Using research in social cognition, particularly attribution theory, this article reviews some heuristics (and inferential biases deriving from them) that are used in establishing a client database, in relating putative causes to manifest client dysfunction, and in formulating problems. Common errors from uncritical use of these heuristics in clinical reasoning are discussed, and their sources are categorized according to conventions that developed in social psychology during the past 25 years. The diagnostic implications of the availability heuristic's pervasive and often uncorrected intrusion into therapeutic judgment are emphasized. Other variables are discussed, such as theoretical set, overgeneralization, the anchoring effect, therapist mood, and levels of inferential content, that can seriously affect diagnosis. Suggestions are offered for reducing the influence of schematic biases and the errors resulting from them. Psychotherapy and the Cognitive RevolutionPsychotherapy and counseling have been significantly altered in the past 3 decades by the sweep of new developments in the cognitive sciences, particularly in the domain of social cognition. The passing of the simple linear models for explaining how people learn, think, process sensory inputs, represent the "world as it is," remember and forget, alter schemata, suppress information, fabricate "excuses," recognize patterns, and engage in other aspects of human mentation has altered the way we as clinicians view our clients and attempt to help them. (See Mahoney, 1991, pp. 67-94 for a short, helpful review of how our models of the human mind have changed.)Among the most influential domains has been that of attribution theory, driven by the creative theorizing and research of

Research in several domains has revealed that when individuals are asked to estimate the probability that their judgments are correct, they reveal an overconfidence effect. Judgments produced in decision environments such as psychodiyrosis, which are by nature complex and ambiguous, appear to be most vulnerable to overconfidence. By implication, this phenomenon threatens the validity of clinical judgment and subjects clients to risks of flawed diagnoses and unsuitable treatments. The authors propose that effective remedies to overconfidence begin in training programs that lead students through problem-solving experiences that can invdidate facile, premature, and dubious dlyrostic judgments. Syndromes det3n.d by the DSM-IV are permeable and fuzzy sets, and students must be trained to assign membership to these sots with caution. Key wods: overconiidence, clinical judgment, debiasing, fuzzy sets. Win Psycho1 Sci Pmc 4d35-345, 1997JWhen psychologists make judgments about clients based on clinical databases that are necessanly incomplete and often ambiguous, they do so, knowingly or otherwise, under condltions of uncertainty. Research has shown that inferences generated under such conditions are ofien expressed with unwarranted confidence. In that perspective, this article presents a selective and brief review of the literature on overconfidence, hghhghting important empirical findlngs that have relevance to this problem in clinical psychology and in other judgment contexts.

Therapists have numerous psychodiagnostic instruments at their disposal for use in assessing patients, although no one can achieve competency in more than a small minority of them. The issues addressed here bear on the penchant of some otherwise competent professional psychologists to use such specialized instruments (a) for which they have not received adequate formal training, (b) without compliance with canons of sound interpretation presented in the research literature as well as the relevant manual, and (c) for which sufficient evidence of validity is lacking. In this descriptive study, use of the Draw-A-Person Test (K. Machover, 1949) by a sample (N = 36) of clinicians provides grounds for these concerns. Some reasons that might dispose psychologists to such practice are proffered, and some implications for clinical praxis as well as for training programs are examined.

The properties of problem representation among experts working in well-structured problem spaces are distinguished from those working in the poorly structured and more complex domain of psychotherapy. The implications of forward and backward causal reasoning, the absence of a consensually sanctioned, canonical domain knowledge, and the reigning confusion about what constitutes a relevant clinical database in a psychotherapeutic relationship are drawn out. A number of normative characteristics of expert performance are examined in view of the manner in which they constitute liabilities for those working in fuzzy problem spaces."An expert," said Frank Lloyd Wright, "is one who does not have to think. He knows." Herein lies a problem. One may suspect that Wright was not referring solely to architects. Whether or not psychotherapists and counselors, most of whom in the course of their careers become highly experienced if not expert, escape the implications of that maxim remains to be seen. In any event, evidence exists (Chi, Glaser & Farr, 1988) that the characteristics of expert systems are similar across disciplines. That means that expertise in psychotherapy may share, in varying degrees, properties that inhere in other skill domains.

When psychologists are asked to estimate the probability that their clinical judgments are correct, they often reveal an overconfidence effect. In an effort to identify sources of unwarranted confidence in clinical judgment, this study examined the relationship between four different inferential biases (dispositionism, confirmationism, datasearch truncation, and narrow problem construal) and diagnostic confidence in the context of a psychological assessment task. Thirtysix clinicians were individually presented a written client case-file to read and clinically interpret aloud. Analyses of participants' verbal protocols revealed that dispositionism alone accounted for a significant proportion of the variance in psychodiagnostic confidence scores. These results indicate that dispositionally-driven assessments tend to be expressed with the highest levels of confidence. The roles that professional psychology and psychology in general play in propagating this bias are considered.