The process by which humans stabilize bipedal stance represents a confluence of changes associated with musculoskeletal maturation and experience-based sensorimotor learning. While investigations have documented a variety of changes with increased bipedal experience, such as reduced velocity and frequency of postural sway and concomitant refinements in muscle activation sequences, the extent to which these changes may be ascribed to growth versus learning processes has not been well characterized. For example, reduced sway frequency is a natural consequence of increasing body height but alternatively, may be explained by active modulations in motor commands specifying the timing and magnitude of muscular activation sequences. It is clear that both types of influences are needed to explain postural development. However, a parsimonious framework for understanding and explaining postural development has yet to be clearly articulated and validated against empirical observations. As such, the purpose of this dissertation was to initiate the development of such an account through a combination of empirical and computational studies.

In this dissertation, data are presented from a longitudinal study of upright posture in infants ranging from the onset of independent sitting until 9 months of walking experience; this dissertation focused on the particular period spanning from walk onset onward. Infants participated in a quiet stance task involving hand contact with a surface that was either static or dynamic as well as an independent stance condition. Empirical analyses were performed to estimate the statistical properties of sway and characterize adaptations to static and dynamic manipulations utilizing the touch surface. An unexpected lack of significance for sway magnitude was observed in all conditions. Robust effects, however, were found across measures of rate properties of sway. Taken in the context of previous literature, the empirical observations were used to guide a final study utilizing computational techniques to test hypotheses regarding potential sources of change in postural development. First, the mechanical and computational requirements for postural stabilization were systematically assessed through a review of extant models of both stance and motor learning. Armed with insights from this review, the final study examined an autonomous reinforcement learning algorithm, that was designed to capture the essence of how a human may stabilize his or her posture under the tutelage of exploratory action. Simulation results provided evidence in support of conclusions regarding changes in rate-properties of postural sway and underlying associations with physical growth as well as calibration of both sensory and motor system parameters. Further, simulations emphasized the importance of inclusion of noise in biologically-relevant aspects of the model, such as in sensory and motor processes, as well as the need to consider physical morphology as a primary constraint on sensorimotor learning in the context of upright postural development.