Object relations in adolescence: A comparison of normal and inpatient adolescents


Hideko Sekiya, MD, Kanto Central Hospital, 6-25-1, Kamiyoga, Setagaya-ku, Tokyo 158-8531, Japan. Email: sekiya@kanto-ctr-hsp.com


Aims:  We aimed to study the development of object relations in adolescents and their correlation with their mothers' defense styles in inpatient and normal adolescents.

Methods:  We administered the Thematic Apperception Test to adolescents in the adolescent unit (junior high, n = 16; senior high, n = 22) and normal controls (junior high, n = 16; senior high, n = 16). Results were analyzed using the Complexity of Representations Scale (CRS). We administered the Defense Style Questionnaire (DSQ40) to the subjects' mothers (patients, n = 38; controls, n = 32) to determine whether adolescents' CRS scores correlated with mothers' DSQ scores.

Results:  There was a nearly significant interaction for group-by-school-year for the children's CRS scores. In the control group, senior high school students' scores (mean [SD] = 3.52 [0.49]) were significantly higher (F [1,66] = 12.3, P = 0.001) than those of junior high school students' (mean [SD] = 3.03 [0.31]). In the patient group, no significant difference was observed between senior high and junior high. For mothers' DSQ40, mature defense scores were significantly higher in the control group than in the patient group (mean [SD] = 10.8 [1.89] vs 9.35 [1.40] in junior high, and 11.8 [1.67] vs 9.36 [1.81] in senior high, F [1,66] = 22.1, P < 0.001, two-way anova). A significant, positive correlation (r = 0.37, P = 0.04) was observed between the mothers' mature defense and the children's CRS scores in the control group only.

Conclusions:  Whatever diagnoses are provided, the problems of adolescents with non-psychotic pathologies are related to the arrest of object relations development. A patient's mother cannot employ mature mechanisms to alleviate signals of anxiety sent by her child.

MOST CHILDREN HOSPITALIZED with non-psychotic psychopathologies in the adolescent psychiatric unit are unable to establish peer relationships appropriate to their age, whatever diagnoses they have been given. In addition, various problems arise in their family adaptation that make outpatient treatment not feasible. This leads to hospitalization.

At the formation of a treatment program, the patient's developmental and family history, focusing on the parent–child relationship, should be taken into account, together with clinical diagnosis based on international diagnostic classification. The program should be grounded on a genetic-psychodynamic formulation that makes the assessment of development possible.1,2 However, to quantify a genetic-psychodynamic formulation is exceedingly difficult. Hence, we considered the possibility of quantifying the object relations development of adolescent patients as one surrogate measure of the genetic-psychodynamic formulation of development. Our hypothesis was that if we were able to make an empirical assessment of object relations, formed under the influence of libido, ego and superego, there would emerge a clearer understanding of the psychopathology of adolescent patients as well as a more objective treatment provision. In order to achieve this, we employed Westen's measures of object representation assessment, the only established methodology of experimental study within the framework of psychoanalytic development. In other words, adolescent inpatient's development of object relations was assessed by using the Social Cognition and Object Relations Scale (SCORS) developed by Westen3,4 and compared with normal adolescents that were free of psychiatric diagnosis.

Westen et al. examined the correlation between the degree of mental health and the internal object representation of borderline personality disorder.5–10 Ackerman11et al. compared patients with antisocial personality disorder, borderline personality disorder, narcissistic personality disorder, and cluster C personality disorder, and concluded that SCORS can be a clinically effective index to differentiate these groups. Ornduff12 also used SCORS to assess object representation of maltreated children. In Japan, Ikegami13 reported that SCORS measures of adolescents placed in age-appropriate nursing homes are significantly lower than those of adolescents living with their parents. These studies support SCORS as an objective psychological evaluation tool for the assessment of object representation development. A high score on Complexity of Representations Scale (CRS), which is one of four SCORS subscales, means favorable development of object relations.

Adolescent development evolves out of the interaction between inherent predisposition and acquired influences from the environment (epigenetic theory). Therefore, we also investigated the impact of the mother's influence on her child by objective assessment of the degree of the mother's mental health, using the Defense Style Questionnaire (DSQ). This was associated with her adolescent child's development of object relations.14–17



The subjects in this study consisted of patients hospitalized in the adolescent unit, their mothers, and normal controls and their mothers. There were 38 junior high school and senior high school inpatients (11 boys and 27 girls excluding those with schizophrenia, bipolar disorder, mental retardation), and their mothers. The patients were diagnosed with DSM-IV18 by clinicians. Among those, six were diagnosed with depressive disorder, four with anxiety disorder, two with obsessive–compulsive disorder, eight with anorexia nervosa, nine with bulimia nervosa, four with conversion disorder, three with hypochondriasis, and two with borderline personality disorder. The patients had been previously admitted to the adolescent psychiatric unit of X General Hospital for 8 consecutive months, in 2009. Normal controls consisted of 32 normal students (10 boys and 22 girls) from Y private junior and senior high schools, as well as their mothers. They were all interviewed to ascertain that they did not have any particular mental health problems. They were matched in sex and age with the patient group.

The average age of the patient group was 14.2 years (SD = 0.8) for junior high school students, and 17.1 years (SD = 0.8) for senior high school students. The average age of mothers of the inpatient junior high school students was 42.5 years (SD = 4.5) and 45.6 (SD = 4.4) for the mothers of the inpatient senior high school students. The average age of the normal junior high school students was 14.2 (SD = 0.8), and that of the normal senior high school students was 17.0 (SD = 0.9). The average age of the mothers of the normal junior high school students was 42.3 (SD = 4.5) and 46.8 (SD = 3.5) for the mothers of the normal senior high school students.

At the start of the survey, we fully explained the purpose of our study in writing, to patients and mothers and obtained written consent from both. Regarding the control group, in addition to the consent of the students and parents cooperating in the study, we explained the purpose of our study to the principals of the junior and senior high schools and obtained their consent also. We conducted this study with the endorsement of Ethics Committee of X General Hospital.



Westen3 developed the SCORS to measure mental representations (images of oneself and others pictured in one's mind) and object relations, using the responses to the Thematic Apperception Test (TAT). SCORS has four subscales. Among those four subscales, Complexity of Representations Scale (CRS) was selected for its usefulness in measuring the object-relations development.

The seven TAT cards (Cards 1, 2, 3GF, 4, 13MF, 15, 18GF) were given at the time of hospitalization. All TAT responses were coded by four raters blind to diagnosis and other clinical information. Coders were given stories typed one to a page in random order, so that the ratings of multiple stories in the same protocol would be independent. Reliability was computed using Pearson's R, with Spearman–Brown correction for multiple coding. Reliability for the scale was 91%. Coders met at regular intervals to discuss independently scored responses in order to prevent coder drift and research consensus scores.

CRS has five levels. Level 1 is most immature and pathologically severe and Level 5 highest in object-relations maturity. At Level l the person does not see others as clearly differentiated or bounded, and/or does not differentiate his/her own thoughts and feelings from those of others. At Level 5 the person sees people in complex ways, making elaborate inferences about their mental states, motivations, points of view, and unconscious processes. People are seen as having conflicting feelings and dispositions, and as expressing different aspects of their personalities in different situations. Thus a high level of CRS indicates healthy development.


The DSQ is a self-filled questionnaire form to measure defense mechanisms. It was released by Bond et al. in 1983 and is designed to overcome the problem of reliability among raters.14–16 DSQ employs the special term, ‘defense style’, instead of defense mechanism, for the self-assessment of defense mechanism that functions unconsciously as a reaction to a stimulus from outside. This study uses the 40-item version of the DSQ (DSQ40), which can be conducted in a shorter time.17

The DSQ40 can measure the following 20 defense styles: splitting, denial, projection, acting out, autistic imagination, devaluation, passive-aggressive, isolation, dissociation, displacement, rationalization, somatization, undoing, idealizing, reaction formation, altruistic surrender, sublimation, humor, prediction and inhibition.

Defense Styles are classified into three classifications based on Kaplan's categories: mature defense (humor, inhibition, sublimation, prediction); neurotic defense (reaction formation, rationalization, displacement, isolation, dissociation); and immature defense (projection, denial, passive-aggressive, acting out, somatization).19

Statistical analysis

Assessment was performed on the subjects' reactions to the seven TAT cards and the average score was calculated. Subsequently, the average score of each subject group was used for later analyses. DSQ40 was conducted on mothers of patients and controls. Assessment was made on their mature defense, neurotic defense and immature defense and scores on each defense style were calculated. To ascertain whether children's CRS scores and mothers' DSQ40 scores were different according to groups (control group and patient group) and school year (junior high school and senior high school), a two-way anova was conducted. Furthermore, correlation between children's CRS scores and mothers' DSQ40 was investigated using Pearson product–moment correlation coefficient. P < 0.05 was considered significantly different (two-tailed test). Statistical software, spss 12.0 (spss, Chicago, IL, USA), was used.


Table 1 shows the results of two-way anova using the children's CRS and the mothers' DSQ40 as dependent variables.

Table 1.  Influence of group and school year on CRS score and DSQ40 score (two-way anova)
 Normal controls (n = 32)Patients (n = 38)Main effect of groupMain effect of school yearGroup by school year interaction
Junior high (n = 16)Senior high (n = 16)Junior high (n = 16)Senior high (n = 22)
MeanSDMeanSDMeanSDMeanSDF (1, 66)PF (1, 66)PF (1, 66)P
  1. CRS, Complexity of Representations Scale; DSQ40, 40-item version of the Defense Style Questionnaire.

Child's CRS Score3.030.313.520.492.830.302.940.4217.1<0.00110.10.0023.80.054
Immature Defense6.662.006.681.267.842.807.692.673.880.0530.080.900.020.89
Neurotic Defense8.501.067.761.297.881.757.
Mature Defense10.81.8911.81.679.351.409.361.8122.1<0.0011.420.240.330.25

When employing the children's CRS scores as a dependent variable, a nearly significant interaction was observed for group-by-school year. The influence of school year on the CRS score tended to be different for normal and patient adolescents. In the control group, the scores of senior high school students were significantly higher (F [1,66] = 12.3, P = 0.001) than those of junior high school students. In the patient group, however, no significant difference was observed (F [1,66] = 0.80, P = 0.37) between senior high school students and junior high school students. Among junior high school students, no significant difference was observed (F [1,66] = 2.21, P = 0.14) between the patient and control groups, whereas among senior high school students, the scores of the control group were significantly higher (F [1,66] = 20.0, P < 0.0001) than those of the patient group.

Using the mothers' DSQ40 as a dependent variable, only the main effect of group was significant for mature defense. Regardless of children's school year, mothers' mature defense scores were higher in the control group than in the patient group. For neurotic defense scores, no significant difference was observed for the main effects of group, of school year, and of group × school year interaction. Concerning immature defense, a nearly significant main effect of group was observed. Regardless of school year, more mothers of child patients tended to show immature defense than mothers of normal children.

Table 2 shows the correlation between the children's CRS scores and mothers' DSQ40 scores. A significant, positive correlation was observed between the mothers' mature defense and the children's CRS scores in the control group only. No significant correlation was observed in the patient group.

Table 2.  Pearson product–moment correlation coefficient (r) between child's CRS Score and mother's DSQ40 score in normal control group and patient group
Mother's DSQ40Normal controls (n = 32)Patients (n = 38)
  1. CRS, Complexity of Representations Scale; DSQ40, 40-item version of the Defense Style Questionnaire.

Immature Defense−0.1780.33−0.0670.69
Neurotic Defense−0.2280.21−0.1160.48
Mature Defense0.3700.040.0110.95


Children's development of object relations

The results of this research indicate two findings. When the control and patient groups are compared, first, the object relations develop in the transition period between junior high school and senior high school for the normal group; and second, in the patient group no object relations development is seen in the same transition period.

The CRS scores were significantly higher in the normal senior high school students than in the normal junior high school students. The result corresponds to findings on the development of object relations by Westen20 where CRS scores were significantly lower in third-year students of junior high than in third-year students of senior high school. On the other hand, the comparison of the CRS scores of the patients in junior high and the patients in senior high school shows that although the mean value of the former is slightly lower, there is no significant difference. This result indicates that the emergence of a certain non-psychotic pathology in adolescence reduces or cuts off the patient's interaction with friends of the same age and sex, which arrests further development of object relations. The parent–child relationship of many inpatients, just before hospitalization, was infantilized (abusive language, violence, not taking meals with the family, sleeping with mothers, taking a bath with the parent of opposite sex and the like). That is to say, after the onset of the illness, peer relationships were cut off or reduced and an infantile relationship with the mother or father emerged. Such a change in the external world of reality cannot provide the proper environment for the development of internal object relations and indeed obstructs it.

Ikegami compared the CRS scores of adolescents (average age of 15.9 years) living in an institution for maltreated children with those of normal children and reported that the scores of the former were significantly lower than those of the latter. In other words, the CRS scores of the group that had been subject to abuse and neglect in early childhood, under circumstances of deprivation, were low. In our group, however, it was only in the senior high school student groups that a significant difference was observed between the normal children and the patients. No significant difference was seen in the junior high school student groups. Ikegami's subjects lived in an institution and initially had lower CRS scores than our subjects. The degree of their development interference was much stronger than ours. Our subjects managed to develop in nearly the same way as the controls until they reached junior high school age. At senior high school, however, their progress to the further stage was disrupted. Although the children had been spared such marked developmental interference as experienced by the group housed at an institution for the maltreated, minor problems in the parent–child relationship are intensified by the sexual development of the child and hamper object relations in adolescence. Our present study has analyzed and qualified this reality. To my knowledge, there is no experimental study comparing the development of object relations between normal junior and senior high school students and patients in junior and senior high school. Ours is the first observation of the kind.

Child's development of object relations and mother's defense style

As shown in Table 1, the mature defense of control group mothers was shown to be significantly higher than that of patient group mothers. Conversely, the immature defense of control group mothers was lower than that of patient group mothers, but not significantly lower. There is a tendency for the immature defense of control group mothers to be less than that of patient group mothers.

As shown in Table 2, no significant correlation was observed for the CRS scores of the patient group and their mothers' DSQ40 scores. On the other hand, there was a significantly positive correlation between the CRS scores of the control group and the mature defense style of their mothers.

These findings are highly suggestive clinically. That is, the mature defense is observed in mothers of controls, which differentiates them from mothers of patients. When mothers encounter words, actions and impulsive behavior from their children that they have never experienced before, mothers of the controls employ mature defense. Hence, they do not throw children's uncontrollable feelings and impulses right back at them and are able to stabilize their children's state of mind. On the other hand, mothers of patients, not being able to employ mature defense, handle children's words, actions and vents of impulse only with neurotic or immature defense. Therefore, they cannot calm down their children but rather argue with them or ignore them only to exacerbate their relationship to a destructive level, building a vicious circle together with their children.

Therapists working on treatment should decrease the mother's feelings of anxiety so that she will not hurl her immature defense at her child as an extension of her own unease. Therapists must also take care not to contribute to the mother's sense of guilt and responsibility. At our institution, we have conducted parent guidance sessions to enable mothers to face their children without resorting to immature defense. The results of our study verify such an approach as appropriate.

Matters of common knowledge in clinical experience and theories have been quantified in the present study, which makes it meaningful. As CRS is closely connected to various functions of ego and superego, object relations score assessment might be useful in evaluating therapeutic effect. This will be our challenge for the future. The use of CRS scores, assessed at the time of hospitalization and discharge, to measure therapeutic effect is a potential subject of our next study.

Limitation of present study

There has been repeated discussion on the male–female difference in object relations development in the clinical field. However, male–female difference using CRS scores is not examined in this study. This constitutes material for further investigation. It should be noted that the findings obtained here derive from a relatively small number of inpatients in a ward of a single hospital. This imposes limits on wider generalization.


In our study, we found out that whatever diagnoses are given, problems of adolescents with non-psychotic pathologies are related to the arrest of object relations development. Actual symptoms are withdrawal from peer relationship and infantilization of parent–child relationship. In such an infantilized relationship, the mother of a patient cannot employ mature mechanisms to alleviate signals of anxiety sent by her child. The present study reveals that here resides the difference between control mothers and patient mothers.

The experimental data obtained through this study correspond to clinical experience in adolescent and young adulthood psychiatry and development theory. The study also clearly defines the goal of our hospitalization therapy as the recommencement of patients' object relations development.


I would like to express my heartfelt gratitude to Professor Kuninao Minakawa, Dr Yuko Miyake, and Professor John C. Maher. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. There is no conflict of interest to declare.