Birdsong has become a widely studied model for vocal learning and motor behavior. It is well understood that sensorimotor area HVC (proper name) is essential for song learning and production, but our understanding of the role of HVC neurons remains incomplete. In one view, HVC serves only as a clock -- a specialized tissue designed to propagate timing information to other areas of the brain. From this perspective, HVC is agnostic about the motor periphery, providing other brain areas with a timing signal from which to construct the carefully sequenced gestures that constitute birdsong. Another hypothesis is that the time-varying activity in HVC contains information about the state of the motor periphery. The gesture trajectory extrema (GTE) model is one of many models that take this second view. Neurons in HVC that match the GTE model were first discovered by using a biomechanical model of the periphery to estimate the motor state during song.,In this thesis, I took three approaches to investigate HVC's role in song. First, I used large, dense micro-electrode arrays to sample from large populations of HVC neurons. I recorded from HVC while the birds were sleeping, and used auditory playback of the bird's own song to entrain the motor system. ,I found HVC activity to be significantly modulated by syllable onsets and vary depending on the syllable type. These observations are,consistent with HVC encoding the gestures of the motor periphery. The time varying population rate appeared consistent with the GTE model, but as yet this has not been validated statistically.,Second, I used similar electrode arrays in HVC, but now combined with simultaneous recordings in the muscles of the syrinx (N=3). By recording from the muscles of the syrinx, I was able to directly examine the motor output of the song system without using the biomechanical model. Perhaps due technical limitations, I was unable to find the expected reliable auditory response to song playback during sleep, yet during sleep the motor system engaged in short spontaneous replay events. Analyzing these replay events revealed a tight coupling between activity in HVC and activation of syrinx muscles. These data suggest that HVC may directly encode features of the motor periphery, a prediction of the GTE model. However it was unclear whether this coupling also existed in the awake state, because the data were collected during sleep.,Finally, in a third set of experiments, I develop a system to record from the syrinx muscles and neurons in HVC simultaneously in a singing bird. I found that the tight coupling between the muscles of the syrinx and neurons in HVC observed during sleep, was also maintained during singing (N=7).,Taken together, the data collected in this thesis suggest that the songbird motor periphery is represented in HVC during song. While these results cannot rule out the existence of a "clock" subpopulation of HVC neurons, HVC may have a premotor role in singing, and is tightly locked to the activity of syringeal muscles. These results are consistent with some of the predictions of the GTE model, albeit the zero-lag prediction of the GTE model remains to be confirmed. The data provide compelling positive results refuting a central prediction of the "clock'" model -- that HVC activity is divorced from muscle dynamics.