In this talk, I integrate scholarship from aural skills pedagogy and cognitive science to advance both the study and practice of musical imagery development. I propose that musical imagery as developed in North American aural skills pedagogical approaches is a form of expert memory, or long-term working memory (LTWM, Ericsson & Kintsch, 1995; Ericsson, 2018). My analysis of the literature shows that pedagogical methods discussed by scholars in the North American tradition can be divided into four discrete categories each with their own distinct function in expertise acquisition. These include schema/chunk formation, semantic encoding, construction of retrieval cues and retrieval structures, and the gradual improvement of memory skill (or LTWM formation). I also integrate insights from auditory imagery psychometrics (Halpern, 2015) through a mapping of imagery processes (generation, manipulation) and their corresponding subjectively available properties (perceived vividness, control) onto the established LTWM framework. To conclude, I introduce my recent work operationalizing musical imagery LTWM as a form of multimodal dual-coding fluency (Paivio 2007).