An oscillatory neural network model for birdsong learning and generation: Implications for the role of dopamine in song learning

Abstract

We present a model of bird song learning and production in which the motor control pathway is modeled by a trainable network of oscillators and the Anterior Forebrain Pathway (AFP) is modeled as a stochastic system. Song learning in many species of birds is divided into two phases. In the first phase, the sensory phase, the male bird listens to the tutor song of another male bird in the colony and memorizes some aspect of the tutor song. In the second phase, the motor learning phase, the bird establishes the songs learnt earlier by rehearsal aided by auditory self-feedback. We hypothesize that: (1) the songbird learns only evaluations of songs during the sensory phase; (2) the AFP plays a role analogous to the Explorer, a key component in reinforcement learning (RL); (3) the motor pathway learns the song by combining the evaluations (value information) stored from the sensory phase, and the exploratory inputs from the AFP in a temporal stage-wise manner. Model performance on real birdsong samples is presented. Impaired song output under conditions of lesions of AFP nuclei, including the Lateral Magnocellular Nucleus of the Anterior Neostriatum (LMAN) and Area X, is studied. The model also proposes a role for dopamine signal in song learning and shows that under dopamine-deficient conditions, similar to those of Parkinson’s disease, song learning is impaired.