A common trend in neuroscience is convergence on selected model systems. Underlying this approach is an often implicit assumption that mechanisms observed in one species are characteristic of all related species. Although the model system approach has been extremely productive, it might not account for all of the mechanistic differences between species that differ behaviourally. Using the neural system that regulates song learning in songbirds as an example, we demonstrate how integrating model system and comparative approaches can lead to a more complete picture of neural mechanisms, and can resolve issues raised by a focus on selected species.