Disturbances in the Self: A Source of Eating Disorders

Disturbances in the development of the self have been identified as important factors that contribute to the formation and persistence of eating disorders. Thus, they are an important focus for intervention to promote attitudinal and behavioral change.

In her early theoretical work, Hilde Bruch argued that anorexia nervosa (AN) is caused by the failure to develop a diverse set of identities or self-definitions.1, 2 Highly controlling and perfectionistic parenting was believed to limit the child’s opportunities to function autonomously and to interfere with development of a clear and richly elaborated self. Bruch suggested that the adolescent turns to body weight as a viable source of self-definition and as a means of compensating for the lack of a clear identity and for associated feelings of powerlessness and incompetence. From this perspective, the adolescent’s fixation on body weight and exaggerated desire to be thin are a maladaptive way of coping with identity deficits and of striving for a sense of self-definition, competence, and control.

Deficits Enhance the Persistence of AN and BN

More recently, other eating disorders researchers have similarly argued that disturbances in the self play a primary role in the formation of both AN and bulimia nervosa (BN). For example, Strober conceptualizes the failure to establish a clear and stable set of self-definitions as the core psychopathology underlying AN.3 However, he suggests that a genetically-based personality style of high stimulus-avoidance, low novelty seeking, and high reward dependence, rather than restrictive parental behaviors alone, inhibits the natural exploration necessary for normal self-development.

Vitousek and Ewald suggest that both genetic and environmental factors contribute to the failure to develop a clear and stable set of positive selves, leading to an over-reliance on environmental cues to define the self. 4 Boskind-Lodahl and, more recently, Schupak-Neuberg and Nemeroff similarly argue that, at its core, BN stems from the absence of a true self and from the overemphasis on physical appearance as a concrete solution to the absence of an authentic self. 5,6

A Lack of Clear Definitions

Although these theories point to disturbances in the self as an important cause of the eating disorders, several problems have limited their effectiveness as guides for research and practice. First, the eating disorder theories have not been based on a theoretical model of the self and generally have failed to provide clear and measurable definitions of the self-related constructs. While terms such as “self,” “self-concept,” and “identity” are addressed in the theories, precisely what is meant by these terms is unclear.

A second and closely related problem is that because the self-related terms are not adequately defined, the precise nature of the self-disturbance is not clear. While the theorists suggest that the absence of a clear, stable, and authentic self is central to the formation of the disorders, the self-construct has no clear referent. Thus, exactly what it means to have an unclear, unstable, or inauthentic self is not known. As a result of this lack of specificity, a majority of research on self-disturbances in the eating disorders has focused narrowly on one dimension of the self—global feelings toward the self or self-esteem. Although these studies have consistently shown that women with eating disorders hold negative feelings toward themselves, this research has not addressed the hypothesis of the absence of a clear and stable self as an important source of the eating disorders.

A Controlled Study of Identity Deficits

Recently, we completed a study using the self-schema model to investigate the role of the self in AN and BN. Within this model, the self-concept is defined as a complex, multidimensional cognitive and affective system. The cognitive components of the self-concept include self-schemas, which are memory structures that incorporate knowledge about the self in specific content areas that reflect particular importance, expertise, and meaningfulness to the person.

Self-schemas are stored in long-term memory, with the self-related information hierarchically organized (Ann Rev Psychol 1987; 38:299). Self-schemas are comprised of three distinct types of knowledge, including: (1) semantic knowledge, generalizations, or abstractions that reflect “who the self is” that are stored at the highest levels of the hierarchy; (2) episodic knowledge, special autobiographical memories nested at the lowest levels of the hierarchy; and (3) procedural knowledge, or action-based memories in the form of skills, rules, and strategies for making judgments, drawing inferences, and accomplishing goals relevant to the domain (Personality and Social Intelligence, 1987).

Self-schemas can be developed about any aspect of the person, including physical characteristics (short, fat, bald), social roles (mother, friend, student), personality traits (independent, outgoing), and areas of particular interest and skill (computer expert, cello player, tennis player). Because of the rich array of information encoded in the self-schema and its repeated activation, self-schemas are stable, enduring structures that are chronically accessible in working memory (J Personality & Social Psychology, 1988; 55:599).

Functional Memory Structures

Once established in memory, self-schemas are functional memory structures. They influence information processing, and organize, motivate, and regulate behavior. Studies have shown that self-schemas enable consistent, competent, goal-directed behavior, stability of the self-view and positive affect states.

In contrast to self-schemas, other, more peripheral, knowledge about the self is stored in less fully developed memory structures. (Figure 1 offers a schematic overview of the cognitive components of the self-concept.) These structures may be comprised of isolated episodic memories of the self in specific contexts and lack the abstract semantic conceptions derived from repeated experience in the content area. As a result, these peripheral self-conceptions are not chronically active in working memory and are less able to function as reliable guides for information processing and behavioral regulation.

Despite having equal ability and intention, persons without a self-schema in a given content area are more susceptible to environmental challenges. They are also less able to reliably translate their intentions into effective behaviors and less able to utilize social information to support their self-view. They also report higher levels of emotional distress and dissatisfaction.

In addition to individual differences in the content of self-schemas, differences also exist in the extent to which each self-schema reflects a positive or negative characteristic of the self. Furthermore, studies have shown important differences in the regulatory consequences of positive and negative self-schemas. Whereas positive self-schemas facilitate goal-directed behavior and positive mood states, negative self-schemas are associated with negative mood states and withdrawn, risk-avoidant behaviors (J Personality & Social Psychology, 74: 1364). The relative proportion of positive to negative self-schemas available in memory may be the cognitive foundation of observed differences in global self-esteem, the affective component of the self-concept. Studies have shown that in normal samples, persons with low self-esteem have more unstable peripheral self-conceptions and fewer positive self-schemas available in memory (Self-esteem: The puzzle of low self-regard, 1993) but these properties of the self-concept have not been systematically examined in clinical samples.

Based on the eating disorder theory of self disturbances and research that has shown that individual variation in the array of self-schemas influences emotional health and well-being, we predicted that women with an eating disorder would have fewer positive self-schemas available in memory and would be more likely to have a fat self-schema available in memory compared to controls. Furthermore, we hypothesized that the number of positive self-schemas would predict the availability of a fat self-schema, which would in turn predict body dissatisfaction and disordered eating behaviors.

Study Population and Methods

Our study included 79 women with a diagnosed ED and 34 women with no history of an eating disorder or other mental disorder (controls). The Structured Clinical Interview (SCID) was used to determine eligibility to participate. Of the 79 ED women, 26 met either full criteria for AN (n = 12) or subthreshold level criteria for the disorder (n = 14). Fifty-three women met either the full criteria for BN (n = 29) or subthreshold level criteria for the disorder (n = 24).

Zajonc’s card-sorting task was used to measure the number of valenced self-schema. Participants were given a stack of 52 blank index cards and asked to write down all descriptors that are important to how they think about themselves. Next, they were asked to rate each descriptor according to: (1) degree of self-descriptiveness, (2) degree of importance to one’s self-description, and (3) whether the descriptor was positive, negative or neutral. In keeping with previous research on self-schemas (Personality & Social Psychology Bulletin, 1997;23,:139), descriptors that were rated highly self-descriptive and highly important (8 to 11 on an 11-point scale) were identified as self-schemas. The number of positive self-schemas was computed by totaling the number of self-descriptors that met the criteria for a self-schema and were rated positive. The same method was used to compute the number of negative self-schemas.

The presence of a fat self-schema in memory was determined by the number of fat-related adjectives endorsed as self descriptive (me vs. not me judgment) and two information processing indicators of the presence of the self-schema in memory. Chronically accessible self-schemas more rapidly influence judgments than does more peripheral self information. Thus,positive judgments (“me” endorsements) about the descriptiveness of fat adjectives are made more quickly and negative judgments (“not me” endorsements) are made more slowly by individuals with a fat self-schema and these response time patterns are viewed as evidence of schema availability. Furthermore, because the fat adjectives presented for judgment “fit” with an existing, well developed self-schema memory structure, they are remembered more readily than when no fat self-schema exists. Thus, the second information processing indicator of the presence of a fat self-schema is the number of fat adjectives recalled after the judgment task.

Results Support the Concept of Disturbed Self

The results provide evidence to support the self-concept disturbance hypothesis. While women in the three groups did not differ according to the total number of self-schemas available in memory, women in both the AN and BN groups had fewer positive schemas compared to women in the control group. For women in the AN group, 57.5% of their self-schemas were positive, for women with BN, 60.5% were positive compared to the 82% of self-schemas that were positive for controls.

The three groups also differed according to the number of negative self-schemas, with both AN (29%) and BN (28.4%) reporting more negative self-schemas compared to control women (7%). Also as predicted, women in the BN group demonstrated a pattern of information processing suggesting that they have a “fat” schema available in memory. Controlling for the effects of BMI and general information processing differences, the BN group endorsed as self-descriptive a greater proportion of fat adjectives, and were slower to make “not me” judgments of fat adjectives compared to the control group. No differences were found in the recall scores.

When considered along with the fact that all but six women in the group were currently within a low normal-to-normal weight range (body mass index, or BMI: mean= 22.2, range, 18.2 – 27.9), these results support the hypothesis that women with BN have an unrealistic conception of the self as fat. How-ever, women in the AN group did not show clear information processing evidence of a fat self-schema.

Finally, fewer positive self-schemas predicted a higher fat self-schema score, which in turn predicted higher body dissatisfaction scores and higher disordered eating behaviors scores. Objectively measured BMI was also predictive of fat self-schema scores, and high BMI predicted higher fat self-schemas scores.

What Is The Proper Focus of Intervention?

The results of this study provide evidence to support the view that the relative absence of a rich and diverse collection of positive self-schemas contributes to the disordered eating attitudes and behaviors that characterize AN and BN, and raise interesting questions about the appropriate focus of clinical intervention. Cognitive-behavioral therapy for the eating disorders focuses heavily on modifying weight related cognitions and eating behaviors. Women are educated about body weight, nutrition, and consequences of weight-reducing behaviors. They are also taught to monitor themselves through the use of written diaries to identify triggers of the dysfunctional behaviors. Because these interventions are largely focused on body weight and food, they may actually heighten accessibility of the fat schema, the proximal source of the disordered eating attitudes and behaviors. In addition, they fail to take into account the associated, and perhaps more basic impairments in the self-concept – the relative deficit of positive self-schemas and the presence of more negative self-schemas.

The results of this study highlight the importance of the total collection of self-schemas as the context for the development of eating disorder symptoms, and offer preliminary evidence to suggest that interventions designed to promote the development of new positive self-schemas may be an important factor in identifying alternative sources of motivated behaviors and promoting recovery from the eating disorders.