Test–retest reproducibility of fMRI patterns during musical and verbal imagery: an N=1 feasibility pilot protocol (revised 2026)

This is a revised, explicitly methodological study protocol (v2.0, October 2026), in English and Hungarian. It supersedes the author's earlier 2025 pilot protocol. The pilot asks a narrow question: within one person, how reproducible across repeated fMRI sessions are the activation patterns associated with internally generated imagery? The core condition is silent imagery of a personal childhood melody that has never been sung aloud, recorded or notated. It is compared with familiar-melody imagery and perception, auditory-verbal and visuospatial imagery, semantic retrieval, a counting baseline and rest, across 10–12 sessions over about 12 weeks. Internal task performance is checked behaviourally (vividness ratings, phrase-end keypresses, random stop probes, EMG check). Primary endpoints (split-session multivoxel pattern reliability and specificity) and regions of interest are fixed in advance. The Dice overlap of thresholded maps is a secondary, descriptive measure with a defined threshold and interpretation. Analysis is blinded and runs in an open, automated pipeline. The study measures pattern reliability only; it does not recover memory content. The revision follows expert feedback received in June 2026 and narrows the claims of the earlier version. No data have been collected. The study will start only after funding, a host-site agreement and ethics approval, and the final protocol will be preregistered on OSF.