Deliberate Practice

Anders Ericsson set out the original criteria with Ralf Krampe and Clemens Tesch-Romer in Psychological Review in 1993, after studying how music students at an elite academy spent their practice time. Writing for a medical audience in Academic Emergency Medicine in 2008, Ericsson described the same conditions: training aimed at improving a particular task, usually designed by a teacher or coach, with immediate feedback, time for problem solving and evaluation, and the chance to perform again to refine the behavior.

Ericsson spent much of his later career objecting to how loosely the term travels, arguing that most of what researchers labeled deliberate practice is better called structured practice because it drops one or more of those conditions. The distinction matters in medical education, where a workshop, a lecture on communication skills, and unguided repetition all get filed under deliberate practice while none of them meet the definition.

Applied honestly, the standard is demanding. A learner needs a task pitched just beyond what they can currently do, one specific goal for the attempt, feedback close enough in time that they can connect it to what they did, and enough repetitions that the behavior changes rather than the learner simply hearing about the change.

A resident who delivers bad news twice a year in front of a standardized patient is being assessed more often than they are practicing. AI patients can add repetitions on the learner's own schedule. What separates that from unguided repetition is Ericsson's feedback condition. Scoring each attempt against one named standard, with the learner's words quoted under the score, gives the learner a clear focus for the next attempt. That frees standardized patients and faculty for the encounters where a human in the room is what counts.