Program

Oral

Probabilistic syntax model of the birdsong - a study based on model selection.

[Speaker] Koumura, Takuya:1
[Co-author] Okanoya, Kazuo:1
1:University of Tokyo (Japan)

Birdsong is one of the most interesting sequential vocalizations. Especially, songs in Bengalese finches are excellent models for complex sequential vocalizations because sequence patterns of song elements are moderately complex, but not completely random or stereotyped. The sequencing rule of song elements are referred to as song syntax.
In this study, we investigated the model of song syntax in Bengalese finches, assuming that it can be modeled by a discrete probabilistic model. We selected the model that predicted the observed sequences most accurately. As results, the selected model had the following properties: element probabilities depended on more than one previous elements; syntax at the beginning of the song was different from that in the other part of the song; repetitions were described in the same model with other sequence patterns. These features were consistent with previous experimental findings and hypotheses. (Supported by Kakenhi #26240019 & #15J09948)

Advanced Search