Spatial scaffolds or spatial dependency? A structured evidence map of AR/MR interfaces for piano learning and practice.

Augmented- and mixed-reality (AR/MR) systems for piano learning and practice provide spatially aligned key highlights, piano-roll cues, virtual hands, performance feedback, and gamified guidance. Previous reviews have already identified unresolved questions concerning sight-reading, cognitive load, retention, transfer, and the possibility of interface dependence. The present review therefore does not treat these issues as previously unrecognized research gaps. Instead, it provides a structured evidence map of empirical studies of AR/MR and spatially augmented interfaces for piano learning and practice and examines how interface design relates to the kinds of learning outcomes that have been measured. Backward and forward citation tracking, supplemented by open bibliographic metadata and publisher searches, identified 36 candidate records. After two duplicate, expanded, or preliminary reports were removed or merged, 34 records were screened, 20 reports underwent detailed eligibility assessment, and 18 empirical reports were retained in the core evidence map. The reports were coded according to two dimensions: notation continuity-notation-preserving, notation-bridging, and notation-replacing interfaces-and scaffold withdrawal-persistent, contingent, progressively faded, or removed guidance. The evidence is strongest for proximal outcomes, including immediate note accuracy, task completion, usability, engagement, motivation, and perceived workload. A smaller number of studies examined repeated practice, motor-skill development, or performance after guidance was reduced or removed. However, delayed unaided transfer to conventional staff notation remains weakly tested. The available evidence therefore indicates that spatial guidance can provide effective immediate task support but does not establish that it consistently produces independent music-reading skills or that it necessarily creates dependency. The review's contribution lies in linking notation continuity, scaffold withdrawal, and outcome distance to clarify what existing studies can-and cannot-demonstrate about learning.