The structure, perception and generation of musical patterns

M. Nyssim Lefford

Interactive Institute, Piteå, Sweden

Structure distinguishes music from noise. When formulating that structure, musical artists rely on numerous mental representations, cognitive processes and sensory perceptions to organize pitch, rhythm, harmony, timbre and dynamics into musical patterns. Creators’ and non-creators’ perspectives of musical structure differ. Non-creators do not generally have access to the decisions, constraints and intentions that shape the composition of musical patterns. We have conducted a study which investigates creators’ perceptions of rhythmic musical structure in the patterns they generate.

The process of generating musical patterns may be compared to playing a game, with goals, constraints, rules and strategies. In this study games serve as a model for the interrelated mechanisms of music creation, and provide a format for an experimental technique that constrains creators as they generate simple rhythmic patterns. Correlations between composer-subjects’ responses and across experiments with varied constraints provide insight into how structure is defined in situ and how constraints impact creators’ perceptions and decisions.

Through the music composition games we investigate the nature of generative strategizing, refine a method for observing the generative process, and model the interconnecting components of a generative decision. The patterns produced in these games and the findings derived from observing how the games are played elucidate the roles of metric inference, preference and the perception of similarity in the generative process, and lead us to a representation of generative decision tied to a creator’s perception of structure.

Two composition game-experiments were developed for this study. The first investigates high-level definitions of structure and explores internal, external, integral and separable structures, and also, commonalities between creators’ strategies. The second game-experiment takes a bottom-up approach using similarity and preference metrics to identify the structural attributes most salient to composer-subjects as they generated rhythmic patterns.