Feher, O. 1 , Smith, K. 1 , Ljubicic, I. 2 , Suzuki, K. 3 , Okanoya, K. 3 & Tchernichovski, O. 2
1 University of Edinburgh, UK
2 Hunter College, City University of New York, USA
3 University of Tokyo, Japan
Birdsong, like language, is socially learned, culturally transmitted, and exhibits structure on multiple levels: species-typical acoustic features define the phonology of song notes that are combined into repeated motifs and higher-level song bouts. Zebra finches raised in acoustic and social isolation do not develop normal wild-type songs. However, as we previously showed, normal songs can evolve from isolate songs by cultural transmission over a few generations. In the present study, we show that wild-type songs appear even in the absence of a tutor song, as a result of iterated learning within a single individual. We raised juvenile birds in social isolation but exposed to their own developing songs: whenever the birds pecked on a key, they heard one of their recently produced songs. For each experimental bird, a sibling was raised in complete acoustic and social isolation. Despite marked family effects, self-trained birds crystallised on more normal and rhythmic songs than their isolate brothers. This suggests that a bird's own developing song is sufficient to enable it to apply its imitation biases, via reinforcement learning, to produce normal song as an adult. This finding can inform us about similar biases and constraints in human language learning.