (23). For example, Loevinger’s measure of ego development, which primarily assesses more affective dimensions such as the experience of relationships as exploitative, correlates with affect-tone of relationship paradigms and capacity for emotional investment in relationships. Complexity of free-response descriptions of significant others, through use of the Blatt measure, correlates with both complexity of representations and social causality as assessed from TAT responses. It has been demonstrated that mean scones (treated as continuous data) and percentage of level 1 responses (hypothesized to be pathological) on all four scales distinguish borderline subjects from psychiatric and normal companison subjects in both adolescent (17) and adult (13) samples. Two studies just completed (see reference 24)
Am
J
Psychiatry
147:8,
August
1990
WESTEN,
TABLE
1. Summary
of Four Aspects of Object
of Measures Complexity
Principle
1
People are not clearly entiated: confusion points of view
2
Simple, unidimensional representations; focus on actions; traits are global and univalent
Level
3
Minor life
Level
4
Level
S
overwhelming
pain
J
of mental
Mixed representations mildly negative
Expanded appreciation of complexity of subjective experience and personality dispositions; absence of representations integrating life history, complex subjectivity, and personality processes Complex representations indicating understanding of interaction of enduring and momentary psychological experience; understanding of personality as system of processes interacting with each other and the environment
Psychiatry
147:8,
AL.
with tone
Mixed representations neutral or balanced
with tone
Predominantly nesentations; enriching
repand
positive benign interactions
August
1990
Understanding Social Causality
of
Scale measures the extent to which others are treated as ends rather than means, events are regarded in terms other than need gratification, and moral standards are developed and considered.
Scale measures the extent to which attributions about the causes of people’s actions, thoughts, and feelings are logical, accurate, complex, and psychologicatty minded.
Need-gratifying
Noncausal or grossly illogical depictions of psychological and interpersonal
or
views social interaction as basically benign and enniching. Malevolent representations: gratuitous violence or gross negligence by significant others Representation of relationships as hostile, empty, or capricious but not profoundly malevolent; profound loneliness or disappointment in relationships
found predicted developmental differences between second and fifth graders and between early and late adolescents. Together, these studies suggest that TAT responses can provide an index of certain dimensions of object relations. TAT responses were coded by two advanced graduate students in clinical psychology and three research assistants with bachelor’s degrees. All cards were double-coded independently by two raters. The stories provided to coders were typed one to a page in random order, so that rating of multiple stories in the same protocol would be entirely independent. Reliability was computed by using the intraclass correlation co-
Am
Capacity for Emotional Investment
Scale measures affective quality of representations of people and relationships. It attempts to assess the extent to which the person expects from the world, and particularly the world of people, profound malevolence or
differof
Level
elaboration or personality
ET
Relations
Affect-Tone of Relationship Paradigms
Scale measures the extent to which the subject cleanly differentiates the penspectives of self and others: sees the self and others as having stable, enduring, multidimensional dispositions; and sees the self and others as psychological beings with complex motives and subjective experience.
Level
BLOCK,
of
Representations of People
Item
LUDOLPH,
profound tion
orientation;
self-preoccupa-
events Limited investment in peopIe, relationships, and moral standards; conflicting interests recognized, but gratification remains primary aim; moral standards primitive and unintegrated or followed to avoid punishment Conventional investment in people and moral standands; stereotypic compassion, mutuality, or helping orientation; guilt at moral transgressions Mature, committed investment in relationships and values: mutual empathy and concern; commitment to abstract values
Rudimentary understanding of social causality; minor logic errors or unexplained transitions; simple stimulus-response causality
Autonomous selfhood in the context of committed relationships; recognition of conventional nature of moral rules in the context of carefully considered standards or concern for concrete people or relationships
Complex appreciation of the role of mental processes in generating thoughts, feelings, behaviors, and interpersonal interactions; understanding of unconscious motivational pro-
Complex, accurate situational causality and rudimentary understanding of the role of thoughts and feelings in mediating action Expanded appreciation of the role of mental processes in generating thoughts, feelings, behaviors, and interpersonal interactions
cesses
efficient, with the Spearman-Brown connection for multiple coding. Unconnected reliabilities ranged from 0.71 to 0.96. Corrected reliabilities were as follows: affecttone, 0.91; complexity of representations, 0.96; capacity for emotional investment, 0.83; and social causality, 0.94.
RESULTS Table 2 reports significant findings on ship between object relations and parental The table presents mean scores on each
the relationpathology. of the four
1063
DEVELOPMENTAL
HISTORY
AND
OBJECT
TABLE 2. Parental Pathology, Traumatic cally Disturbed Adolescent Girls Affect-I
RELATIONS
Childhood
Experiences,
one of Relationship Paradigms
Complexity Representations Score
Score Item
N
Mean
Parental pathology Maternal psychiatric illness Yes No
21 74
2.33
0.50
2.65
0.69
t
alcohol
SD
Malevolent Responses (%)
164b
-
Maternal
and Scores on Four Scales Assessing
Mean
SD
Capac
of of People
Poorly Differentiated Responses (%)
1.81 2.15 1.96l
a -
-
-
-
_
-
-
-
Relations
ity for Emotional Investment
SD
0.50 0.51
Score Mean
38.80 20.10
Yes
-
-
-
(%)
18.80 2.40
167b
217b
-
-
-
-
-
-
-
-
-
-
8.10 0.10
-
-
-
-
-
-
214b
Yes
23
-
-
-
-
-
-
-
-
-
No
12
-
-
-
-
-
-
-
-
-
1.99 2.27
-
-
-
-
-
-
-
-
-
200b
-
-
-
-
-
-
-
-
8
t
-
-
-
-
-
-
25.80 8.60
psychiatric
t
-
0.44
17.20
0.28
2.80 180b
criminality
Yes No t
Childhood
SD
-
27
Paternal
Grossly lllogical Responses
abuse
No Paternal illness
for 36 Psychiatri-
Understanding of Social Causality
NeedGratifying Responses (%)
Score Mean
Object
7
-
-
-
-
-
28
-
-
-
-
-
9.30 0.90
-
-
-
-
-
-
2.42’
experiences
Neglect
Yes No
11 25
t
-
Grossly inappropriate parenting Yes No
9
t
Maternal
2.20 2.58 1.8?’
-
0.59
-
-
-
-
2.78
27
-
-
-
2.26
-
-
-
-
049
-
0.58
-
2.67’
0.73 0.42
-
-
-
-
separation
No
Yes
12 24
2.15 2.61
t
-
2.42c
Paternal
0.51
2.18 2.50
0.42 0.59
166b
1.83
0.40
2.19 267d
0.37
17
-
-
-
-
-
-
1.79
0.54
41.60
-
-
No
18
-
-
-
-
-
-
2.13
0.49
19.80
-
-
-
-
-
-
-
-
198b
t
-
Adoption Yes
3
No t
Extrusion Yes No
from
31
1.77 2.58
-
255b
8 28
56.10
-
-
-
-
54.40
0.50
15.40
-
-
1.60
-
-
24.30
1.58 2.13
_
_
4.21e
_
_
167b
252d
8.10
1.S7e
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
21.70
0.22 0.37
47.80 8.30 355C
-
-
-
-
180b 1.79
abuse
10
No
21
2.48 2.64
t
-
180b
physically
4 31 -
Psychological physically
0.52 0.58
-
abusive
t
1.97
0.64
2.57 193b
0.57
2.16 2.60
0.44 0.61
40.00 16.70 201b
16.20
1.70
-
-
-
-
27.50
7.50
314d
188b
father abusive
Yes
13
No
22
t
one-tailed
0.85
-
Yes
aDashes
199b
home
t
Mother Yes No
6.60 259d
separation
Yes
Sexual
26.80
-
indicate
nonsignificant
-
214b
differences.
All data
analyzed
by
t test,
df=29-35
(depending
on
missing
data),
two-tailed
for grossly
inappropriate
parenting,
elsewhere.
b