September 27, 2004 —
Drum the tip of a finger on a typewriter key quickly: “eeeeee.” Now, stop and
type “e,” take a moment, type “e,” take another moment, type “e” again. The motion
in both cases is exactly the same, performed by the same finger.
But according to a study done by a University of Southern California neural specialist
and colleagues, the brain processes that make them happen are utterly different.
HE'S GOT RHYTHM: A FMRI scan measures brain activiity as an experimental subject
flexes his wrist, either in a continuous rhythmic movement, or in individual steps,
each flex orchestrated up or down. with a pause after each action.
The insight may lead to better movement control by humanoid robots, new ways
of movement rehabilitation for people and it even offers insight into the effect
Stefan Schaal, associate professor of computer science at the USC Viterbi School
of Engineering led the international team that used functional Magnetic Resonance
Imaging (fMRI) scans to test a longstanding question regarding “rhythmic” versus
“Rhythmic movements like walking, chewing or scratching are found in many organisms,
ranging from insects to primates,” notes Schaal in an article published in Nature
Neuroscience Sept. 26. “In contrast, discrete movements like reaching and kicking
are behaviors that have reached sophistication in young species, particularly
Schaal is a robotics expert with a deep background in neuroscience who draws
inspiration for robot controls from biological models. He notes that researchers
have historically treated the two different kinds of movement as fundamentally
the same in terms of control, assuming that one is a special form of the other.
Thus specialists studying discrete movement have considered rhythm a subset of
discrete movement – the same thing speeded up and repeated. But behaviorists
studying rhythmic movement like walking have considered discrete movement as the
same thing slowed and aborted after only a single act of rhythmic movement.
In a carefully arranged set of experiments, Schaal and co-workers from Pennsylvania
State, and ATR Computational Neuroscience Laboratories in Kyoto, Japan showed
that control mechanisms for the two types of movement are drastically distinct.
THE COLORED AREAS show the differences -- that is the changes - in brain area activity in the
three diffeernt states measured: rhythmic movement, discrete movement, and rest
(see text for definitions). The blue area, for example, shows the most active
areas after a switch from discrete to rhythmic activity; the green the reverse.
If the two types of movement were controlled by the brain in the same way, these
changes would not be as marked.
The study monitored 11 volunteer subjects, who performed a simple flex of the
wrist while undergoing fMRI monitoring. A visual signal instructed the subjects
to do one of three actions: rhythm – flexing the wrist repeatedly at a comfortable
pace, back and forth; discrete – flexing the wrist, pausing, flexing it back,
and rest. Another set of experiments had the timing of the rhythm dictated to
the subjects by a metronome.
The resulting fMRI records displayed far-reaching differences. Rhythmic activity
created activity only in the motor areas of the opposite brain hemisphere and
Discrete activity was much more extensive, including numerous areas on both sides
of the brain, and including “planning areas” not directly connected with motor
The difference held up even when careful controls made sure that the amount of
actual activity – the number of up and down flexes, and their velocity – was the
“We believe that these results provide strong evidence to refute the hypothesis
that rhythmic movement is generated with the help of the discrete movement system,”
the authors wrote. However, the opposite is not the case: the authors found that
“discrete movement could indeed be generated with the help of the rhythmic movement
“What our results indicate is that we really deal with two very separate systems
in movement,” says Schaal. “There is an automatic system that, literally, functions
without any thought; and a separate cognitive system that orchestrates more complex
And music? “Computational neuroscientists theorize that rhythmic movements are
generated from oscillator circuits in the brain, and it may be that these inherently
rhythmic neural systems make it to easy for us to swing to the rhythm of music,”
said the scientist.
Meanwhile, Schaal and his colleagues are working on converting their results
to humanoid robot algorithms that capture such behavior, and which could give
future robots a bit more rhythm in their stride, “but don’t look for them right
away in hiphop videos,” he says.
Dagmar Sternad of the Penn State department of Kinesiology and Rieko Osu and
Mitsuo Kawato of ATR-Kyoto were co-authors of the study, which was funded by the
National Science Foundation, Japanese Science and Technology Agency, and ATR.