The utilization of the Objective Structured Clinical Examination (OSCE) as a tool for assessing the ability of students entering their senior year to identify psychiatric problems in patients in a medical setting has had mixed results (1, 2). In one instance, when students were presented a straight-forward psychiatric case in a mix of standardized patients with medical-surgical problems, they performed poorly in being able to ask questions that would help them clarify, explore, and identify the psychiatric problem, so much so that the case had to be dropped in the overall assessment of performance (1). In another analysis, a low percentage of students were able to complete a suicide risk assessment of standardized patients when asked to consider psychiatric diagnoses in a primary care setting (2). As a result, concern has been voiced whether it is reasonable to expect students to demonstrate integrative clinical skills, particularly by the end of their 3rd year during which clinical training is imparted, in most institutions, by dividing up the clerkships into discrete specialty blocks (A. Brodkey, personal communication, 2005). This argument raises two questions. Is psychiatric history-taking skill identical to medical history-taking skill? If not, can students reliably and consistently generalize psychiatric skills to other clerkships and settings?
Is psychiatric history-taking skill identical to medical history-taking skill?
In an attempt to answer the question, it is useful to review the nature of a “medical” problem and a “psychiatric” problem, the nature of learning, the nature of the educative process, and its application to medical education.
Patient A, a 40-year-old married woman, seeks an appointment with her primary care physician for vague abdominal discomfort that yields no physical examination findings or laboratory and imaging corroboration. The symptoms remit, only to return some months later and prompt a revisit to the clinician. The physician elicits history of an abusive relationship of many years with her husband who has been more irritable and abusive owing to his precarious employment-related stress. One might consider a number of explanations for this patient’s presenting problem. She may have a conversion disorder, might be depressed or resigned to a deplorable marital relationship, or her visit might be the “cry for help” of a fearful spouse.
Patient B, a 48-year-old man, comes to the emergency room with chest pain and a history of serious cocaine abuse and homelessness. The physicians suspect that he is malingering in order to seek a brief hospitalization. However, upon physical examination, he has tachycardia and is hypertensive and an ECG reveals an acute inferior wall myocardial infarction. His cardiac enzymes are elevated. This patient, despite an initial suspicion of malingering, is admitted for the management of an acute myocardial infarction.
Problems lie on a continuum of definition. A well-defined problem has only one solution and a guaranteed method of solving it (e.g., a quadratic equation). An ill-defined problem has more than one acceptable solution and no universally agreed upon strategy for reaching it (e.g., global warming) (3, 4). According to this view, the first problem described here may lie closer to the ill-defined end of this spectrum in comparison with the second. It may, therefore, be argued that “psychiatric” problems are less well defined than “medical” problems because the objective measurement of the former is hampered by a lack of easily accessible tools such as an ECG, and their concrete verification is limited by the lack of clinically viable imaging or laboratory studies.
Interest in problem-solving developed early in the 20th century among educators and psychologists. In 1911, Thorndike (5) experimentally established that cats learned to escape from a wooden crate by pressing a lever largely in a trial-and-error manner. He observed that problem-solving in cats was not intentional and argued that human learning occurred similarly and that success occurred incrementally as a function of trial-and-error attempts. John Dewey (6), in contrast, believed that learning to solve problems was a conscious and deliberate process that occurred sequentially and could be taught in a step-by-step manner. The Gestalt psychologists provided a third model by conducting studies on chimps (7). In one famous experiment, a chimp was put in a cage in which bananas were hung from the ceiling just beyond his reach. Various wooden crates were also randomly placed in the cage. After many unsuccessful attempts and some deliberation, the chimp managed to stack the crates and reach the bananas. It was thus proposed that problem-solving was a result of the development of “insight” in the chimp. Duncker (8) introduced the concept of “functional fixedness” in 1945 by demonstrating that preexisting knowledge may actually inhibit such solutions to a problem.
Nature of the Educative Process
The three successive models of educational psychology—the associative model of Hull (9) and Spence (10), the “learning-by-doing” constructivist model (including, perhaps, the more radical, directed behaviorism model of Skinner ) and the cognitive psychology model (12)—may well loosely trace their origins to the three models of problem-solving proposed by Thorndike, Dewey and the Gestalt psychologists, respectively. It is clear, however, that the cognitive model now serves as the dominant theme for educational psychology. Such cognitive models of education emphasize that learning is constructive, not repetitive; is driven by motivation and beliefs that must be integrated into cognition; requires mental frameworks or schema to organize memory and guide thought; entails extended practice; requires the capacity for critical thinking which, in turn, is based upon the development of self-knowledge and self-regulation; calls for social instruction as being a fundamental and assumes that knowledge, strategies and expertise are contextual, not mechanistic (12).
Application to Medical Education
So if medical problems are construed as somewhat better-defined than psychiatric problems, one must look at how the dominant cognitive model is applied to solving a “medical” problem and how this application differs in the case of a “psychiatric” problem. The three key steps in solving a medical problem consist of identification, understanding, and solution (or treatment). Identification of a problem is the first and the most difficult step. It requires background knowledge, creativity, persistence, and willingness to ponder without committing to a premature conclusion. In the case of a psychiatric problem, this first step is made more difficult by the limitations of measurement and verification described above. Without the tools necessary for identifying a psychiatric problem with the degree of confidence comparable to his or her medical colleagues, the psychiatrist may only try to realize the crucial first step by weighing evidence, analyzing competing views, gathering new information, and with some luck, reidentifying or “finding” the problem (13).
In the second step, the physician “understands” the problem by an external (or, graphic) representation of the problem with the help of laboratory and imaging techniques. For psychiatric problem-solving, the only tool available to the psychiatrist is an internal or reflective one. Consequently, the task of breaking up the psychiatric problem into its components and distinguishing relevant from irrelevant constraints is harder and the depth of the “problem space” required to solve it is greater (3). Thus psychiatric problem-solving appears to be different from medical problem-solving in that it requires a more complex approach to interviewing and thinking about the problems that a patient brings to the doctor.
Can students reliably and consistently generalize psychiatric skills, which appear to be somewhat different from general medical skills, to other clerkships and settings?
The ability to utilize distinct domain skills interchangeably in a variety of settings requires a critical mass of “shared general knowledge” (history and physical examination skills, objective measurement, concrete data from laboratory tests) between domains being considered. If the domains have differences in shared general knowledge (mental status examination versus physical examination skills, “finding” rather than identifying a problem, reflection instead of representation), then “problem-solving transfer” from one distinct domain to another may not occur because expertise is welded to specific domains according to Detterman (14). In his elegant experiment demonstrating the concept of “functional fixedness,” which may provide yet another explanation for this assertion, Duncker (8) provided subjects with a candle, a box of matches, and a few tacks. To one group, he gave the matchbox with the matches inside it and to the second group, he gave the matchbox and matches separately. He asked them to attach the candle to a wooden door. The problem could be solved by using the empty matchbox as a platform for the candle by attaching it with the tacks to the door. The group that received the empty matchbox solved the problem faster. This led Duncker to conclude that the empty matchbox group solved the problem more efficiently because they were more likely to see the box as a potential platform rather than as a receptacle for matches. The matches in the box activated the preconceived notion of what a matchbox is and what it can be used for, thus inhibiting novel problem-solving.
Cox (15) and Halpern (16), on the other hand, believe that “problem-solving transfer” is possible if certain conditions are met. According to them, distinct domain skills must be utilized and practiced interchangeably. In the context of clerkship training in the 3rd year, this might, for example, include structured practice exercises, such as the Objective Structured Clinical Examination, and diagnostic reasoning exercises that reinforce automated skills. It might also include supervised use of domain skills from medicine or surgery clerkship in psychiatry clerkship and vice versa, and integrated walking-rounds such as those prescribed by the liaison model and training in combined medical-psychiatric units. Norman and Schmidt (17) have proposed that problem-based learning curricula too may enhance transfer of concepts to new problems. More importantly, the use of metacognitive critical thinking must be promoted. To achieve this, not only must knowledge, expertise, and strategies be acquired in a contextual rather than in a mechanistic manner, but the student must develop self-awareness and self-regulation of how he or she learns. According to the work of Bloom (18) and Ericsson (19), however, successful acquisition of critical thinking may well require extended practice to the tune of thousands of hours.
So, to develop an adequate degree of self-awareness and self-regulation, the student must have the opportunity to reflect on what is learned and how it is learned. He or she must acquire the ability to monitor the effect of conclusions in order to strategize and manage these skills in a flexible manner. Such reflection and monitoring occur best in the context of discovering what works and what does not. Learning and memory thus acquired must be constructed by the learner in a social and historic context, not a mechanistic one. This degree of complex problem solving would necessitate learning well beyond the 3rd year of training, into the 4th year subinternships and possibly into residency, before such generalization of skills is possible and fairly testable.
Thus, by encouraging students to hasten the task of arriving at a psychiatric diagnosis in 10–15 minute long, contrived, integrative Objective Structured Clinical Examination exercises, the process of learning might itself be short-circuited. Groopman (20) has eloquently pointed out the “representativeness” and “availability” type errors based upon thinking that is overly influenced by what is typically true, and, upon the ease with which relevant examples come to mind, respectively. The resulting failure to consider possibilities that contradict existing mental templates may well be reinforced by representative standardized patients. The consequences of learning promoted by such mechanistic exercises are dubious enough for the acquisition of general medical skills but may be disastrous for the fostering of psychiatric problem-solving skills in the medical context.
The author is grateful to Drs. Burt Landau, Bryce Templeton, and Sheila Vaidya for their help in the conception and development of this article.