Mental health articles

Principles of Classification abnormal behavior

A classification scheme attempts to divide natural phenomena into mutually exclusive as well as exhaustive subsets. As mentioned previously, the f irst purpose of classifying is to permit accurate communication that will allow developing, explaining, predicting, or controlling the events that constitute a particular phenomenon. To accomplish this goal, certain rules must be followed regarding the development of a conceptual scheme for classification. When psychologists attempt to classify their observations, they must keep in mind the conceptual nature of their efforts—for example, the idea of concept formation, which in reality is simply a model of natural events that simplifies the complex phenomena under study. Because classification systems are first and foremost communication devices, it is vital that the characteristics defining a category be clearly detailed and described.

The scientific or data language is the basic prerequisite for any classification scheme, and complex concepts need to be operationally defined so that measurement of the attributes of a particular category is possible. In the initial stages of developing a classification system, it is probably best to begin with what has been called alpha taxonomy or lower order categories . In essence, the construction of an alpha taxonomy requires a thorough description of the observable attributes of a phenomenon within a specific set. It should be demonstrated that the pattern exists with a cluster of covarying characteristics that are consistent over time and can be observed by various methods. As many subsets as possible should be constructed, so that all aspects of the phenomenon being classified may be included. The reliability of these subsets is greatly affected by the degree of care taken in specifying the definable aspects of the subsets and by how mutually exclusive they are. Another important aspect of developing classif ication schemes is the unit used to assign an event to a given category or subcategory.

Unfortunately, in the area of abnormal behavior, these units have tended to be abstract (e.g., personality structures, traits, or defense mechanisms rather than observations). An example is the first two editions of the American Psychiatric Association’s  Diagnostic and Statistical Manual, which were saturated with psychoanalytic terms and concepts. This problem has since been largely eliminated in subsequent versions. Classification units should be simple, capable of being observed, and amenable to measurement techniques. It will be argued that the appropriate units for classifying abnormal behavior are responses and their param eters. Furthermore, categories and subcategories of an adequate classification scheme of abnormal behavior should consist of response patterns or dimension of response patterns rather than individual persons as units of measurement. The advantage to this approach can be seen, for example, in using the more response-specific phrase person with alcohol dependence rather than labeling the person as alcoholic with its sweeping implications. Ideally, the defining properties of subsets should establish which properties at what values are necessary prerequisites for membership in that particular class. It should be noted that when categories require only some group of attributes in some value, therefore allowing the same attribute to cause a phenomenon to be classified in more than one subset, then the categories are said to be disjunctive. Disjunctive categories can cause confusion about the way an event should be classified, and ultimately, problems of lower reliability, redundant categories, and decreased ability to identify specific etiologies and treatments may result. For example, if an individual exhibits depressed mood and depressed mood is a defining characteristic for a number of categories (e.g., schizoaffective disorder, dysthymic disorder, and major depressive disorder), the person could be given any of these diagnoses. The disjunctive categories in DSM-III led to considerable overlap in symptomatology attempted to reduce the problems inherent in disjunctive categories by deleting some of the overlapping items. For instance, idealization/devaluation was deleted from narcissistic personality disorder and angry outbursts/tantrums deleted from histrionic personality disorder because these were not central features and both items overlapped borderline personality disorder. This was an attempt to pursue more conjunctive categories, in which rules specify the intrinsic attributes and specific levels thereof necessary for a phenomenon to qualify for membership in a particular class or category. However, there are still considerable problems in the DSM-IV because it typically uses such terminology as the patient has to exhibit ‘‘six to ten of the following symptoms,’’ a method that ensures disjunctive categories. Conjunctive categories are one alternative to the problems inherent when disjunctive categories are used. In this case, the absence of one or more attributes or a change in value of the attribute(s) would disqualify the phenomenon from membership. The end result is more homogeneous categories, which allows more precise scientific communication. Another benefit of this type of category is that borderline cases are clearly specified. For these reasons, conjunctive categories should be most useful scientifically. Unfortunately, they can be quite difficult to obtain. It appears that most naturally occurring categories (e.g., furniture, birds, or fruit) tend to be rather fuzzy sets with heterogeneous membership, overlapping boundaries, and many borderline cases.

Natural categories are characterized by a set of correlated features that are imperfectly related to category membership (i.e., the defining features are not singly necessary). Some members of the category possess all of the defining features, but many—if not most— do not. Indeed, in the sciences of biology, zoology, and archaeology, the criteria tend to be polythetic (requiring only a specified number of a larger set of items, as in the DSM-IV) rather than nomothetic (requiring that all items are present). In the field of abnormal behavior, growing awareness of the difficulties in striving for homogeneous syndromes with distinct boundaries led to major revisions in the DSM-III, III-R, and IV. As Widiger, Frances, Spitzer, and Williams noted, the nomothetic nature of many of the DSM-III and III-R’s criteria sets were overly restrictive, whereas the adoption of polythetic criteria for all disorders in DSM-III-R (and IV) improved reliability by not requiring the presence of every item in every disorder for making a diagnosis. Reliability was further improved by deleting some overlapping items, though this can become problematic if it cuts too big a swath through the essential features of related conditions.

Many of the personality disorders, for example, inherently correlate with each other and necessarily share overlapping behaviors or characteristics. Thus, it is realistic to acknowledge this at a basic level. Consider social withdrawal, which is shared by avoidant, schizoid, and schizotypal personality disorders. Deleting this feature from one or two of these categories would improve differentiation but would virtually eliminate clinical usefulness. It can be seen, then, that there are advantages and disadvantages to both disjunctive and conjunctive categories. Although the ideal scientific approach would be conjunctive, the reality of nature as well as practical considerations are moving the field toward polythetic and somewhat disjunctive criteria to emphasize the validity of categories. As noted before, however, efforts are continuing to work toward eliminating nonessential overlap to arrive at more homogeneous categories (as required in a conjunctive approach). Another important principle of classification is the requirement to use a single classifying principle.

Classification schemes dealing with human behavior have the option of sorting behaviors according to observed behavior, the cause of behavior patterns (etiology), development over time (prognosis), or response to treatment. Although ideally there should be classification schemes for all of these parameters of human behavior, initially they should be independent to avoid conceptual contamination. The descriptive classification of behavior, for example, should not include the causes of a particular behavior pattern as part of its defining factors. In the early stages of a psychological science, behavior, etiology, and response to treatment should be classified independently, and a specific type of behavior should be related to a specific etiology on the basis of empirical facts or research. This results in a higher order of classification, termed beta taxonomy by Bruner et al., and through this approach the relational categories can then be empirically related to a particular treatment approach. A good example of this type of development is the identification of syphilis, which was followed by the demonstration that it is caused by spirochetes, and then by the development of penicillin to treat the disorder. If an empirical relationship among behavior, etiology, prognosis, and response to treatment can be established, then a more scientific classification system results. On the other hand, if this approach is not adopted, and behavior, as well as the presumed causes of the behavior, is used in classification, then criterion contamination results. One good way to reduce criterion contamination is by using a multiaxial system that classifies behavior on one axis, etiological factors on a second axis, and so forth. In summary, several principles of classification are important, including the use of adequate terminology or data language, simple classification units, conjunctive categories (as far as possible), and a single classification principle. Inadequate terminology, overly disjunctive categories, and preconceived notions about etiology that influence the development of a classification system will result in difficulties in the system’s reliability and validity.

Tags: abnormal, abnormal behavior, behavior, classification