Is analytic information processing a feature of expertise in medicine?

Diagnosing begins by generating an initial diagnostic hypothesis by automatic information processing. Information processing may stop here if the hypothesis is accepted, or analytical processing may be used to refine the hypothesis. This description portrays analytic processing as an optional extra in information processing, leading us to questions if it actually contributes to diagnostic performance, and whether it heralds expertise or a lack of expertise. When we encourage students to solve problems using analytic processing -- as is our teaching tradition -- are we helping or hindering diagnostic performance and the evolution of expertise? The relationship between information processing, expertise and diagnostic performance is complex. At least four additional variables affect this relationship: context; task difficulty; clinical domain; and experimental conditions. Therefore, we cannot make a generic statement about the relationship between processing, expertise and performance -- we can only say that when given a problem containing certain information, of certain difficulty, in a certain clinical domain and under certain experimental conditions, analytic processing appears to improve diagnostic performance and be a feature of expertise. No single processing strategy is a panacea. Training students in both automatic and analytic processing offers flexibility in information processing and better prepares them for the wide variety of problems that they may encounter.