The Influence of Context on Categorization Decisions for Mental Health Disorders Jessecae K. Marsh (jessecae.marsh@TTU.edu) Department of Psychology, Texas Tech University, MS 42051 Lubbock, TX 79413 USA Andres De Los Reyes (adelosreyes@psyc.umd.edu) Department of Psychology, University of Maryland at College Park, 1147 Biology/Psychology Building College Park, MD 20742 USA Abstract Mental health clinicians engage in an important form of real- world categorization as they diagnose their patients with mental disorder diagnoses. How are clinicians affected by the context within which diagnostic criteria of a patient present when making diagnostic evaluations? The classification system clinicians are instructed to use is structured around a statistical approach to assessing diagnosis and does allow for the interpretation of criterial features through influences like context. The following experiment tests whether clinicians are affected by the context within which non-diagnostic information about a patient is presented. We tested clinician’s diagnostic judgments for symptoms of Conduct Disorder that were presented either in a context that should be perceived as being associated with Conduct Disorder or in a context that should not be perceived as being associated with Conduct Disorder. We found that clinicians were influenced by context, but in surprising ways. Clinicians lowered their diagnostic judgments for symptoms presented in a low associative context but did not change their estimates for high associative contexts as compared to baseline. The effect of context was also found to vary over the criterial symptoms that were presented, and this variation was associated with clinicians’ idiosyncratic ratings of the criterial symptoms. These results have interesting implications for how clinicians view their patients and for how context affects categorization more generally. Keywords: Categorization; clinical reasoning; theory-based reasoning. Introduction Mental health disorders provide an interesting domain within which to study real-world categorization. The Diagnostic and Statistical Manual of Mental Disorders (DSM) presents a guideline for mental health clinicians to use in the categorization of patients into mental disorder categories (American Psychiatric Association [APA], 2000). Most mental disorders included are described as lists of criterial symptoms, with membership in a given category being gained by displaying a certain number of the described criterial features. For example, in order to qualify for a diagnosis of the childhood disorder Conduct Disorder a patient must meet a series of required criteria (e.g., minimum amount of time showing the problematic behaviors, marked impairment in life functioning) and then possess any 3 of a possible 15 other criteria that typify clinically impairing levels of aggressive behavior and a disregard for rules or social norms. The exact symptoms or combination of symptoms generally do not figure into categorization as long as the required number of symptoms is met. In order to collect information about the criterial features that are present in a given patient, mental health clinicians must go through a difficult process of gathering and synthesizing different types of information. Patients do not necessarily provide their diagnostic symptoms in a clear-cut and easy to parse fashion. Instead, such criterial features are embedded within a set of extraneous, biographical, non- diagnostic information. Although the DSM acknowledges that the context within which these behaviors occur should be taken into account when making diagnostic decisions (e.g., a child that frequently gets into fights at school for the sake of self-defense likely does not meet criteria for Conduct Disorder), such non-diagnostic contextual information that does not take the form of actual diagnostic criteria does not play a formal, codified role in diagnostic classification as delineated in the DSM. The question becomes, do clinicians use the context within which diagnostic criteria present to make their category decisions? Evidence from the cognitive psychology field suggests that people’s categorization decisions can be affected by the context in which the features are displayed. For example, Medin, Goldstone, and Gentner (1993) found that the context within which an ambiguous stimulus was presented affected the interpretation of its ambiguous features, and consequently its categorization. Similarly, Ahn, Novick, and Kim (2003) found that providing clinicians with an explanatory context for a person’s mental disorder symptoms results in the person being categorized as more ‘normal’. Likewise, the stereotype literature has shown that an ambiguous behavior will be categorized differently depending on contextual information such as the ethnic background of the person engaging in the behavior (Gawronski, Geschke, & Banse, 2003). Generally speaking, this evidence suggests that people are influenced by the context within which categorization information is presented. The actual structure of DSM categories may further influence clinicians to rely on contextual information when evaluating patients. The statistical structure used in the 1953