An overview and, to some extent, a re-reading of a model of cognitive processes involved in real time listening to a piece of music—the “cue abstraction model”—is given here. Music listening is considered as a schematization process based on cue abstraction where a fundamental principle—the principle of similarity and difference—comes into play in listening over long time spans. The basic lines of this model were proposed about 15 years ago (Deliège, 1987a, 1989, 1991) and specific experimental procedures were developed to test the validity of its different aspects.