鳴禽の発声学習機構:雑音回避実験からの知見

抄録

Song learning of songbirds provides us a unique opportunity to study detailed mechanisms for vocal learning in various species, including humans. Recent studies in the behavioral neuroscience field have shown accumulated evidence indicating that their song learning is based on reinforcement via the auditory feedback of their own voice. The present review introduces an experimental paradigm that can elicit additional learning in bird's songs as a response to perturbations in the auditory feedback with noise presentation. This paradigm, named the noise-avoidance (NA) experiment, is quite useful for understanding the mechanism for song learning. Here I summarize findings obtained from the NA experiments, and review the current understanding of behavioral and neuroscientific mechanisms for feedback-based vocal learning. Additionally, I discuss computational aspects of the NA behavior in light of the reinforcement learning framework, and how the NA paradigm can be associated with the operant conditioning.