Human Performance and Goal-Directed Tasks

We are goal-directed creatures. Often there are clear assessment metrics that relate to performance (e.g., hitting a tennis shot where desired; successfully trading off speed versus accuracy). The process of learning skills is often viewed as an optimization process, and detailed discussion in beyond the scope of this class. Here we focus on one small aspect that is often of importance for dynamic movements, and helps us develop an appreaciation for the remarkable capacity of the brain to learn, and perform many tasks well: scaling of movements.

Goal-directed skilled movement often requires scaling of dynamic movements. For instance, a fast tracking movement can be made different distances. In such cases neuromotor pulse widths and/or heights must scale. Similarly, there is scaling of a golf or tennis swing, a kick, etc. Thus a "motor program" associated with the skill cannot always be retrived and implemented directly, without first being adjusted to current aims and circumstances. Typically the cerebellum and basal ganglia are involved in planning such movement scaling, and typically there are "success" criteria related to performance. In an optimization problem, such "performance criteria" are used by an algorithm (e.g., within the brain) to "sculpt" and scale a planned movement.

As a concrete example, consider the classic neuromotor implementation of a simple point-to-point target tracking movement involving one joint, such as an elbow flexion-extension movement. From a neuro-control perspective, moving quickly from one point to another is initiated by strongly activating the agonists (prime mover) while relaxing any drive to the antagonists. This causes a muscle-induced moment across the joint, which will cause a movement to be initiated toward the new, desired target position. The acceleration will depend on the inertia of the segment to be rotated (Newton's law) and the velocity will increase as this acceleration is integrated. After reaching roughly the half-way point, there is a need to start decelerating the limb segment, i.e. to start planning how to stop. One approach might be to simply turn off the agonist drive at just the right time and "coast" into the new position (taking advantage of some natural "viscosity" (velocity-dependent friction) within the muslce-joint apparatus. But in general, there is a need for active braking, or "clamping" of the movement through a temporary change in sign of the moment. This is accomplished by activating the antagonist muscles while lowering the neuromotor drive to the agonists. The movement rapidly slows toward zero. But since the new moment is not functioning as a frictional brake, there is often a need for an additional agonist clamping pulse. We've just described the classic "tri-phasic burst" pattern (agonist-antagonist-agonist) for very fast goal-directed tracking movements that has been observed across many joints (e.g., seen in EMG's for isolated movements of the elbow, shoulder, wrist, horizontal head, ankle). Typically the first (agonist) burst is large in both pulse magnitude and width (e.g., over half of the movement time), the second (antagonist) burst starts just before the first agonist burst ends and is nearly as high in magnitude but for less time, and the third (agonist second burst) overlaps some with the end of the second. Also, there is often a degree of coactivation around and shortly after the end of the movement (temporarily "stiffening" the joint) that gradually dies down. Finally, there needs to be a sustained "step" increase, typically small, in the agonist relative to the antagonist so as to maintain the new joint position and not drift back towards the old position.

Clearly the experienced human without functional impairments relevant to this type of movement task can easily adjust movement speeds, if desired. Also, the human can adjust their strategy to targets of different magnitudes. Thus both time-scaling (or speed-scaling) and magnitude-scaling are possible. How does such scaling occur?

Let's first consider time-scaling, starting from the fastest movement. While the details depend on biomechanical and neuromotor factors for the given system, in general as the desired movement time increases (speed lowers), the initial agonist burst will have a lower magnitude (for some systems making large movements it may first remain saturated at maximum and start by scaling its pulse width). The second (antagonist) burst will also scale down in magnitude and often change its on-off timing. The third burst will gradually disappear, and for moderate-to-fast speeds the neurocontrol becomes "bi-phasic." For moderate-speed movements, the agonist and antagonist bursts continue to scale down in magnitude, and the antagonist may even end before the movement is completed. With moderate-to-slow movements, at some stage the antagonist clamping burst is no longer mechanically needed and the entire movement (and its magnitude) becomes governed by a "pulse-step" strategy of the agonist. Thus the shape of the neurocontrol signals differs drmatically between a tracking movement performed quickly and performed at a moderate-to-slow pace; but the strategy is very much that which would be predicted based on analysis of biomechncial factors. To some extent the brain "learns" mechanics.

To scale movement magnitude during fast movements, say for a larger-magnitude movement, the pulse height (magnitude) of the initial agonist burst is adjusted scaled up (unless it is already satuarated at maximum neuromotor drive), causing a higher acceleration and velocity during the early phase of movement. Often the pulse width is also increased. Researchers have also studied other scaling phenomena, such as with added inertia (e.g., balls of different sized in the hand). In all cases, the neuromotor solutions make biomechanical sense.

Of note is that the above neuromotor scaling strategies assume that the neuromotor system has perfectly learned mechanics. This is typically close to the case. However, in reality the human does "many activities well" yet none perfectly - even NBA basketball players routinely miss free throws. The more challenging the task and the greater the mechanical sensitivity to internal neurocontrol "noise" or error in formulating these neurocontrol signals, the more likely that accuracy and performance will suffer. One form of this reality is the classic "speed-accuracy trade-off" that is captured under what is called "Fitt's Law" - faster-speed movements generally will exhibit greater target error. From a rehabilitation perspective, a key point is that there is a tremendous degree of neural-mechanical coupling in goal-directed performance. Thus your biomechanical background is very relevant to understanding and addressing the role of functional impairments in goal-directed performance and skill acquisition.

| gift-touching | movement dynamics | movement synergies | performance | disability/access |