Birdsong is a classic model in neuroethology, and continues to provide valuable insights into organization of brain and complex learned behavior. Here we address two central issues: the organization of the motor system for song production, and the realtime representation of feedback. A popular model in birdsong reflects a top-down command-control view, the standard for motor control. Instead, we show that activity in the “HVC” premotor cortical structure has information of the fine-grained dynamics of vocal movements, including near-zero delay relative to muscle movements. These results can be explained by interactions between top down and bottom up influences on motor control, analyzed from the perspective of nonlinear dynamics. To understand how changes in singing behavior are represented in realtime, we turned to an unlikely preparation, HVC whole cell slice recordings. We show that the intrinsic somatic neuronal properties of a class of HVC neurons show unanticipated homogeneity within individual birds and variation between individuals. The distribution of intrinsic properties are regulated during song development, are modified in adults by changes in singing that are driven by abnormal auditory feedback, and in adults these changes occur rapidly, at most taking hours and perhaps only seconds. These data provide compelling evidence that intrinsic somatic electrophysiological properties contribute to behavioral plasticity in a fashion previously conceptualized only for synapses. Collectively, these results have broad implications for neural coding.