A series of experiments investigated cognitive processes involved in listening to a piece of music, focusing in particular on the abstraction of surface features (here referred to as cues). Subjects listened to an unfamiliar piece in a familiar musical idiom, and their sensitivities to aspects of the just-heard piece were used to elucidate the nature of their representations of the piece in recent memory. The study also sought to assess the capacities of subjects to use any declarative knowledge of aspects of tonal structure that they possessed in organizing musical material. Three experiments made use of different procedures to address these issues, using either a single short tonal piece—Schubert's Valse sentimentale, D. 779, op. 50, no. 6—or a variant of this. The first two experiments used nonmusician subjects and examined (1) the cues abstracted in listening to the piece and (2) subjects' ability to identify the temporal location of segments of the piece after listening. The third experiment explored the constructional abilities of musician and nonmusician subjects, requiring them to create a coherent piece by ordering the segments that made up the original piece. The results of these experiments indicated that although the abilities of musicians differed from those of nonmusicians, both groups of subjects exhibited a weaker sensitivity to features of musical structure than to cues abstracted from the musical surface.
Lerdahl and Jackendoff propose a grouping theory that can apply to both the global and the local structures of the process of listening to music. It is a set of rules expressing the intuitive organization of groups in music perception. The "proximity rules" describe the length differences and the "change rules" describe the modifications in the acoustic or temporal state of sound structures, in relation to Gestalt Theory. As such, they propose a testable hypothesis on certain aspects of music perception. Two experiments are reported, which do not go beyond segmentation into two levels of grouping. They compare the grouping behavior of two categories of subjects, nonmusicians and musicians. Four questions are raised: (1) Do the segmentations reported by subjects answer in all respects the predictions of the rules? (2) Are they available to both categories of subjects? (3) Do they cover all grouping situations in music? (4) Are they of equal perceptual salience? The first experiment used material taken from compositions in the Western art music repertoire (Bach to Stravinsky). The second one put the rules in conflict in simple melodic sequences, where a combination of all possible conflicts between pairs of rules was designed. The results show the validity of the rules. Nonmusicians had poorer performances with repertoire music sequences. Yet the two categories of subjects do not show a radically different grouping behavior. New rules are suggested by the segmentations that were not in accordance with the theory. They also show some difficulties for the length rules deriving from the Gestalt Similarity law.