We were glad to hear from Dr. Cook whose work we are well aware of, having used his measure several years ago, along with a measure of guilt developed by Karen Kugler and Warren Jones, in a study of eating disordered women. In this study, which was conducted by a student working with us, Debbie Grebel, we found that eating disordered women suffered from both shame and guilt. We consider both shame and guilt to be primary contributors to psychopathology, because pathogenic beliefs can lead to guilt and to shame, either alone or more usually, in combination.
Dr. Cook seems to be presenting the view that shame is the primary self-conscious emotion related to psychopathology, that guilt is a variant of shame and secondary in all cases. He claims that shame and not guilt is a biologically based emotion (by which we assume he means physiological), and that guilt is merely "cognitive" (as well as secondary). We would question his assumption; in our view, both guilt and shame have biological as well as cognitive components (and in fact, Dr. Cook's description of his measure of internalized shame makes the point that shame becomes internalized due to its cognitive component). Antecedents to both guilt and shame appear early in life. Focusing on guilt, as Dr. Cook has already specified the early presence of shame, it has been reported that babies as early as six months exhibit a distinct distress response at witnessing the distress of another, and infants make efforts to engage and perhaps cheer up depressed mothers. We do not see how this could reduce to shame. Furthermore animal studies have suggested antecedents to survivor outdoing guilt may occur in higher apes. Specifically, chimpanzees share food in the wild. And in the laboratory it was found that unrelated chimpanzees, when given food that is not also given to the chimp caged next to him or her, will respond in some cases, by sharing their food. It was also noted that sometimes this occurs after a period in which the deprived chimpanzee begs their neighbor with food, suggesting that some emotion related to guilt is induced in the fed animal. This is another case that could hardly be explained by shame. We would suggest that guilt is an emotion selected by evolution, that serves to maintain group organization and cohesion, as is shame. This has recently been discussed by Paul Gilbert, as well as by ourselves. Both guilt and shame lead to psychopathology when they are associated with pathogenic beliefs or pathogenic expectations.
Dr. Cook and other theorists and researchers who emphasize shame and attempt to minimize or even exclude guilt as a potentially problematic and primary emotion are perhaps being constrained by an individualistic and competitive perspective on human nature, as found in western social thought and well exemplified by 19th century thinkers such as Freud. This world view has led to what may be an excessive emphasis on the concept of the self. If one believes that the "self" is the main unit of study and of the social group, then emotions focused on the self --or threats to the self-- become primary. However if one considers that groups may be an equally significant unit organization and of selection, then an emotion related to the group, such as guilt, may be primary. Weiss's theory of the mind, of psychopathology, and of psychotherapy incorporates an understanding of man as a group-based social animal, with empathy and the need to belong as a primary motivation. The need to belong as a fundamental motivation has also been discussed in detail by Roy Baumeister and Mark Leary.
In an effort to investigate the significance of guilt and shame and their relationship to psychopathology, we began our series of empirical studies using a measure developed by June Price Tangney, which included a subscale of shame and of guilt. We found that her measure of guilt was related to what we consider to be adaptive guilt, whereas the Kugler and Jone's measure mentioned above seemed to tap into the kinds of guilt related to psychopathology. Neither of these measures however, examined the kind of guilt that is emphasized in Weiss's theory , which is why we decided to develop our new measure.
Without going through all the details (though we'd be glad to send anyone interested our articles which do that), we set out to find out if guilt, as defined as the fear of harming others, was associated with psychopathology as Weiss's theory suggests. Challenged by Tangney's results using her measure, that shame and not guilt was the problem emotion, we conducted studies using our measure and hers (as well as studies with Jones' measure), along with measures of psychological distress, symptoms and depression. In one study we replicated Tangney's method, including her method of data analysis, which included the use of part correlations to examine the role of "guilt-free shame", and "shame-free guilt". In this investigation we found that when controlling for shame, survivor guilt continued to have a significant relationship with psychological symptoms, however when controlling for survivor guilt, shame lost its significant correlation with symptoms. This did not lead us to consider that guilt and not shame are important in the development and maintenance of psychopathology; however it did lead us to the conclusion that the assertions of Tangney, like those of Dr. Cook in his note, could not be substantiated empirically, when using a measure of maladaptive interpersonal guilt.
Dr. Cook mentioned the issue of the "competence and credibility of the psychometric research carried out by Weiss and his associates" (by the way Dr. Cook, we prefer to be called by our names; we conducted this research and are first and second authors on all reports of it). While we have not checked out Dr. Cook's measure in our data sets, for its psychometric properties, we can assure you that we have certainly examined our own. Psychometrically, our subscales look good in terms of internal consistencies of items. Alpha coefficients are reported in all write ups of this work, and one article describes the development of the measure (in press, Journal of Clinical Psychology). The IGQ was developed using the usual procedures based on classical test theory (creating test specifications, experts generating a large pool of face valid items, administering the items to a sample, selecting items based on item-total correlations, calculating alphas, re administering the shortened version to new samples, etc). Recently we used item response scaling procedures (fitting item data to a Rasch polytomous item response model developed by David Andrich; often called "the Rating Scale Model") to assess the IGQ subscales more thoroughly at the item level. With one exception, items for each subscale fit the Rasch model, indicating that items are defining unidimensional, interval level variables. The item reliabilities for the four subscales were all greater than .90; the case reliabilities (analogous to alpha coefficients, but corrected for error) were all greater than .75. These results are being prepared for publication.
Dr. Cook's suggestion that guilt and shame are significantly correlated is correct, however the coefficients are not so high as to suggest they are the same construct. In several studies, guilt (not only survivor guilt, but guilt as assessed by Kugler and Jones' Guilt Inventory) has been moderately correlated with shame. But it will come as no surprise to Dr. Cook that shame has also been moderately correlated with guilt. Confounding of constructs is a difficult problem to handle, especially in self-report inventories. This is true of guilt measures, but also of shame measures. What's the difference between proneness to shame and low self-esteem, Beck's negative self-statements, or any "me bad" construct? As we reported above we used part correlations following Tangney's methodology to attempt to separate some of these constructs, and in most cases, found survivor (outdoing) guilt but not shame to hold up.
Dr. Cook believes that guilt is a defense against shame; we believe that in many cases, shame is a defense against unconscious guilt. Survivor guilt and its associated pathogenic beliefs are often unconscious when patients initially present their problems. In fact many patients frequently first describe themselves as inadequate or deficient, a symptom which equates with shame and self-hate, and which after a period of therapy comes to be understood as a way of reducing survivor guilt. That is patients may use persistently negative self-talk, self put-downs, and feelings of shame to keep themselves from feeling that they are better off than family members, siblings, or other loved ones. As patients come to be more aware of their survivor guilt and the pathogenic beliefs that cause it, they are able to counter these beliefs in their work with their therapists, in self-help groups, or other corrective relationships or experiences. As they modify these beliefs, their survivor guilt is reduced, and with it their symptoms. Weiss has specifically suggested that patients who first present with symptoms that appear to be most related to shame, are often struggling unconsciously with survivor guilt, and that the feeling of shamefulness may have the effect of reducing or concealing this guilt. He describes shame in some cases as an indicator of powerful and unconscious survivor guilt.
To argue about the relative importance of guilt and shame may reflect us psychologists fighting over turf. We would suggest that both guilt and shame are painful emotions that serve as important organizers of social life, that often occur together, and when linked to pathogenic beliefs become excessive, crippling, and potentially pathological.