Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2010.Cataloged from PDF version of thesis.Includes bibliographical references (p. 201-211).An infant reaches out for her new toy, struggling day after day to simply grasp her fingers around it. A few years later, she hits a tennis serve, perfect in the timing of its intricately choreographed movements. How does a young brain learn to use the muscles it controls, to properly coordinate motor gestures into complex behavioral sequences? To a surprising extent, for many advanced vertebrate behaviors this knowledge is neither innately programmed nor acquired via deterministic developmental rules, but must be learned through trial-and-error exploration. This thesis is an investigation of the neural mechanisms that underlie the production and maturation of one exploratory behavior - the babbling, or subsong, of a juvenile zebra finch. Using lesions and inactivations of brain areas in the song system, I identified neural circuits involved in the production of subsong. Remarkably, subsong did not require the high vocal center (HVC) - a premotor structure long known as the key region for controlling singing in adult birds - but did require the lateral magnocellular nucleus of the nidopallium (LMAN) - the output region of basal ganglia-forebrain circuitry previously considered a modulatory area. Recordings in LMAN during subsong revealed premotor activity related to the vocal output on a fast timescale. These results show, for the first time, the existence of a specialized circuit for driving exploratory motor control, distinct from the one that produces the adult behavior. The existence of two neural pathways for singing has raised the question of how motor control is transferred from one pathway to another and, in particular, how the control of song timing develops in these pathways. I found that early singing can be decomposed into mechanistically distinct "modes" of syllable and silent gap timing - randomly-timed modes that are LMAN-dependent and developmentally-acquired, consistently-timed modes that are HVCdependent. Combining acoustic analysis with respiratory measurements, I found that the consistently-timed mode in gap durations is formed by brief inspiratory pressure pulses, indicating an early involvement of HVC in coordinating singing with respiration. Using mild localized cooling - a manipulation that slows down biophysical processes in a targeted brain area - I found that the circuit dynamics intrinsic to HVC and LMAN are actively involved in controlling the timescales of distinct behavioral modes. In summary, this work demonstrates the existence of two motor circuits in the song system. These circuits are specialized for the generation of distinct types of neural dynamics - random exploratory dynamics, which are dominant early in life, and stereotyped sequential dynamics, which become dominant during development. Characterization of behaviorally-relevant dynamics produced by neural circuits may be a general framework for understanding motor control and learning.by Dmitriy Aronov.Ph.D.