How the Brain Implements Movement
PASADENA, Calif. — We give little thought to reaching for an object. Should something be in the way, we simply use our arm to go over or around whatever it is to grab whatever we want.
If you think about it, says Elizabeth Torres, a postdoctoral scholar in biology at the California Institute of Technology, it's a remarkable talent. The simple act of reaching and grasping involves a cascade of rapid-fire events. First, one of our senses, vision, eyeballs the object's location and distance. Next, the brain determines what "degrees of freedom" of the limb are needed to grab it--what angle of the elbow, what rotation of the shoulder, how far to turn the wrist and bend the fingers. Finally comes the determination of the appropriate speed for extending the arm, then stopping precisely at the object's location.
But what's long puzzled neuroscientists like Torres is how the brain translates two different "languages" into action: the sensory cues that come from our vision system--which feeds the brain information about the location, orientation, and shape of the object of desire--and the motor execution, the movement of bone and tissue that actually grabs it.
The answer? Geometry, suggests Torres. A transitional, geometric stage between sensory perception and motor actions, in which the brain simulates a task such as grasping an object without actually moving the arm. Her research, which provides further insights into the marvelously complicated workings of the brain, has been published in the May issue of the Journal of Applied Physiology and was selected as a "Highlighted Topic" by the journal.
The conventional wisdom for solving such goal-directed behavior as reaching and grasping has relied on the assumption that the brain has exquisite timing, knowing exactly how long any movement will take before it initiates it. Reaching for a glass of water a foot away takes a certain amount of time at an appropriate speed, a speed different from (way slower than) the speed needed to rescue a glass that's about to tip over. That assumption about timing came from prior biological motor-control studies, and made sense, says Torres, because the data has always come from experiments that use very simple, straightforward reaching movements. In humans, such simple movements are mastered at about the age of seven months. Indeed, such motions become so automatic after a while that the timing for each repetition of the same movement is consistent down to the millisecond.
However, the way in which more complex movements evolve as the system learns to perform them is a different story. In everyday life we are faced with the need to solve new motions all the time. For instance, the familiar and simple task of opening the door to our home requires a new motion if we are carrying a tall stack of books that blocks our view of the door's lock. In that case, the old strategy that would elicit an automatic movement with an exact duration would fail us under the new circumstances.
Torres's approach ignores timing, proposing that during learning, the spatial and the temporal components of the motion can be separated.
In her view, the brain first determines a path that, in the case of arm movements, leads the hand along the most optimal direction to solve the task geometrically in a given context. This means the brain takes into consideration the relationships of other objects to the goal (the telephone in front of the water glass), and the amount of physical space the arm will have to work in. This geometric path is special in that it can be traversed with different timing and still be the same path. For instance, the straight line between two points in space creates a path that is the shortest distance joining them. This path is unique in that no other path qualifies as the shortest one between these two points. Thus, no matter how one travels along it, fast, slow, or slowing down and speeding up, it is still the shortest path.
Only later, says Torres, does the brain learn how to adjust the timing to best traverse the path. This is analogous to the way in which a dancer learns the choreography of a particular dance, or a musician learns to perform a piece. First, they master the motion that gives the desired outcome, and then they adjust the timing of this motion to bring the movement to perfection. Eventually, they become so proficient at it that they can put emotion into it, thus reaching a new level of performance.
More recently, behavioral data collected in the lab of Caltech's Richard Andersen, the James G. Boswell Professor of Neuroscience, lends support to Torres's theory. The data shows how nonhuman primates learn to avoid obstacles. Indeed there is a clear separation between learning or forming a geometric solution to a particular object, and that solution becoming automatic. Learning consists of forming a geometric strategy, the best path to use in space with respect to a particular set of goals. It becomes an automatic movement only later, when the best time profile is uncovered that best traverses that particular path.
Torres says that without the electrophysiological data from the brain, it's hard to confirm whether this theoretical idea makes sense as a brain solution for learning goal-oriented behaviors. Thus her current research in Andersen's lab focuses on the monitoring of learning and adaptation as they evolve in the non-human primate, in a part of the brain known as the posterior parietal area.
"Dr. Torres is a truly remarkable scientist," says Andersen. "As a graduate student at UCSD she developed a completely novel geometric theory of motor control. She is now testing her theory with experiments here at Caltech and already has results that support her ideas."
"As in a game of chess, through action simulation one can anticipate various outcomes before making a definite move," adds Torres. "Finding evidence in support of this geometric stage, where actions are simulated before actually being executed, would be a first step toward understanding how new motor skills may be acquired all the time."
