Birdsong requires fine coordination of the muscles in the vocal organ and respiratory system. The song system, which is the neural architecture dedicated to song production in oscine birds, poses a beautiful opportunity to model the activity of a circuit of interconnected neural nuclei for the production of complex behaviour.
Such a model has been developed previously by our laboratory. Our model makes specific predictions about the timing of population neural activity in different nuclei needed to produce song and can reproduce respiratory patterns observed in canary song phrases of different types.
In this work, the model was used to generate respiratory patterns that correspond to two types of phrases of canary song, given a proposed sparse activity in nucleus HVC (proper name), one of the nuclei involved. This telencephalic nucleus plays a key role in the production of motor commands, but the neural code it uses is still under debate.
We put our model to the test by recording extracellularly from nucleus HVC in singing canaries, isolating single-units that fire during particular phrase types and analyzing at which song instances these neurons fire.
We found that grouped activity across animals is in good correspondence to the population activity that can produce realistic respiratory patterns in our model for the two types of phrases studied. Furthermore, the experimental data show activity in additional song instances that will enrich further iterations of the model.