Complexity, time and music

The concept of complexity as considered in terms of its algorithmic definition proposed by G.J. Chaitin and A.N. Kolmogorov is revisited for the dynamical complexity of music. When music pieces are cast in the form of time series of pitch variations, concepts of dynamical systems theory can be used to define new quantities such as the {\em dimensionality} as a measure of the {\em global temporal dynamics} of a music piece, and the Shanon {\em entropy} as an evaluation of its {\em local dynamics}. When these quantities are computed explicitly for sequences sampled in the music literature from the 18th to the 20th century, no indication is found of a systematic increase in complexity paralleling historically the evolution of classical western music, but the analysis suggests that the fractional nature of art might have an intrinsic value of more general significance.

💡 Research Summary

The paper revisits the notion of complexity, traditionally defined in algorithmic terms by Chaitin and Kolmogorov, and adapts it to the study of musical structure. Rather than attempting to compress a musical score directly, the authors first translate each piece into a one‑dimensional time series of pitch values. This conversion is performed on a representative corpus spanning the 18th to the 20th century, covering Baroque, Romantic, and early modern works. Once the pitch series is obtained, two complementary dynamical‑systems measures are applied.

The first measure is a dimensionality estimate derived from phase‑space reconstruction. By selecting an appropriate embedding dimension and time delay, the authors compute either the correlation dimension or the box‑counting dimension of the reconstructed attractor. This quantity reflects the global temporal dynamics of the piece: higher dimensions indicate a richer mixture of periodicities, non‑linear interactions, and rapid modulations, whereas lower dimensions suggest a more regular, predictable evolution of pitch.

The second measure is the Shannon entropy of successive pitch transitions. By constructing a probability distribution over adjacent pitch intervals, the entropy quantifies the local unpredictability of the melody. In this context, entropy captures short‑range variability, while the dimensionality captures long‑range structural complexity.

Both metrics are calculated for each work in the dataset, and the results are aggregated by historical period. Contrary to a common narrative that classical Western music has become progressively more complex, the analysis reveals no systematic increase. Dimensionality values cluster between roughly 1.2 and 2.1 across all periods, and entropy values remain within a narrow band of about 0.8 to 1.4 bits per note. Statistical tests show that period‑to‑period differences are not significant.

The authors interpret these findings through the concept of “fractional nature” of art. They argue that artistic objects occupy an intermediate zone between complete randomness (high entropy, high dimension) and strict determinism (low entropy, low dimension). This zone is continuously reshaped by cognitive constraints, cultural conventions, and the composer’s aesthetic goals. Consequently, complexity is not a linear evolutionary marker but rather a proxy for the level of “acceptable uncertainty” that human listeners tolerate.

Methodologically, the paper contributes a novel quantitative framework to musicology. By importing tools from nonlinear dynamics—phase‑space reconstruction, fractal dimensions, and information‑theoretic entropy—the study moves beyond traditional analyses based on harmony, form, or affective rating scales. It demonstrates that music can be examined as a dynamical system whose global and local properties are measurable and comparable across historical epochs.

Nevertheless, the authors acknowledge several limitations. The analysis isolates pitch while ignoring rhythm, timbre, and polyphonic interaction, all of which are essential components of musical texture. The corpus, though diverse in time, is relatively small and confined to Western classical repertoire, limiting the generality of the conclusions. Future work is suggested to incorporate multi‑modal time series (pitch, onset timing, spectral features), expand the dataset to include non‑Western and popular genres, and perhaps correlate the dynamical metrics with perceptual experiments on listener surprise or aesthetic preference.

In summary, the study provides empirical evidence that the global temporal dynamics (dimensionality) and local melodic unpredictability (Shannon entropy) of Western classical music have remained remarkably stable over three centuries. This stability supports the view that artistic creation balances order and randomness, maintaining a “fractional” level of complexity that aligns with human cognitive and cultural constraints rather than following a simple trajectory of increasing intricacy.