Elsevier

Clinical Psychology Review

Volume 23, Issue 8, January 2004, Pages 1001-1022
Clinical Psychology Review

Attentional biases in eating disorders: A meta-analytic review of Stroop performance

https://doi.org/10.1016/j.cpr.2003.09.004Get rights and content

Abstract

The Stroop task has been adapted from cognitive psychology to be able to examine attentional biases in various forms of psychopathology, including the eating disorders. This paper reviews the research on the Stroop task in the eating disorders research area in both descriptive and meta-analytic fashions. Twenty-eight empirical studies are identified, which predominantly examine food and body/weight stimuli in bulimic, anorexic, or dieting/food-restricted samples. It is concluded that there is evidence of an attentional bias in bulimia for a range of stimuli but that the effect seems to be limited to body/weight stimuli in anorexia. The evidence to date is that there is no attentional bias in dieting samples. Limitations of the methodology employed in the extant literature include small sample sizes, unstandardized Stroop methodology, restricted gender, and a general lack of consideration of individual differences variables. Recommendations for future research are provided.

Introduction

The models that have developed in the area of the eating disorders Fairburn & Brownell, 2002, Garner & Bemis, 1985, Vitousek & Orimoto, 1993 emphasize multiple possible causal factors in these problems. Among these models, one of the notable developments in psychopathology research has been the growth of investigations and models that emphasize cognitive processes and content (cf. Dobson & Kendall, 1993). In part spurred on by the general “cognitive revolution” in psychology and in part by the success of cognitive-behavioral therapies, this emphasis has yielded several varied and rich technologies to examine psychological processes.

Among the range of cognitive factors that have been implicated in the eating disorders are such issues as attitudes and beliefs about ideal body weight, body dissatisfaction, and body image and perception; perfectionism have received considerable attention. For example, in a recent examination of the prediction of binge eating and purging (Byrne & McLean, 2002), it was found that overconcern with body weight and shape, coupled with the adoption of purgative behaviors, was predictive of binge eating and purging. Similarly, a recent review of the issue of body size dissatisfaction in anorexia nervosa (Skrzypek, Weheimer, & Remschmidt, 2001) found that although body size estimation is not impaired in this condition, it is the discrepancy between perceived body size and ideal size that is significantly associated with anorexia nervosa. Findings such as these highlight the role of negative attitudes and beliefs about food and body shape in the eating disorders and suggest the possibility of disordered information processing in these conditions. For example, individuals who have eating disorders have undue concerns about food or body image; it may be that these concerns could be demonstrated through a variety of methods that relate to different cognitive processes. One of the cognitive processes associated with eating disorders that has been the focus of considerable research is that of selective attention. Consistent with the idea that individuals with eating disorders have negative beliefs about food and body shape, it has been suggested that these individuals are also more attentive and responsive to these relevant stimuli than non-eating-disordered individuals. A body of research using a modification of the Stroop task has evolved in which selective attentional processes have been examined. In this paper, we review the use of the Stroop task in examining eating-disordered attentional processes and examine in meta-analytic format the existing data. Conceptual and methodological issues in the extant literature and directions for future investigations are provided.

The original or Classic Stroop task (Stroop, 1935) was developed as a means to study basic human attentional and informational processes. This task consisted of the presentation of colors printed on either neutral words or incongruent color words. Participants first named the color name of the stimuli and then the color of the stimuli. This method was later revised to include color words (e.g., red and blue) that were printed in either the corresponding color (congruent condition) or other competing colors (incongruent condition). As these tasks are relatively easy, they form a basis for comparing the interference created by naming the color of the stimulus when the actual color is competing (e.g., the word “blue” written in green). Scores derived from the Stroop task consist of either the latency (typically in ms) to name each stimulus or the “interference effect,” which is calculated as the average time taken to name the competing cards divided by the sum of the average time for the two other sets of stimuli. In some studies, the number of correct color names is also recorded.

When the Stroop task was first explored (Stroop, 1935), the interference effect was discussed as an example of cognitive competition and how such competition retards the production of correct color names. Replications of the Stroop interference effect have shown that it is reliable Dyer, 1973, Franzen et al., 1987, MacLeod, 1991. Both Stroop (1935) and MacLeod (1986, as cited in MacLeod, 1991) have reported that in normal controls the latency to name color-only stimuli is about 40% of that for naming incompatible color-word stimuli.

MacLeod's (1991) extensive review of the Stroop task identified “more than 700 Stroop-related articles in the literature” (p. 163), many of which focus on the perceptual, cognitive processing, and potential neurological processes underlying the Stroop phenomenon. According to MacLeod, a set of 18 reliable findings emerges in this literature, which relate to the conditions that optimize or suppress interference effects, the effects of cues on Stroop performance, and the effect of different localizations of stimuli, practice effects, and other performance aspects. Notably for the current review, MacLeod's authoritative review does not examine the effect of individual differences on Stroop interference, except to examine gender and age.

Despite the much research conducted with the Classic Stroop task on normal participants, it remains unclear whether the Classic Stroop effect is created by difficulties associated with the relative speed of processing color versus language (Klein, 1964), the automaticity of language Logan, 1980, Posner & Snyder, 1975, the perceptual encoding of the proper stimulus attribute (i.e., is a result of perceptual interference; Dyer, 1973, but see MacLeod, 1991 for evidence refuting this hypothesis), or the interference related to the differential strength of the competing pathways being processed (see Cohen et al., 1990, Cohen et al., 1992, Logan, 1980, Logan, 1985, MacLeod, 1991, Williams et al., 1996 for discussions of parallel distributed processing). Debate on the issue of which model best explains the Stroop interference effect continues, and “the Stroop effect will continue to be a challenging phenomenon for cognitive psychologists to explain for many years to come” (MacLeod, 1991, p. 193).

Notwithstanding the controversy in cognitive psychology about the mechanism that accounts for the Stroop interference effect, the Stroop task has been used in psychopathology research for a considerable period of time. It has been suggested that the Stroop task is a valuable tool for examining cognitive processes in psychopathology Segal, 1988, Williams et al., 1996 and that it affords an opportunity to compare disordered samples with normal controls in a task with relatively tight experimental control. Stroop research can be in such diverse areas as anxiety disorders, depression, schizophrenia, eating disorders, substance abuse, and studies that relate personality variables to cognitive interference. Recent years have witnessed a dramatic increase in the amount of Stroop research conducted in psychopathology (Williams et al., 1996).

In most of the psychopathology Stroop research, the focus is not often on general attentional interference but is rather on the attentional bias that disordered participants exhibit relative to comparable control stimuli. As a result, the Stroop task has often been modified to use both classic and “emotional” or disorder-appropriate stimuli (e.g., anxiety words for studies of anxiety-disordered samples and food words for studies with eating-disordered samples). In effect, the model that has been tested is that appropriate emotional stimuli are more salient to individuals with a particular disorder and that such stimuli will be selectively attended to (i.e., processed more efficiently) than for either individuals without that condition or mismatched stimuli. The large number of studies that has been conducted using this paradigm affords a unique opportunity to make comparisons about the information-processing mechanisms among diverse clinical conditions. For example, Williams et al. (1996) reviewed the Stroop task in depression and anxiety and emphasized the mechanisms (cognitive and artifactual) underlying its interference effect. These authors demonstrated that attentional biases are found in depression and anxiety and argued that the connectionist (parallel distributed processing) model (Cohen et al., 1990) is useful for understanding these biases.

The purpose of this study was to review in both descriptive and meta-analytic formats the Stroop task in the eating disorders. There has been a recent “explosion” of research on the Stroop task in eating-disordered samples, as fully 23 of 26 studies discovered in the literature search for this review were published in or since the 1990s. Much of this work has been precipitated by cognitive formulations of eating disorders Fairburn & Garner, 1988, Polivy & Herman, 1987 particularly the awareness of the sensitivity of patients with eating disorders to food as well as perceptions of body size and weight. The use of the cognitive measures has been increased because of concern that simple questionnaires do not well measure attitudes associated with eating disorders (Vitousek & Hollon, 1990); the use of the Stroop task in particular has been spurred by the development of sets of adjectives specifically designed for this population. Thus, the development of both a “Food Stroop” and a “Body Stroop” (sets of words with food or body shape terms, respectively; Ben-Tovim et al., 1989, Channon et al., 1988) has enabled researchers in this area to focus on specific matched stimuli rather than the Classic Stroop words. Rather than exploring the mechanisms thought to produce the Stroop interference effect (see Williams et al., 1996), this review assesses the Stroop findings and methodology to further our understanding of psychopathology and to enhance future clinical research. In the literature review that follows, a distinction will be made between “Classic Stroop” and other “Emotional Stroop” methods, where the latter manipulate the content of the stimuli employed. Comparisons will be made between eating-disordered and other groups as well as between eating-related stimuli and other comparison stimuli, where such data exist.

The research in eating disorders has focused on three different problem areas: bulimia, anorexia nervosa, and dieting/restrained eaters. There appears to be a uniformity hypothesis in the literature in that the rationale provided in different studies for the use of the Stroop task with different samples is analogous. However, as this distinction is consistently made in the empirical literature, we review the literature on dieting/food-restricted samples separately from those individuals who meet formal diagnostic criteria for an eating disorder and further distinguish eating-disordered samples into those who suffer from either anorexia nervosa or bulimia (American Psychiatric Association, 1994). Explicit tests of the uniformity hypothesis exist and are discussed below.

Section snippets

Method

The strategy employed for locating published research was to use the PsycINFO and Index Medicus indices for research articles, chapters, and books for the period reviewed. Keywords included any combination of “Stroop” and “eating disorder,” “bulimia,” “anorexia,” “diet,” or “restricted.” Further, relevant published citations from all obtained research articles were pursued. We were able to obtain a copy of every citation that appeared to include data on the Stroop task in eating disorder. Thus,

Qualitative comments on the literature

As can be observed in Table 1, the predominant research method in this area has been the comparison of eating-disordered samples with one or more types of control subjects. The general finding in this literature is that patients with bulimia demonstrate an interference effect with the Food Stroop task relative to normal controls (Ben-Tovim & Walker, 1991, Ben-Tovim et al., 1989, Black et al., 1997, Cooper et al., 1992, Fairburn et al., 1991, Jones-Chesters et al., 1998, Perpiñá et al., 1993;

Discussion

The results from the eating disorders research suggest that whereas bulimic participants consistently show attentional biases on the Stroop task across a range of stimuli, the results are limited to the area of body/weight stimuli for anorexic subjects and are trivial to modest for dieting/food-restricted subjects. These results should provide cause for concern about using the Stroop task as a measure of attentional bias in nonclinical samples. They further suggest, however, that there may be

Conclusion

Before the adoption of methodologies derived from experimental cognitive psychology, clinical investigators were unable to directly measure the cognitive processes or structures involved in maladaptive functioning. Instead, indirect inferences were often drawn, from self-report data, about the logical underlying cognitive operations associated with psychopathology (Ingram & Kendall, 1986). As Segal (1988) aptly stated, “the strategy of relying on [patient] self-reports to validate a construct

Acknowledgements

The support of the Alberta Heritage Foundation for Medical Research (Dr. Dobson) and the Ontario Mental Health Foundation (Dr. Dozois) are noted.

References (78)

  • D.I. Ben-Tovim et al.

    An adaptation of the Stroop test for measuring shape and food concerns in eating disorders: A quantitative measure of psychopathology?

    International Journal of Eating Disorders

    (1989)
  • C.M.D. Black et al.

    Selective processing of eating disorder relevant stimuli: Does the Stroop test provide an objective measure of bulimia nervosa?

    International Journal of Eating Disorders

    (1997)
  • G.H. Bower

    Mood and memory

    American Psychologist

    (1981)
  • B.P. Bradley et al.

    Covert and overt orienting of attention to emotional faces in anxiety

    Cognition and Emotion

    (2000)
  • S.M. Byrne et al.

    The cognitive-behavioral model of bulimia nervosa: A direct evaluation

    International Journal of Eating Disorders

    (2002)
  • F.A. Carter et al.

    Changes on the Stroop test following treatment: Relation to word type, treatment condition, and treatment outcome among women with bulimia nervosa

    International Journal of Eating Disorders

    (2000)
  • K.L. Cassiday et al.

    Cognitive processing of trauma cues in rape victims with post-traumatic stress disorder

    Cognitive Therapy and Research

    (1992)
  • S. Channon et al.

    The effect of short-term fasting on processing of food cues in normal subjects

    International Journal of Eating Disorders

    (1990)
  • S. Channon et al.

    Selective processing of food words in anorexia nervosa

    British Journal of Clinical Psychology

    (1988)
  • J.D. Cohen

    Statistical power analysis for the behavioral sciences

    (1977)
  • J.D. Cohen et al.

    On the control of automatic processes: A parallel distributed processing account of the Stroop effect

    Psychological Bulletin

    (1990)
  • J.D. Cohen et al.

    A parallel distributed processing approach to automaticity

    American Journal of Psychology

    (1992)
  • M.J. Cooper et al.

    Selective processing of eating-, shape-, and weight-related words in persons with bulimia nervosa

    Journal of Abnormal Psychology

    (1992)
  • M.J. Cooper et al.

    Selective processing of eating, weight and shape related words in patients with eating disorders and dieters

    British Journal of Clinical Psychology

    (1992)
  • M.J. Cooper et al.

    Demographic and clinical correlates of selective information processing in patients with bulimia nervosa

    International Journal of Eating Disorders

    (1993)
  • M.J. Cooper et al.

    Changes in selective information processing with three psychological treatments for bulimia nervosa

    British Journal of Clinical Psychology

    (1994)
  • M. Cooper et al.

    Selective processing of three types of stimuli in eating disorders

    British Journal of Clinical Psychology

    (1997)
  • K.S. Dobson et al.

    Specificity and stability of self-referent encoding in clinical depression

    Journal of Abnormal Psychology

    (1987)
  • D.J.A. Dozois et al.

    Information processing and cognitive organization in unipolar depression: Specificity and comorbidity issues

    Journal of Abnormal Psychology

    (2001)
  • F.N. Dyer

    The Stroop phenomenon and its use in the study of perceptual, cognitive, and response processes

    Memory and Cognition

    (1973)
  • C.G. Fairburn et al.

    Selective information processing in bulimia nervosa

    International Journal of Eating Disorders

    (1991)
  • C.G. Fairburn et al.

    Diagnostic criteria for anorexia nervosa and bulimia nervosa: The importance of attitudes to shape and weight

  • E.B. Foa et al.

    Processing of threat-related information in rape victims

    Journal of Abnormal Psychology

    (1991)
  • D.M. Garner et al.

    Cognitive therapy for anorexia nervosa

  • G.V. Glass

    Primary, secondary, and meta-analysis of research

    Educational Researcher

    (1976)
  • S.G. Gowers et al.

    Development of weight and shape concerns in the aetiology of eating disorders

    British Journal of Psychiatry

    (2001)
  • M. Green et al.

    Impaired color naming of body shape-related words in anorexia nervosa: Affective valence or associative priming?

    Cognitive Therapy and Research

    (1999)
  • Cited by (220)

    View all citing articles on Scopus
    View full text