Relatedness and self-definition (communion and agency) develop normally in a synergistic dialectic interaction throughout life (1–3) and are central dimensions in personality development (1, 4, 5). While most individuals try to maintain a balance between these two major foci of experience, individuals within the normal range usually place somewhat greater emphasis on one or the other, creating two basic personality styles in which people experience and engage life differently (1, 6).

Severe disruption of this normal dialectic developmental process may result in a distorted, one-sided preoccupation with one of these dimensions at the expense of the other (1, 3). Intense preoccupation with establishing and maintaining interpersonal relatedness is expressed in preoccupations with experiences of merger, closeness, trust, caring, intimacy, and sexuality to the neglect of the development of a sense of self and is associated with psychopathologies of an anaclitic configuration, including undifferentiated schizophrenia and borderline, dependent, and histrionic personality disorders. Preoccupation with establishing and maintaining a viable sense of self is expressed by feeling separate, autonomous, independent, and worthwhile and is associated with psychopathologies of an introjective configuration, including paranoid schizophrenia and paranoid, obsessive-compulsive, self-critical depressive, and narcissistic personality disorders (3). Clinical judges can reliably differentiate these two types of patients from intake case reports (7, 8).

Anaclitic and Introjective Psychopathology

The value of the anaclitic-introjective diagnostic distinction has been demonstrated in research on depression and personality disorders. Studies using the Depressive Experiences Questionnaire (9, 10) and similar instruments, such as the Sociotropy-Autonomy Scale (11), have identified two types of depression, one focused on interpersonal loss and feelings of abandonment (anaclitic depression) and the other on issues of self-worth (introjective depression) (6, 12), and have identified early and current life situations that contribute to the onset of these two types of depression, their unique personality and clinical characteristics, and their differential responses to therapeutic interventions (6, 13). Investigations of outpatients (14, 15) and inpatients (unpublished 1995 paper of K.N. Levy et al.) with personality disorders found that the axis II personality disorders are organized around issues of relatedness and self-definition. Patients with dependent, histrionic, and borderline personality disorders (anaclitic disorders) had greater preoccupation with issues of relatedness. Those with paranoid, schizoid, schizotypic, antisocial, narcissistic, avoidant, obsessive-compulsive, and self-defeating personality disorders (introjective disorders) had greater preoccupation with issues of self-definition.

Anaclitic and Introjective Psychopathology and Therapeutic Interventions

Judges in several studies (7, 8) reliably differentiated anaclitic and introjective patients on the basis of intake evaluations and investigated the patients’ differential responses to brief and long-term intensive outpatient and inpatient treatment. Although introjective patients did poorly in brief outpatient psychotherapy and with imipramine treatment (13), they were responsive to interpretive exploration in long-term intensive psychodynamic treatment (7, 8, 16, 17), whereas anaclitic patients seemed responsive primarily to supportive relational aspects (8, 16). Although few differences in symptom severity are observed initially between anaclitic and introjective patients (7, 8, 18), these two patient types express therapeutic change along different dimensions in long-term intensive psychodynamic inpatient treatment. Seriously disturbed anaclitic patients changed on measures assessing the quality of interpersonal relatedness, while introjective patients had significant increases in cognitive efficiency (IQ) and reduction in symptoms on dimensions most relevant to their personality organization (7).

One important and puzzling inconsistency was that while it was expected that thought disorder would diminish primarily in introjective patients because of their ideational orientation, this measure significantly declined only for anaclitic patients (7). In this study, we explore this inconsistent finding by deconstructing the composite thought disorder measure into its components.

Anaclitic and Introjective Personality Organization and Therapeutic Change in Types of Thought Disorder

Thought disorder is a persistent and important topic in severe psychopathology. Blatt and Ritzler (19) differentiated three major types of thought disorder according to degrees of boundary disturbance and demonstrated that these types of thought disorder form a continuum of severity. Boundary differentiations include the capacity to differentiate between independent objects, including self and nonself, and between the actual object and the mental representation or verbal signifier used to designate the object—that is, between outside and inside. Failure to maintain these fundamental differentiations occurs primarily in psychosis and severe borderline personality disorder (20–23). Difficulties in maintaining these boundary differentiations are expressed in thought-disordered responses to the Rorschach test—in contaminations, confabulations, and fabulized combinations (19).

Contaminations, considered pathognomonic of schizophrenia (21–23), express an inability to establish and maintain a fundamental boundary between independent objects, events, or thoughts. Percepts become fused in a distorted response (e.g., an accurately perceived “hand” and “rabbit’s head,” on the bottom of Rorschach card X merge into an idiosyncratic “rabbit’s hand”).

Confabulations express an inability to maintain a boundary between inside and outside, between what is perceived and what one thinks or feels about the perception. Confabulations are characterized by extensive arbitrary ideational or affective elaborations (personal reactions) that seriously distort an accurately perceived response (e.g., “two fetuses, representing good and evil, heaven and hell,” on the top of card II). Confabulations occur in seriously disturbed borderline patients (22, 24).

Fabulized combinations, the least serious of the three thought disorders (19, 21, 22), reflect an attribution of arbitrary relationships between independent objects because of spatial or temporal contiguity (e.g., “two elephants dancing on a butterfly” on card II). They occur primarily in more organized outpatient borderline and depressed patients (22), often with paranoid features.

These thought disorders can be scored reliably (7, 19, 22). In a study of therapeutic change in long-term intensive inpatient treatment of seriously disturbed patients, Blatt and Ford (7) constructed a composite score in which these three thought disorders received differential weighting based on the relative severity of their psychopathology. Contaminations and contamination tendencies received the highest weighting (scored 6 or 5), confabulations and confabulation tendencies received moderate weighting (scored 4 or 3), and fabulized combinations and tendencies received the lowest weighting (scored 2 or 1). This weighted composite thought disorder measure diminished significantly in long-term intensive treatment only in anaclitic patients (7), contrary to prediction.

The Present Study

We deconstructed the weighted composite thought disorder measure to evaluate possible differential changes in its three components. We expected therapeutic change in anaclitic patients to occur primarily in contaminations because this type of thought disorder indicates tendencies to merge and fuse with an other, an expression of primitive anaclitic difficulties with interpersonal relatedness (3, 7, 19, 21). In contrast, therapeutic change in more ideational introjective patients was expected to occur primarily in fabulized combinations because this type of thought disorder reflects a tendency toward referential thinking that expresses the overideational orientation of introjective disorders (3, 7, 21, 24). We were unclear, however, about confabulations in the midrange of boundary disturbances—whether they express the boundary disturbances and tendencies to merge and fuse seen in anaclitic patients or the referential tendencies seen in introjective patients. Thus, this aspect of the study was exploratory.

Participants and Design

Our analyses were based on clinical data that are routinely gathered as part of long-term psychoanalytically oriented intensive inpatient treatment of seriously disturbed, treatment-resistant young adults (ages 18–29 years at admission) at the Austen Riggs Center. We used data from 90 patients for whom Rorschach responses were available early and later in the treatment process. Most patients had extensive prior outpatient treatment (on average, 28 months), and on average, patients had 1.5 brief psychiatric hospitalizations before admission to the treatment program. Two judges reliably distinguished anaclitic and introjective patients (kappa >0.75) based on initial clinical case records, independent of the psychological testing. The judges agreed on 17 of the 18 cases they had both rated to assess interrater reliability. Judges making the anaclitic-introjective distinction in two studies of long-term treatment (7, 8, 25) reported that the primary criterion they used in making this distinction was the content of the patients’ predominant concerns (i.e., interpersonal relatedness or self-definition) and the predominant defenses the patients utilized (i.e., avoidant, such as withdrawal, denial, and repression; or counteractive, such as projection, intellectualization, reaction formation, and overcompensation) (1, 3).

Of the 90 patients we studied, 42 were rated as anaclitic and 48 as introjective. Approximately 30% of the patients had psychotic diagnoses (e.g., schizophrenia and psychotic depression); 60% were diagnosed as having severe personality disorder and 10% as having severe depression. The proportion of patients at these three levels of psychopathology was not significantly different in the anaclitic and introjective groups.

Measures and Procedures

A total of 180 Rorschach protocols were scored—two for each patient, one at the beginning and the other at midtreatment (on average, after 15 months of treatment and 10 months before discharge). The judge who scored them had established acceptable levels of interrater reliability in scoring the three types of thought disorder (intraclass correlations >0.70). The judge rated the protocols in random order, uninformed about patients’ background, about which protocols were from the same patient, and about when the Rorschach protocols were administered. Differences in therapeutic response between anaclitic and introjective patients were explored by evaluating change in the three types of thought disorder from pretreatment (time 1) to 15 months later in the treatment process (time 2).

At times 1 and 2, contamination and contamination tendency responses represented the contamination factor, confabulation and two forms of confabulation tendency responses represented the confabulation factor, and fabulized combination and two forms of fabulized combination tendency responses represented the fabulized combination factor. These thought disorder scores at times 1 and 2 were residualized for overall response productivity on the Rorschach and subjected to principal components factor analysis. The results indicated that the two types of contamination responses loaded on one factor at time 1 (Eigenvalue=1.38, loadings of 0.83 and 0.82, accounting for 68.8% of the variance) and at time 2 (Eigenvalue=1.32, loadings of 0.80 and 0.81, accounting for 66.2% of the variance). The results also indicated that the three types of confabulation scores loaded on one factor at time 1 (Eigenvalue=1.63, loadings of 0.86, 0.45, and 0.88, accounting for 54.2% of the variance) and at time 2 (Eigenvalue=1.72, loadings of 0.81, 0.66, and 0.79, accounting for 57.2% of the variance). Finally, the results indicated that the three types of fabulized combination scores loaded on one factor at time 1 (Eigenvalue=1.47, loadings of 0.85, 0.86, and 0.43, accounting for 48.9% of the variance) and at time 2 (Eigenvalue=1.31, loadings of 0.78, 0.77, and 0.40, accounting for 43.8% of the variance). Standardized scores of each of the three factors were computed for times 1 and 2 and served as the factor scores for contamination, confabulation, and fabulized combination responses in a general linear models (GLM) repeated-measures multivariate analysis of variance (MANOVA).

Changes in the three types of thought disorder over the course of treatment were compared in a 223 GLM repeated-measures MANOVA with the two patient groups (anaclitic and introjective) as the independent variable, the two times (pretreatment and later in treatment) as the within-subjects independent variable, and the three mean standardized thought disorder factor scores (contamination, confabulation, and fabulized combination) as the within-subjects repeated-measures dependent variables. (The means and standard deviations for contamination, confabulation, and fabulized combination responses for anaclitic and introjective patients at times 1 and 2 are available in a data supplement that accompanies the online version of this article.) Anaclitic patients at time 1 tended to have more contamination and confabulation responses than fabulized combinations, while introjective patients had primarily fabulized combination responses and very few contamination and confabulation responses. Although these differences at time 1 were not statistically significant, the results indicated significant patient differences at time 2, with introjective patients having significantly fewer fabulized combination responses than anaclitic patients (F=10.34, df=1, 88, p<0.002) and anaclitic patients having significantly fewer confabulation (F=14.09, df=1, 88, p<0.0003) and contamination (F=5.08, df=1, 88, p<0.03) scores than introjective patients.

The GLM repeated-measures analysis of variance yielded four statistically significant effects (the data from the analysis are available in the data supplement that accompanies the online version of this article). According to h_p² (partial eta-squared), a measure of effect size, these four effects accounted for 60% of the total variability in thought disorder scores:

1. A main effect for time (F=16.58, df=1, 88, p<0.0001; η_p² effect size=0.17, observed power=0.98).

2. Two 2-way interactions: (a) thought disorder by patient group (F=8.55, df=2, 176, p<0.0003; η_p² effect size=0.16, observed power=0.96); and (b) time by thought disorder (F=3.47, df=2, 176, p<0.03, η_p² effect size=0.06, observed power=0.54).

3. As indicated in Figure 1, a highly significant higher-order three-way interaction of time by thought disorder by patient group (F=11.84, df=2, 166, p<0.00002; η _p² effect size=0.21, observed power=0.99).

Since the main effect and the two-way interactions are nested within the higher-order three-way interaction, we focus on the higher-order interaction.

Probing the Significant Three-Way Interaction

Consistent with earlier findings with other measures (7, 8), differences between anaclitic and introjective patients on all three types of thought disorder at time 1 were not statistically significant. Results of the GLM analyses, however, consistent with our main hypothesis, indicated a highly significant decline in confabulations and contaminations for anaclitic patients (F=12.69, df=1, 88, p<0.0006 and F=9.96, df=1, 88, p<0.002, respectively), but with no significant change in their fabulized combinations. Comparisons between the significant declines in confabulation and contamination responses in anaclitic patients indicated no significant difference between the changes in these two types of thought disorder responses in anaclitic patients.

Results also indicated a highly significant decline (F=20.74, df=1, 88, p<0.00001) in fabulized combinations in introjective patients, but no significant change in their confabulations and contaminations. Comparisons between the significant decline in fabulized combinations in introjective patients and the significant decline in confabulations and contaminations in anaclitic patients indicated a significantly greater reduction in fabulized combinations in introjective patients than in confabulations and contaminations in anaclitic patients (F=22.31, df=1, 88, p<0.00001 and F=12.41, df=1, 88, p<0.00001, respectively) (Figure 1). Although seriously disturbed anaclitic and introjective patients exhibited no significant difference in types of thought disorder at baseline, they expressed therapeutic progress in significantly different ways, indicating that therapeutic change in seriously disturbed anaclitic patients involves a reduction in primitive forms of relatedness expressed in themes of merger and fusion and that therapeutic change in seriously disturbed introjective patients involves a reduction of referential thinking.

Differential changes in thought disorder in seriously disturbed anaclitic and introjective patients in long-term intensive inpatient treatment add to accumulating empirical findings and clinical observations indicating the value of this diagnostic distinction for understanding psychopathology and the processes of therapeutic change. This distinction has enabled investigators to identify two groups of patients who are differentially responsive, sometimes in different ways and possibly through different mechanisms, to different types of therapy (7, 8, 13, 16, 26, 27). The anaclitic-introjective distinction also seems useful in investigating possible mechanisms of therapeutic action in the interaction between different types of patients and treatments as well as for possibly matching therapy and patient (8, 13, 16, 27).

In contrast to the descriptive symptom-based differences in DSM-IV, the differentiation between anaclitic and introjective configurations of psychopathology is based on psychodynamic considerations, including differences in primary motivational focus (libidinal versus aggressive), types of defensive organization (avoidant versus counteractive), and predominant character style (emphasis on an interpersonal versus self orientation and on affect versus cognition). The anaclitic and introjective configurations of personality development and psychopathology provide a comprehensive theoretical structure for identifying fundamental similarities among many forms of psychopathology and for maintaining conceptual continuity among processes of psychological development, normal variations in character or personality organization, and different forms of psychological disturbance (1–3). In this view, psychopathological disorders are in part compensatory exaggerated and distorted responses to severe developmental disruptions of the reciprocally balanced normal dialectic development of interpersonal relatedness and self-definition. Severe disruptions of this developmental process result in exaggerated attempts to achieve equilibrium through either an intense distorted preoccupation with the quality of interpersonal relatedness or exaggerated defensive efforts to consolidate the sense of self. This approach is consistent with the recent emphasis (28) on seeking a dimensionally organized taxonomy of psychopathology based on a few broad, overarching constructs or multiple dimensions of disordered thought, affect, behavior, temperament, and personality (29–31) that “transcend a putative distinction between more normal and more abnormal psychological phenomena” and the “official nosologies such as DSM” (28, p. 491).

This study’s findings should be interpreted within the context of specific limitations. We did not examine the data for possible third variables to account for the findings. In addition, because of the lack of specification about the details of the treatment process and differences among therapists, we were unable to explore possible mechanisms of therapeutic change. Future studies should include measures of dimensions of the therapeutic process and how these dimensions might account for differential changes in anaclitic and introjective patients. Finally, random assignment of patients as well as a control group might facilitate examination of complex questions, including what kind of treatment is best for what kind of patient (8, 27), with what kind of therapist (6, 13, 27), and leading to what kind of therapeutic change (7, 8).

In summary, the findings of a differential change in thought disorder in anaclitic and introjective patients, consistent with findings from a number of other studies, indicate that formulations of anaclitic and introjective configurations of personality development and psychopathology may provide a comprehensive theoretical structure for identifying fundamental similarities among many forms of psychopathology as well as their continuity with dimensions of personality development, variations in normal personality organization, and processes of therapeutic change (32).

An Improved Anaclitic Woman

At admission, “Ms. I,” a woman in her late teens, showed histrionic, obsessional, and paranoid qualities and a degree of interpersonal withdrawal. The working diagnosis at admission was of a long-standing depression in a histrionic character with serious borderline features. Ms. I’s parents had divorced when she was about 10 years old. At age 16, a year before her mother remarried, Ms. I gave birth to a child, placed it for adoption, and seemed to have suffered a postpartum depression.

At the case conference at time 2, her therapist described a long beginning phase of the therapy in which Ms. I behaved as if she were a small child, requiring the therapist to enact the role of protective parent. However, when the therapist refused to tolerate her coercive regression, Ms. I swung abruptly from being a clinging child to becoming a promiscuous woman, albeit without much feeling and without clear definition.

Particularly noteworthy in terms of thought disorder on the Rorschach protocol was the change of a response to card IV. At time 1, Ms. I reported perceiving an “angel fetus”: “This part in here (small white area in the center) sort of looks like a fetus. It’s got an umbilical cord, sort of misplaced a little, I think. A fetus with wings. (She laughs.) Like it’s an angel fetus.” (Angel fetus?) “The head and the little ass, and its feet were sort of coming down; they weren’t curled up. I guess it couldn’t be inside the womb in that position. It was a little high up, but it was attached to something in the center.” This elaborate response of a humanoid figure (a fetus) in a small white space embedded in the center of the blot was of inaccurate form and extensively elaborated with contamination tendency and confabulation (an “angel fetus”) and fabulized combination (a “fetus with wings”) thought disorder. At time 2 testing, this extensively described, securely ensconced thought-disordered “angel fetus with wings” becomes an unenclosed, largely undefined “person.” . . . “Some man. Who, I don’t know. He’s not pretty.”

At time 1, Ms. I’s autistically elaborated, symbiotically ensconced tiny human figure within a largely undefined other is passive and helpless as well as powerful and grandiose. By time 2, Ms. I used this same area of the blot to reveal an impoverished representation of an undesirable adult figure in an ambiguous, somewhat more benign surround. This shift in human representation corresponds well with Ms. I’s movement from a primitive merger transference to a relatively more distant, more separated involvement with her therapist, a movement partially disguised by her adolescent promiscuity.

This example of a decrease in more primitive forms of thought disorder (contamination tendency and confabulation responses, typical of anaclitic psychopathology) in elaborating an inaccurately perceived humanoid form (a “fetus”) to a non-thought-disordered, less defined, unelaborated, but more accurately perceived “person” is consistent with indications in the clinical material of a move toward a higher level of organization within the anaclitic configuration, a move from a more dependent borderline to a more integrated histrionic level of organization.

Ms. I left the hospital prematurely 12 months later, primarily because of financial considerations, but it was also several months after a very close friend had left. While it was felt that she had definitely improved, her functional capacity to maintain commitments was seen as limited, indicating a continued need for the sustained involvement of significant adults to help modulate her ongoing severe sense of shame and doubt. Since these conditions clearly indicated her need for further treatment, her prognosis was guarded, depending on the quality of the therapeutic environment she might be able to discover.

An Improved Introjective Woman

“Ms. K,” a woman in her mid-20s, had experienced a growing sense of futility, depression, and detachment. She was becoming seriously confused and was no longer able to function in the daily tasks of work and living. Her diagnosis at admission was of decompensation in a long-standing severe obsessive character disorder with depressive and masochistic features. Sad and angry-looking, she hid her femininity. Clearly depressed, she withdrew from others, at times appearing tense and flushed, as if she were in a rage. She described herself as having two parts: her controlled, ordinary self and “something else.” She feared she could become violent. At her most depressed, Ms. K described herself as being completely detached and without feelings. She revealed great sexual confusion and was very much frightened by the prospect of intimacy. When she allowed the expression of some hope, she said that she wanted to establish a more effective identity—to be in contact with herself, to be a woman and not an overgrown child.

Her treatment was described at her time 2 case conference as having proceeded well. Crucial elements were her male therapist’s flexibility, playfulness, and availability for gradual attachment and for an increasing number of partial identifications.

In terms of thought disorder on the Rorschach protocol, the change in her responses to card VIII was noteworthy. At time 1, she reported, “This thing up here is a genie.” (Genie?) “Partly because it’s a peculiar-looking thing coming up behind two curtains, as if it might come out of a bottle. An ethereal sort of thing. That much is behind the curtain” (fabulized combination). “The rest of it is some kind of curtain—except for his head and arms—that he’s hiding behind. I don’t know what he’s doing.” (Curtain?) “It doesn’t particularly, but it’s just something that the genie is coming out of.” (Out of?) “No, I mean he’s coming over the top of it. Kind of like the wizard in The Wizard of Oz. The wizard comes out from behind the curtain” (confabulation tendency). At time 2, she responded to the same card: “Upside-down it looks like a torso, opened, so it’s flat and these two things . . . are lungs, and some ribs in the middle, and that’s [at bottom] a pelvis, and this, the spine [as she runs her finger up the middle of the card], and that’s [blue] muscles in the back, and this part [orange] muscles in front, and these two [center pink] are breasts, and it hasn’t got a heart. That’s unfortunate.” (Where had she seen something like this?) “Perhaps in my science lab when I was in ninth grade. I can’t think of any place since then. I have in my mind a fancy anatomy textbook with plastic layers in it and all done in bright colors.” Among the evidence of change are the reduction of thought disorder and Ms. K’s greater control over the urgency of her associations. At time 1, she reacted to the greater stimulation of the color of the Rorschach card as helplessly as to an unfolding drama. At time 2, the stimulus impinges less on her, and she is in more control in her intellectualized response of a fancy anatomy textbook with plastic layers.

This change from less primitive forms of thought disorder (confabulation tendency and fabulized combination responses, typical of introjective psychopathology) in a mystical powerful quasi-human form (a “genie”) hidden behind two curtains to a more intellectualized non-thought-disordered response of a layered anatomy textbook is consistent with indications in the clinical material of a move toward a higher level of organization within the introjective configuration, a move from a more grandiose paranoid to a more integrated obsessive-compulsive level of organization.

When Ms. K left the hospital some 4 months later, she was seen as consolidating her gains and advancing significantly toward a more mature feminine character. She evinced a steady pulling together of resources, increasing her explorations into both the world of people and the world of work. At the time of the mutually agreed-on discharge, she thought she would like to plan her future without seeking further psychotherapy.