A comprehensive longitudinal study aims to characterize the behavioral manifestations of musical WM. Musicians and non-musicians will complete 17 variations on a delayed match-to-sample WM task in which participants are presented with a pair of melodies and determine whether they are identical. The task addresses either simple WM maintenance or manipulation (retrogradation), maintenance with ipsimodal or allomodal distractors, and in the auditory or visual modalities. Each of these eight tasks is also administered with letter sequences mapping the original melodies to test for the effects of domain-specific stimuli, which results in 16 different tasks. An extra task using visual polar representations of the melodies contrasts the influence of dynamic and static visual stimuli. Within-participant data across the whole series of experiments enables examining the role of WM type (maintenance or manipulation), distractors (ipsimodal or allomodal), sensory modality (auditory or visual), and stimulus kind (melodies or letter sequences), at the individual and at the group levels.

This approach allows us to test specific hypotheses such as the following:

• Experts have greater WM capacity and manipulation abilities for domain-specific expertise-related stimuli. Preliminary results confirm our expectation that musicians outperform non-musicians in all musical tasks but have similar abilities regarding comparable non-musical dynamic and static stimuli. Moreover, difficulty in the musical tasks correlates more strongly for musicians, while difficulty in the non-musical tasks shows stronger correlations across musical and non-musical tasks for non-musicians, which points to a special WM mechanism operating more prominently in musicians.
• Experts rely on multimodal representations of expertise-related stimuli. Preliminary results suggest that the impact of task on performance does not differ significantly between groups. However, an advantage of musicians in auditory maintenance with visual distractors might point to a more robust multi-sensory integration for sequential stimuli in musicians.

All paradigms are implemented online using PsyNet, an in-house experimental platform for psychological research. The data from healthy adult musicians are collected through CAP Recruiter, an in-house recruiting system targeting special populations—professional musicians in this case. The data from non-musicians are collected via Prolific and are representative of the general population. Item response theory (IRT; Mellenbergh, 1994) provides a reliable metric of item difficulty and individual ability generalizable across participants, experimental manipulations, and scientific approaches. We leverage the psychometric properties of our metrics to select stimuli for subsequent studies and develop a computer adaptive testing (CAT; Meijer et al., 1994) staircase paradigm for use in subsequent studies. Finally, we validate our paradigm with standardized metrics of musical and general WM.