Abstract

One of the most universal features of music is the construction of melodies using a restricted set of pitch categories – i.e. musical scales. Despite this universality, we know little about how scales have evolved, cross-culturally. Some analogies can be drawn between the evolution of scales and biological evolution: they both involve linear sequences of discrete units; these units undergo random change and directional selection. Scales are more complex in the mechanisms of change: in biology there are a limited set of changes (mutations, insertions, deletions, inversions), while scales can be created entirely anew without reference to any previous template. However, the possible space of scales is tiny (typically 7 or fewer notes, just-noticeable-differences of at least ~5 cents results in millions) compared to biological sequences (a protein with 100 amino acids can have any one of 4300 sequences). Further, there are some proposed theories on how scales are selected. It thus ought to be possible to study their evolution, yet there are only a few quantitative studies of the evolution of scales. To study of the evolution of scales, we first scoured the ethnomusicological literature of the last century to assemble a cross-cultural database of scales. The data point to convergent evolution, which is clear evidence of selection pressures. We then consider a set of selection biases, and test (using a generative inference approach) which hypotheses may best predict the distributions of scales. Finally, we think about how the different forces (physical, perceptual, cultural) involved in drift and selection may vary considerably between and within societies, across geography and time.