We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Measuring context dependency in birdsong using artificial neural networks.
- Authors
Morita, Takashi; Koda, Hiroki; Okanoya, Kazuo; Tachibana, Ryosuke O.
- Abstract
Context dependency is a key feature in sequential structures of human language, which requires reference between words far apart in the produced sequence. Assessing how long the past context has an effect on the current status provides crucial information to understand the mechanism for complex sequential behaviors. Birdsongs serve as a representative model for studying the context dependency in sequential signals produced by non-human animals, while previous reports were upper-bounded by methodological limitations. Here, we newly estimated the context dependency in birdsongs in a more scalable way using a modern neural-network-based language model whose accessible context length is sufficiently long. The detected context dependency was beyond the order of traditional Markovian models of birdsong, but was consistent with previous experimental investigations. We also studied the relation between the assumed/auto-detected vocabulary size of birdsong (i.e., fine- vs. coarse-grained syllable classifications) and the context dependency. It turned out that the larger vocabulary (or the more fine-grained classification) is assumed, the shorter context dependency is detected. Author summary: We investigated context dependency in over ten-hour recordings of Bengalese finches' songs using a neural-network-based language model, whose flexible fitting enabled non-parametric analysis of the birdsong. For this aim, we proposed an end-to-end unsupervised clustering method of song elements (syllables) based on a statistical optimization married with an artificial neural network. The proposed method enabled individual-invariant classification of Bengalese finch syllables, even though substantial individual variations were present in the raw acoustic features. In the meantime, the clustering results were kept consistent with the individual-specific classification that have been used in previous studies. Based on these clustering results, the detected length of context dependency in Bengalese finch song was consistent with findings from behavioral and neuroscientific studies, while it went beyond the dependency found in the previous computational analyses based on Markovian modeling. We also found that the context dependency became shorter as a greater number of syllable categories were assumed.
- Subjects
ARTIFICIAL neural networks; BIRDSONGS; MODERN languages
- Publication
PLoS Computational Biology, 2021, Vol 17, Issue 12, p1
- ISSN
1553-734X
- Publication type
Article
- DOI
10.1371/journal.pcbi.1009707