The Emergence of a Universal Rhythmic Feature: Simple Models Can Produce Categorical Rhythms.

Across musical cultures, rhythm consists of discrete categories of interval durations. Such rhythmic categories are also increasingly quantified in various nonhuman species' displays. However, their evolutionary origins are still largely unknown. Complementing cross-species comparative work with computational modeling can help us understand the cognitive mechanisms underlying the emergence of this universal rhythmic feature and the minimum requirements for producing it. This study investigates whether minimal computational models can produce rhythmic categories. We compare two computational models: a single spiking neuron model representing a minimal neural system, and a model of cricket stridulation as a minimal synchronization mechanism. Both models transform a random temporal sequence into a more structured, isochronous rhythm; that is, randomly distributed temporal intervals get more similar in duration. An isochronous input sequence, in contrast, combined with the models' intrinsic bias, has a more complex effect on the produced temporal patterns. At frequencies that closely relate to the models' intrinsic frequency, the models produce stable temporal patterns with rhythmic categories. Our results show that rhythmic categories can emerge from simple mechanisms, likely shared across species, especially when multiple individually isochronous mechanisms interact. As such, we should expect to find categorical rhythms across an even larger range of animal displays.