On the control of unstable objects: the dynamics of human stick balancing.

Although the dominant focus of human motor control and learning has been on sequence production and force field adaptation, the control of unstable objects has received much less attention. In this review, we summarize a recent body of work on the control of unstable objects using the paradigmatic example of stick balancing. In a series of four studies, we examined learning-induced changes in balancing an inverted pendulum at the fingertip. First, we identified the statistical distribution that captures the spatio-temporal dynamics of hand displacement trajectories. The probability distributions for changes in transverse plane fingertip speed revealed Levy-distributed dynamics that changed as a function of learning. Following this, we quantified the nature of the intermittent control mechanisms that gave rise to these Levy-distributed fingertip trajectories. The results pointed to the existence of a drift-and-correct mechanism. In a subsequent set of studies, we report task-specific changes in the coupling between posture and hand displacements as subjects became more proficient at stick balancing. Such changes revealed the presence of a hierarchical control mechanism that intermittently switched between coordinative and individual subsystem control. Finally, we show the effects of attentional focus and concurrent cognitive tasks in expert stick balancers. Our results are discussed in terms of an intermittent, drift-and-correct mechanism that is sensitive to the cognitive and physical task demands of the stick-balancing task. Dynamical system’s approaches to motor control and learning are discussed in detail.