Diagnostic discrepancy and clinical inference: a social-cognitive analysis.

This review posits that clinical inference be studied as an instance of human inference. Normative rules for clinical inference may be derived from those for human inference in general, but clinicians choose not to adhere to them, relying instead on possibly biasing and suboptimal methods. A seven-phase model of clinical inference, with each phase containing particular tasks, is presented. Performance on these tasks depends on the effect of cognitive and environmental influences and the application of cognitive strategies, with erroneous performance leading to impaired clinical inference.