Investigating complexity and preference for structure in Savannah sparrow song by Kate Swann ’26

About this event

Investigating complexity and preference for structure in Savannah sparrow song by Kate Swann ’26, Friday May 8, 1:00 – 1:50pm, North Science Building 015, Wachenheim, Mathematics Thesis Defense Savannah sparrow (Passerculus sandwichensis) songs are socially learned and undergo cultural evolution. The song’s middle segment usually consists of 1-3 syllables. In a well-studied population, middle segment types follow a power-law distribution. Random copying models can predict a power-law distribution, but cultural transmission can also give rise to such a distribution. We used Bayesian Iterated Learning Models to test two hypotheses: 1) Savannah sparrows copy middle segments randomly, or 2) they select segments with a regular structure that facilitates learning. In our model, the prior distribution encodes juvenile Savannah sparrows’ biases about how syllables may be combined. Juveniles go through two rounds of song learning, once during the two months after hatching and again at 9-10 months of age, just before their first breeding season. During those two phases, young birds hear songs and generate a posterior distribution. That posterior distribution then affects which middle segment the juvenile sings, which in turn influences the next generation. We compared our model predictions to patterns in empirical data to evaluate our hypotheses about song learning. We measured entropy to estimate the proportion of syllable combinations generated by free choice and by statistical rules, respectively, to further test our hypotheses.