Sean Collins

Clinical reasoning requires a patient-specific representation capable of supporting explanation, prediction, intervention, and continued inquiry. Scientific knowledge, however, is inherently general: it represents causal mechanisms, relationships, and interventions across classes of systems or patients rather than any particular individual. This paper argues that the transformation from generalized scientific knowledge to a patient-specific explanatory representation constitutes a distinct conceptual problem that existing accounts of clinical reasoning often presuppose but rarely examine explicitly. I call this transformation clinical instantiation. A generic causal model represents current scientific understanding of the causal structure believed to characterize a class of systems, whereas an instantiated causal model represents the clinician's provisional explanatory understanding of a particular patient at a particular time. Instantiation is the representational process through which generic causal knowledge and patient-specific evidence jointly constrain construction of that individualized model. The resulting representation supports mechanistic reasoning, prediction, intervention, and decision making while remaining uncertain, fallible and revisable. Clinical inquiry is therefore recursive: evidence constrains instantiation, the instantiated model guides reasoning and action, and the consequences of action generate new evidence that may prompt subsequent instantiation. This framework complements Bayesian reasoning, structural causal models, illness-script theory, decision theory, and related approaches by identifying the representational transition on which their patient-specific use depends. It also distinguishes personalization as representation from merely accumulating individualized patient data. Making this transition explicit clarifies the relationship between population knowledge and clinical judgment and provides a conceptual basis for future work in personalized medicine, clinical education, patient-specific causal modeling, and intelligent clinical systems.