How does expertise develop in motor skills? Through research conducted on a challenging task--balancing a stick on the fingertip--some of the secrets of motor expertise have been discovered. Counter-intuitively, random perturbations may be a key component of the solution. Research conducted by John Milton and his students at Claremont McKenna College (a primarily undergraduate institution) suggests that expertise depends on an underlying "drift and act" control strategy, in which unconscious corrective actions are made only when deviations become sufficiently large. Voluntary corrective movements are more likely to be disruptive than effective. The investigators will extend this work to balancing on a wobble board in order to establish the fundamental importance of intermittent control for maintaining balance in an unpredictable, noisy environment.
With support of the National Science Foundation, Dr. Milton brings together an international team of scientists to work with undergraduate students to advance our understanding of expertise development in balancing. The investigators focus on balancing tasks for three main reasons: 1) the tasks are sufficiently difficult to allow identification of levels of expertise; 2) they are sufficiently constrained to permit careful comparisons between observation and prediction; and 3) expertise can be dramatically increased with just a few days of intensive practice. High speed motion capture technologies are used to simultaneously monitor the positions of the stick, body and eye movements (a measure of visual attention) as expertise develops and key elements are varied.
The research project is designed to excite and motivate undergraduate students, while at the same time teach them to work effectively in international teams. Because proper balance control is essential for the expert performance of many motor tasks, it is anticipated that studies of balance control will provide potential solutions for problems ranging from minimizing the risk of falling in the elderly to the development of two-legged robots and novel rehabilitative strategies.
This grant studied stick balancing at the fingertip. There were two goals: 1) Investigate how the nervous system controls this complex voluntary motor task; and 2) Develop a program that integrates research and education in a manner that both excites and motivates undergraduate students. Stick balancing is a difficult motor task because the nervous system must learn to stabilize an unstable position using time-delayed feedback. The importance of time-delayed feedback is demonstrated by the observation that longer sticks are easier to balance than shorter ones: once the stick becomes sufficiently long its rate of movement becomes slow relative to the time required to make a corrective movement. Our starting hypothesis was that the nervous system used a "drift and act", or safety net control strategy in which corrective actions were made only when deviations from the vertical position became sufficiently large. Such control strategies are robust in the presence of noisy perturbations. Drift and act control implies the existence of a sensory threshold. We showed that there was sensory dead zone in the detection of the vertical displacement angle of 3-4 degrees in the anterior-posterior direction which arises because humans have poor depth perception. Thus the feedback is turned off ("drift") when the angle is less than 3-4 degrees and turned on ("act") when the angle is greater than 3-4 degrees. There is also a threshold for sensory input from mechanoreceptors. A sensory threshold has no effect on large-scale stabilization. Mathematical modeling and computer simulations demonstrated that the nature of the feedback can be identified by measuring the time delay and length of the shortest stick that can be balanced. The time delay was measured using a visual blank out technique. Briefly, sensory blank outs of variable length were produced by balancing the pole on a table tennis paddle to mask inputs from cutaneous mechanoreceptors and then blanking out visual input by applying a voltage to LCD optical shutters. The delay measured from the recovery of balance after the blank out is terminated was determined to be 0.23s. The shortest stick that can be balanced by expert stick balancers for greater than 4 minutes was 0.3m. Together, these observations support the suggestion that the nervous system uses an internal model of stick balancing to compensate for the time delay. We showed that in the presence of time-delayed feedback, a sensory threshold effects dynamics. First, the interplay between time-delayed feedback and a sensory threshold can produce transiently stabilized upright positions lasting minutes even though the stick eventually falls. Second, there can be the generation of complex dynamics such as limit cycle oscillations and micro-chaos. Finally we demonstrated the feasibility of developing strategies to improve balance control. First, computer simulations and experiments on virtual stick balancing demonstrated the presence of a signature, or trigger, for impending stick falls suggesting the possibility that stick falls can be predicted before they occur. Second, we showed that we showed that low frequency (< 50 Hz), low amplitude (0.1mm) vibration of the fingertip can increase the length of time that a transiently stabilized balance position can persist. This observation was verified for balancing on a wobble board. Broader impact: This project resulted in the training of 27 undergraduate students from Claremont McKenna, Harvey Mudd, Pitzer, Pomona and Scripps colleges (which together comprise the Claremont Colleges) and two students from other institutions. There were 8 student co-authors on peer-reviewed publications. Our findings may also have impact on: 1) Engineering: The stabilization of an inverted pendulum is an important benchmark for assessing the robustness of feedback control mechanisms in engineering applications. The effects of dead zones and the possibility of transient stabilized states may have practical applications.; 2) Mathematics: The effects of dead zones on the dynamics of delay differential equations in the presence of noisy perturbations are not well understood. The observation that such equations may arise in some aspects of neural control should spark mathematical interest.; and 3) Human resources: The concept of drift and act control for stick balancing is similar to the ankle-hip-step strategies used to maintain postural balance in response to increasingly large perturbations. Thus our observations may motivate research into strategies to minimize the risk of falling in the elderly.
