Within Grasp

David Pacchioli
January 01, 2000
2 hands gripping rope in tug-o’-war position
James Collins

It's morning—already—and you're sitting in front of your terminal, struggling to remember just what it is you're supposed to be doing. Your trusty mug of coffee is steaming away at your side. How much do you have to think about what happens next?

Hand reaches out, thumb and fingers encircle the handle, grip tightens, hand raises mug toward lips. . . . Oops. Better get the paper towels. Maybe you should have thought about it a little bit more.

Grasping—the ability to take and hold objects in our hands—is one of the hallmarks of being human. It is something we do better than any other creature (after we've had our morning coffee). Our manual dexterity is something that is more important to us, more empowering, than almost any other physical ability we possess. It is also the sort of fundamental skill we tend to take for granted.

That's understandable. Grasping is learned very early: most babies can curl their hand around an adult's little finger by the age of four months, and by ten months can daintily take a piece of cracker in thumb and forefinger. After we master these basics, it seems, we tend not to think much about grasping until something goes wrong—our grasp is limited by injury or old age—or until we call on our hands to do something they haven't done very often, like picking up a pen in the offhand and trying to write our name.

There's a lot involved, however, in even the simplest grip; that's why modeling the hand is so difficult a task for a robotics expert. At the most basic level, any motor task requires a sequence of preliminary events: a certain level of attention in the central nervous system, an external stimulus, a decision to respond, the selection of an appropriate response, and initiation of that response. That's all before the muscles get to carry out the movement. Practice makes perfect, whether you're wielding a paintbrush or throwing a curveball. But here's something else to chew on: No matter how long you practice, you never do even the simplest motor task exactly the same way twice.

There are many more muscles than joints in our fingers, for one thing. This anatomical redundancy greatly enriches our potential for movement, increases our options, allows us to express incredible versatility and intelligence through our hands. But this same built-in complexity makes the hand and its workings very hard for scientists to understand. What exactly are the controls on such marvelous dexterity? What are the specific muscles and coordination of muscles required to grasp and draw back and release a bowstring, or pick up a quarter, or unscrew the lid from a peanut butter jar? What are the nerve signals involved? And how exactly are these tasks and a million others "organized" by the brain?

hand holding pick to strum guitar
James Collins

John Napier, the British evolutionary biologist who pioneered the modern study of the hand back in the 1950s, simplified matters considerably by identifying two basic types of human grip: precision and power, he called them, based on the number of fingers involved. Researchers in the field have worked according to Napier's grouping ever since.

Today, however, kinesiologists are at last beginning to push beyond Napier's categories, to consider more complicated, and realistic, approximations of grasping functions, and a more nuanced understanding of what controls them. They are asking new questions, and to answer them they are using a remarkable variety of approaches and techniques.

Robert Gregory's long-held obsession is what he calls the "pure physics" of human movement.

"Even as a kid I had this interest," he says. "My parents used to say ëYou'll never get a job studying how muscles work.'" Gregory accordingly majored in premed as an undergraduate at Notre Dame, but, "When I was looking at med schools I came across some information on biomechanics programs. I knew that was it."

At Penn State, where he recently completed his Ph.D. in kinesiology, Gregory worked with Vladimir Zatsiorsky, professor of kinesiology and director of the Biomechanics Lab, located in the basement of Rec Hall.

"Basically," Gregory says of the lab in general, "we're looking at variability in human performance, why it occurs." Not just variability among individuals, he explains, but within individuals, too. "Look at a world-class long jumper like Carl Lewis," he says. "Why does he sometimes jump 30 feet, the record, a supermaximal performance, and sometimes only 27? Why can't he do the same thing the same way every time?"

Making sense of variability, for the biomechanist, involves understanding as completely as possible all of the body's working parts. "Our main focus right now," Gregory says, "is the function of the hand." Specifically, he and Zatsiorsky, often collaborating with professor of kinesiology Mark Latash, have developed a pair of experimental devices for measuring the force produced by the individual fingers as they encounter various grasping tasks. The first of these devices, which Gregory pauses to demonstrate, is an upside-down-T-shaped aluminum contraption fittingly referred to as the gallows. "You clamp your hand in here," he says, placing his hand between two bars at the base of the T and tightening them across his bottom finger joints. "Then you put your fingers in these"—rubber loops hanging from a bar at the top. At a signal, the subject presses down with the suspended digits, and a transducer wired to the loops measures the amount of force exerted on each one.

In a study he presented at Penn State's 1999 Graduate Exhibition, Gregory used the gallows device to separate the effects of two distinct types of force variability. He instructed his subjects to press as hard as they could against the loops, using all four fingers, for a total of 30 trials. By comparing the variations in total force produced against the variations in force exerted by each finger, he isolated performance variability from compensated variability. The first, he says, "is associated with the intensity of the central command from the brain." Here, the amount of exertion remains evenly distributed among fingers; only the total changes. "Variation is due to a stronger or weaker intensity of that central signal." Compensated variability, on the other hand, is when total force output stays constant, but the contributions of the individual fingers vary. "With compensated variability, the command arrives fine at the hand, but then it gets shunted more to one finger than another."

To better approximate "real-life" grasping tasks, Gregory uses another device, which consists of a metal hand-grip attached to the middle of a long bar, not unlike a carpenter's level. With forearm resting on a table, hand out over the floor, the subject holds onto the grip while Gregory hangs weights at different places along the bar to make it tip in one direction or the other. "The subject is asked to right the bar, to bring it level," he explains, and very sophisticated three-dimensional transducers on the finger and thumb pads of the grip precisely measure the forces involved. "Again," Gregory says, "we're looking at how the force required is distributed among the fingers, and now also at how this distribution changes with different turning stresses on the wrist."

blond head; hand in position to throw football
James Collins

To measure the activity of individual muscles during these bar-balancing trials, Gregory uses electromyography—a technique similar to that used in stress tests. "For this experiment," he says, "I put one set of electrodes on the inside of the forearm, to measure the collective activity of the four finger flexors used in grasping; a second set on the back of the forearm, to get the finger extensors; a third on the thumb flexor; and a fourth on the outside of the pinky."

The devilish element in this whole set-up, as Gregory explains, is that it pits the brain against itself: "Our main goal is to understand how the central nervous system deals with two opposing tasks—grasping the grip hard enough not to drop it, while at the same time trying to keep the bar level."

Paola Cesari's questions about the hand, and the tools she uses to answer them, are of a very different order. Cesari's specialty is motor control: her focus is less on muscles and more on behaviors—and what close observation of those behaviors, aided by mathematical analysis, can tell about the workings of the brain in connection with the body. After earning her Ph.D. from Penn State in May 1999, Cesari went back to her native Italy, and a research position at the University of Trento. In August, she returned to University Park for an international conference on motor control. Her former adviser, Penn State professor of kinesiology Karl Newell, was one of the conference organizers.

Cesari is particularly interested in what she calls grip configurations, that is, the ways in which we grasp different objects. What determines these choices, which happen so quickly and seem so automatic? What information do we need to enable such decisions?

"What makes this difficult to study is that the hand has so many degrees of freedom," Cesari says. "We can accomplish grasping in so many different ways." As a result, she says, most of the studies reported up to now have limited themselves to examining what John Napier identified as "precision" grips: those involving only the forefinger and thumb. "We wanted to test grasping in a more natural way."

Cesari opens a tall metal cabinet neatly stacked with cubes of various colors, ordered by size: from big Jack-in-the-Box-sized blocks on the bottom shelf to cubes as small as dice up near the top. For her thesis, she used 62 of these cubes, spanning the whole range of sizes and made of different materials, light to heavy, from balsa to cork to brass. The differences in size and mass were intended to elicit different grip configurations when Cesari asked each of her subjects, while seated at a table, to pick up cubes from one spot and move them, as quickly and accurately as possible, to another marked spot. Cesari videotaped these efforts, then went back and counted the number of fingers making contact with each cube for each move.

There are standard grip configurations, she had hypothesized, relatable to the size and mass of the object to be picked up. But she wanted to see when exactly, at what size and weight of cube, those configurations would change—from two fingers to three, say, or from three to four. And were those changes predictable by anything other than the features of the object being grasped?

Cesari chose her subjects for variety of hand size; in the first trial, she tested adults ranging in size from tres petite to extra large, and then she repeated the experiment with children, and also with small men and large women, to account for any possible gender effects. What she found was that when she adjusted for body size, all of her subjects used the same grips for the same tasks, and all changed grips at the same (relatively speaking) cube size. Using a technique called dimensional analysis, she devised an equation that expressed this "dynamic scaling factor."

Her results, Cesari says, suggest something important about how the brain processes information. "Given that the hand-grip configuration is organized prior to contact with the object," she writes, "these findings suggest that information about the dynamic scaling relation"—that is, the relationship between hand size and object size—"is picked up visually," and that this information alone is enough to determine the appropriate grip. In other words, Cesari argues, "All the brain needs in order to make a decision about which grip to use is these four pieces of information: the size and mass of the object, and the size and geometry of the hand it is controlling.

"What it means is that perception and action are one coupled process," she says. "The brain is not calculating this complex motion step by step, considering hundreds of individual variables. There's no time for all that. It's more like what an airplane pilot does when he's coming in for a landing—he picks up an optical flow," an overall visual pattern, much of which he has encountered before, parts of which are slightly different. Perceiving a pattern "decreases the number of variables to a manageable level by coupling them together," with the result that making a decision takes far less mental work. "That's why," says Cesari, "the simple motions that are so complicated for any biomechanist or any robotics person to map out are motions that a small child can do very well."

one hand holding juicer; one hand juicing lemon
James Collins

Behavioral work like what Paola is doing makes an inference about the activity of the brain," Matt Rearick says. "I'm trying to measure that activity directly." Rearick works in the Psychophysiology of Movement Lab, also in the basement of Rec Hall. The lab's director is his adviser, assistant professor of kinesiology Semyon Slobounov.

"One of our specialties here," Rearick says, "is EEG and movement." Electroencephalograms, he explains, provide electrical signals from different regions of the brain, an indication of the location and intensity of brain activity. Controlled experiments can link those signals to specific body movements.

At the moment, Slobounov is preparing a young woman for an EEG tracing. "You," he says brightly, with a strong Russian accent, "are the guinea pig today." He fits her head with what looks like a red cloth bathing cap, covered with white circular patches. Under each patch is an electrode, wired to a nearby computer. Next door, its open frame almost filling what looks like a large walk-in closet, is a cage made entirely of copper, containing a desk and chair.

"It's called a Faraday cage," Rearick explains. "EEG signals are very small. Any vibration can impede the signal. Somebody bouncing a ball upstairs in the gym could throw things off." The copper barrier, he says, will dissipate such interference.

Once the subject is in the cage and the closet door is closed behind her, Slobounov sits down at the terminal, ready to monitor the electrical signals streaming from her brain as she completes a prescribed set of movements. "Today she's doing a finger task," Rearick whispers as the trial begins. Soon he points to a dip in the linear readout. "There she blinked," he says. "We have to do about 50 trials to account for any noise like this, then superimpose them. That's called increasing your signal-to-noise ratio."

For his dissertation, Rearick is using the same set-up to investigate the brain's activity during a series of simple grasping tasks. "I'm looking at slow-wave potential," he says, "which is a very slow negative rise of electrical activity." The region he's targeting is the motor cortex.

From a nearby desk drawer he produces a grip dynamometer: two aluminum bars joined by a short cross piece and fitted with force transducers. The task he sets his EEG-wired subjects, Rearick explains, is to squeeze the grip for five seconds—in one trial using only two fingers and thumb, and then again with all five digits. For each grip configuration, they are to squeeze first at ten percent of their maximum force, then at fifty percent, as determined by a target line displayed on a computer screen in front of them. "Once they get to the target line," Rearick says, "we're looking to see how well they can control the force trace, to keep it exactly on that line. How accurate they are—and what the brain is doing when they're accurate as versus when they're not."

His findings so far have been something of a surprise. "In the five-finger configuration," he reports, "we found that people were more accurate at fifty percent force than at ten percent. No one's really sure why, but it looks like a situation of overkill—like the difference between picking up a textbook with all five fingers and picking up a coffee can lid."

In terms of brain activity, Rearick has found, "across the board, the ten-percent tasks required more cortical excitation than the fifty-percent tasks." This result runs counter to most of the existing literature, which correlates higher slow-wave activity with greater use of force. Instead, Rearick says, "I found that controlling the grip while exerting a lower force level is a more difficult problem for the brain to solve.

"I'm leery about drawing too many conclusions yet," he quickly adds. "This study is still in its infancy." Over the next couple of years, he says, he plans to look at still lower force levels: "Do they require even more brain activity?" And also at higher levels, "say eighty percent, to see what happens there. Is that harder? Is it easier? Going from past research, it should be harder, which would tell us that medium-force grips tend to be the least variable, and to require the least cortical resources to produce. It might tell us something about why we use a particular grip."

All of Rearick's subjects so far, he notes, have been college age. "Down the road I'd like to look at other populations," he says, "and see what that might tell us about developmental psychology. Do the very young and the very old have different responses? Are they the same in terms of their differences, as work in other areas suggests? Are development and degeneration a mirror image?"

Part of the program at the conference that brought Paola Cesari back to University Park last August was a "mini-symposium" on hand and finger control: a two-hour session of brief talks by kinesiologists from around the world on issues related to grasping. One by one, a dozen researchers stood and presented their work. Once again, the variety of approaches they presented suggested the complexity of the subject, and the size of the task they are engaged in.

Francisco Valero-Cuevas of Stanford, who organized the session, showed pictures of disembodied cadaver hands, wires running from their tendons: just a small part of his effort to understand the mind-boggling degree of muscle coordination that exists between our fingers.

Jose Gomez of the University of Minnesota trained three Rhesus monkeys to reach out and grasp 16 objects of varying shapes and sizes. Electrodes placed in the monkeys' brains, he reported, showed that the firing patterns of cortical neurons changed significantly with the shapes and sizes of the objects grasped.

At the University of Osaka, Hiroshi Kinoshita told the audience, he is using positron emission tomography (PET scans) to image brain function during grasping. So far he has identified seven brain regions that are activated when subjects lift a small object between thumb and forefinger.

In one of his recent experiments, Marco Santello, another University of Minnesota researcher, asked subjects wearing a sensor-laden cyberglove to pretend they were grasping objects including an ashtray, a zipper, a frying pan, and a computer mouse, so that he could precisely measure hand "postures" used in various activities. He concludes that in addition to a set of basic postures, there appears to be a separate fine control mechanism for making small adjustments in fingertip force.

M. H. Schieber of the University of Rochester also uses a cyberglove, he reported, but he does so in combination with a video camera. Schieber's focus is on independence: the freedom, or lack thereof, of our fingers to move around all by themselves. The thumb and forefinger, he acknowledged, "are very highly individuated"; but try wiggling your middle finger, he challenged his audience, and keeping the other fingers completely still.

Part of the problem, Schieber asserted, is passive mechanical coupling. "Some musicians go so far as to have surgery to cut some of the tendons involved," he said, adding, "This doesn't seem to work well." But maybe, he went on to suggest, there's something else at work, too:

"The nervous system that controls the hand has evolved, all the way from when the hand was used as a fin to today's sophisticated uses," he said. "Maybe it just hasn't yet got to the point of completely individual digits."

Robert W. Gregory received his Ph.D. in kinesiology in fall 1999. He currently holds a postdoctoral fellowship at the University of Michigan. His adviser at Penn State, Vladimir M. Zatsiorsky, Ph.D., is professor of kinesiology and director of the Biomechanics Laboratory in the College of Health and Human Development, 267F Recreation Building, University Park, PA 16802; 814-863-3772; vxz1@psu.edu. Paola Cesari received her Ph.D. in kinesiology in May 1999. She is currently a researcher at ECUS University in Trento, Italy. Her adviser at Penn State, Karl M. Newell, Ph.D., is professor and head of the department of kinesiology, 146 Recreation Bldg.; 863-1163; kmn1@psu.edu. Matthew P. Rearick is a Ph.D. student in kinesiology, 266 Recreation Bldg.; 863-5608; mpr133@psu.edu. His adviser, Semyon Slobounov, Ph.D., is assistant professor of kinesiology and director of the Psychophysiology of Movement Laboratory, 267E Recreation Bldg.; 865-3146; sms18@psu.edu. The international conference, "Progress in Motor Control II: Structure-Function Relations in Voluntary Movement," was held at Penn State, August 1922, 1999.

Last Updated January 01, 2000