Bimanual coordination

Bimanual coordination:
electrophysiological and psychophysical study.
Thesis for the degree
“ Doctor of Philosophy”
By
Anna Gribova
Submitted to the senate of the Hebrew University in Jerusalem
December 2001
This work was carried out under the supervision of
Prof. Eilon Vaadia
Prof Hagai Bergman
Acknowledgments
I would like to thank my supervisor prof. Eilon Vaadia for the long way we went
together, for encouraging me to find and express my own ideas during my PhD work and
for his all time available support and supervision during my work.
I would like to thank my second supervisor prof. Hagai Bergman for very helpful advises
and support during my work.
I would like to thank all my collaborators, especially Simone Cardoso de Oliviera and
Opher Donchin for sharing the years of experimental work.
I would like to thank my parents, especially my mother for giving me support and
strength to believe in myself.
And finally, I would like to thank my Yoga Teachers: Andre Siderski, Shandor Remete,
Zipi Negev and Orit Sen-Gupta who help me to go my own way through many others.
I thank Luxemburg foundation for the price I got. I thank Mifal ha-Paise foundation by
M.Landau for the fellowship during the last year of my research.
I thank Alexandra Mahler and Gil Alroy for the help in editing this work.
Bimanual coordination:
electrophysiological and psychophysical study.
ABSTRACT
This work combines psychophysical studies in humans and monkeys with neural
recordings from monkeys, in an attempt to study the mechanisms of coordination and
execution of bimanual movements. The results of our study suggest that bimanual
movement is a different entity from unimanual movement. The notion that bimanual
movement is planned as a whole may explain the “bimanual coupling” paradigm and
“bimanual specific activity”, found previously in our lab. In addition, it leads to a
more holistic view of motor planning problem, its preferences and restrictions.
Using the non-structured scribbling task in humans, we demonstrated that both hands
are naturally coupled when special restrictions are not imposed. We advanced one
step further by showing that non-symmetric bimanual movement is unexpectedly hard
to perform and requires special learning. These simple experiments revealed the
existence of a common plan for both hands.
In the center-out task in both humans and monkeys we continued to study the
dynamics of inter-arm coupling and found that it is high even before movement starts,
and that it gradually decreases towards the end of the movement. This type of
dynamics supports the notion that according to the brain original blueprint the two
arms are coupled and that decoupling occurs due to a separate proprio-receptive
negative feedback to each arm. As a possible neural correlate to this phenomenon, we
found in monkeys that the LFP cross-correlations are high before movement starts and
decrease as the movement is performed. The time scales for the inter-arm coupling
decrease and the LFP cross-correlation decrease were very similar.
In the same center-out task we compared bimanual performance with the unimanual
movement from which it is composed and showed that the bimanual movement does
not have a higher reaction time (RT) then the unimanual one. Furthermore we
demonstrated in monkeys that the preparatory neural activity is significantly longer
for unimanual non-dominant movements and equal for the bimanual and unimanual
dominant movements. These findings support the notion that bimanual movement is
not a combination of unimanual ones.
Supplementary motor area (SMA) lesion study in monkeys revealed some transient
changes in behavior; followed by some compensatory changes in primary motor
cortex (MI). We found that MI activity in the absence of SMA is involved in the
control of both sides of the body. The present study indicates that each hemisphere
has information about both arms. Under normal conditions, this control is most likely
obscured by the inhibition imposed by SMA on MI.
A comparison of right and left MI reveals that there are functional differences
between the two sides of the brain. These differences are small and further research
with a larger monkey database is required to substantiate our conclusions.
Contents:
Introduction
1. General introduction and working hypothesis
Prolog………………………………………………………………….1-4
Scientific background…………………………………………………4-10
Working hypothesis…………………………………………………..10-13
2. The interconnection between the various parts of this work
….....14-15
Experiments and results
Chapter A: Psychophysics of Scribbling
1. Introduction ………………………………………………………….16-22
2. Methods ……………………………………………………………...17-18
3. Results ………………………………………………………………...18-22
4. Intermediate summary, discussion and conclusions …………………..22-24
Chapter B: Psychophysiology article (in preparation)
Chapter C: SMA lesion in two monkeys
1. Introduction ………………………………………………………………25
2. Methods ………………………………………………………………..25-35
3. Results …………………………………………………………………35-46
4. Intermediate summary, discussion and conclusions
…………………….46-49
Chapter D: Comparison of right and left arm-related MI in four monkeys
1. Introduction ……………………………………………………………..49-50
2. Methods …………………………………………………………………50-53
3. Results
………………………………………………………………….54-59
4. Intermediate summary, discussion and conclusions ……………………..59-60
Chapter E: General conclusions and discussion:
1. Overall summary and conclusion ……………………………………61-64
2. Further suggestions for implementation of this research
………….64-66
3. Appendix 0: Protocol of writing experiments.
Reference list.
………………………………………………………..67-77
Appendix: Additional publications
Introduction
Section 1: General introduction and working hypothesis.
Prolog.
On the course of this work, emphasis was placed on combining psychophysical and
physiological approaches in order to study the mechanisms involved in the
coordination and execution of bimanual movements.
Natural movements almost always require the coordination of different limbs. For
example, when one observes how the two arms move, one can see that there is a
natural tendency to move both arms in some correlated fashion. Yet, we can clearly
operate each arm alone. Let us consider the following example. If you are a righthanded person, try to write a word using the left hand; now do it using both hands as
a simultaneous drawing. Which of the two tasks is easier? For most subjects, the
second is easier. In general, tasks in which the two hands move in a similar way are
easier. For example, if you try to write using your dominant hand, while the other
hand is typing the same text on the keyboard – you will find that this is an impossible
task.
Arm movements can be divided to three categories:
1) Unimanual movements where only one of the arms moves. Some movements
are typically unimanual: e.g. writing, drawing.
2) Symmetric bimanual movements where both arms move together or in opposite
directions around a plane of symmetry. Examples of symmetric limb movements
are hand clapping, walking, hand movements associated with speech.
1
3) Non-symmetric bimanual movements where the two hands move in a nonsymmetric fashion, e.g.: tying shoelaces, playing the piano or violin, some kinds
of dancing.
If one executes a movement, which involves more than one limb – one actually
performs at least two movements. Thus: inter-limb-coordination may be considered
as generation of a “unitary motor action” from elements of several movements. Taking
the example of bimanual movements, it is common knowledge that the more different
the movement executed simultaneously by the two hands, the more difficult it is. For
example, it is very easy to draw a triangle or a circle using one hand. But, to draw
simultaneously a triangle with one hand and a circle with the other is very difficult.
Yet, to draw two circles (or triangles) simultaneously, using both hands is, again, an
easy task. These simple observations reveal some of the restrictions and preferences of
the motor system. This shows that there are different types of bimanual movements
and that they are not merely the sum of two unimanual movements. All these
observations should be included in the concept of coordination.
To execute any movement, the motor system has to compute desired values for each
point in time for each muscle, joint and limb. This complex computation is, no doubt,
a time-consuming process. Therefore, it also reasonable that the nervous system
contain internal representation of compound bimanual movements as unitary motor
actions.
An important feature of the system is the redundancy in the degrees of freedom.
Namely, each movement can be executed in several different ways. Bernstein defined
motor coordination as “the process of mastering redundant degrees of freedom of the
moving organ, in other words its conversion to a controllable system ” (Bernstein,
1967,1990). In other words, this mastery means finding some optimal time-efficient
algorithms to control high degree of freedom (DF) system.
This “controllable system” is a real time computational one. Therefore, minimization
of the time required for each computation during the movement is of prime
2
importance. Let us assume that “easy” motor tasks (like symmetrical bimanual
movements) are less demanding computationally, while harder tasks requires more
complex, time-consuming computation. To apply this approach to definitions of
difficulty for “bimanual coordination,” we may define coordination that require less
computations – as “easier.” Looking at the same issue in terms of DF, we may describe
“difficulty” as follows: The number of DF of a movement proportional to the number
of joints associated with it. In relation to bimanual coordination, we introduce the
term: “Functional DF (FDF),” which is the actual number of independent uses of the
joints while a certain movement is executed. This is to indicate that the system may be
designed to reduce the actual number of DF while executing a certain movement.
Note that the number of independently moving joints is always equal to or less than
the anatomical number of joints in the given limb (DF ≥ FDF). The difficulty in
bimanual coordination is, therefore, proportional to the level of interlimb
independence in the movement: namely, it is proportional to the FDF. In summary, I
suggest that the level of difficulty of interlimb coordination reflects the need to
increase the number of controlled degrees of freedom in order to execute the
particular movement. To demonstrate the interesting impact of such an approach, let us
briefly consider an example from robotics.
What are the differences between robots and humans from the mechanical point of
view? The most striking difference is the perfectly smooth and very rapid
performance of multi-joint and multi-limb manipulations by humans as compared
with robots. The main advantage of the real brain is probably its ability to coordinate
the joints in time and in space in a very efficient mode. The timing problem is readily
solved, but the space problem (where to place the joint) remains a restricting factor in
modern robots. To generate a finer movement by a robot (particularly in space) – we
need to increase the number of DF. The result: The cost in computation demands rise
exponentially (!). These computational problems limit the possible number of joints in
all artificial systems. In simplistic terms, one could say that the robots are limited by
the fact that the number of FDF is always equal to the DF (the sum of the degrees of
3
freedom of the joints). This, I believe, is the main advantage of the living brain. It can
reduce the number of DF to number of FDF according to functional demands. This
means that our brain does not have to continuously calculate the coordinates for each
joint, but only for the independent joints. Therefore, we find that human (and other
animals) can easily coordinate joints and limbs on a very short time scale. This notion
serves as a guideline throughout this study.
Scientific Background
Psychophysics.
Whenever we perform simultaneous movement of multiple body parts, their
movement has to be coordinated in some way. One straightforward suggestion could
be that isolated movements, e.g. of one arm, are easier to produce than bimanual
movements. If true, movement coordination requires an extra process and it should be
reflected in greater processing time. Indeed, some studies have reported that bimanual
movement requires longer reaction times then unimanual ones (Kelso et. al., 1979,
Garry et. al., 2000). Other studies, however, did not confirm this hypothesis, since
they observed this effect only in certain bimanual tasks (Aglioti et. al., 1993, Anson
et. al., 1993).
Studies of interlimb coupling in 5-12-month-old infants (Fagard et. al., 1997, Bresson
et. al., 1977, Corbetta et. al., 1996, Gesell et. al., 1947) demonstrated that although
there are fluctuations between uni- and bimanual reaching during the first year of life,
frequent bimanual reaching (5-6 months, end of the first year) alternate with periods
of frequent unimanual reaching (7-9 months).
Yomanishi et. al. (1980) first found that two coordination patterns of arm movements,
corresponding to in-phase (symmetric) and anti-phase (opposite) were “preferred” by
the motor system. This preference is expressed in two ways: first, movement
variability is lower, and second, other phasing patterns tend to shift toward in-phase or
4
anti-phase. Later, other studies revealed strong tendency to produce spatially (Franz
et. al., 1996,2001, Swinnen et. al., 1997, Bogaerts et. al., 2001) and temporally
coupled arm movements (Franz et. al., 1997, Eliassen et. al., 2000, Kelso et. al., 1984,
Semjen et. al., 1995, Byblow et. al., 1998, 2000, Walter & Swinnen 1990, Swinnen et.
al., 1996, 1997 a&b), supporting the hypothesis of unified, rather then separate,
coding of bimanual movement.
Cunningham et. al. (1989) studied the human psychophysics of movement trajectories
in different map transformations. They showed that a transformation of 180º has
significantly smaller effect on performance, in comparison with other transformations.
This may explain why human subjects make opposite movements as easily as
symmetrical ones.
Tuller and Kelso (1989) examined between-hand rhythmic movement patterns in
normal (musicians and nonmusicians) and split-brain subjects. Only two phase-locked
states, in-phase and anti-phase, were shown to be stable in all subjects. Split-brain
subjects showed an even greater tendency toward these patterns, thus refuting the
notion that reduced cortical interactions between the hemispheres allows independent
visuomotor control of the hands in such a task. However, in spatial coupling, interhemispheric interactions may play a more central role. Split-brain subjects are more
adapt then normal subject at decoupling spatial aspects of bimanual movements (Franz
et. al., 1996, Eleassen et. al., 2000).
It has been suggested that spatial coupling may be a simple byproduct of
biomechanical coupling of two arms, but two studies contradict this hypothesis. Franz
and Ramachandran (1998) showed that amputees with phantom limb effects produce
spatial coupling even though the amputated limb cannot move, and similar results
were obtained in normal subjects who move one arm, while imagining movement of
the other (Heuer et. al., 1998).
To summarize this section, there are two main contradictory hypotheses put forward
by psychologists, relating to bimanual motor planning:
5
According to one hypothesis there is a "coordinating schema" (originally suggested
for the interaction of reach and grasp, Hoff & Arbib, 1993), which coordinates
between two independent motor plans by rescaling the individual movements such
that they are temporally aligned. Proponents of this hypothesis claim it is conceivable
that the coordinative schema requires additional cognitive resources, and probably
also additional time to fulfill its task. Demonstration of longer reaction time for
bimanual as compared with unimanual movement could support this notion.
The other hypothesis is that bimanual movements are not generated on the basis of
two independent motor plans, but rather by a single common movement plan, a
"generalized motor program" (GMP) (Schmidt 1975, Schmidt et al., 1979). As a
variant of this notion, the term "coordinative structure" was coined for collectives
(synergies) of muscles that can be controlled jointly as a single functional unit
(Easton, 1972, Turvey, 1977). Like the GMP, the "coordinative structure" enhances
coding efficiency by reducing the number of separately controlled degrees of freedom
(Bernstein, 1967,1990, Turvey, 1977), and explains the tendency towards common
timing of bimanual movements. In this framework, a bimanual movement, similarly
to a unimanual movement, should be a motor pattern, and, therefore, no differences
between bimanual and unimanual movements should be expected.
Physiology.
Physiological studies of bimanual coordination mainly focused on two cortical areas.
One is the supplementary motor area (SMA) and the other is the primary motor area
(MI). MI is related to movement parameters such as force (Evarts et. al., 1983),
direction of movement (Gourgopolous et. al., 1982, 1995, and Schwartz et. al, 1988),
muscle activity (Scott et. al., 1997), velocity and acceleration (Fu et. al., 1995), and is
also involved in bimanual coordination (Donchin et. al., 1998, 2001 a&b, Kermadi et.
al., 1998). SMA is a secondary motor cortex thought to be associated with learned
non-automated complex motor tasks (Halsband et. al., 1993; Lang et. al., 1990,1988),
initiation and programming of sequences of complex movements (Brinkman and
6
Porter, 1979; Luppino et. al., 1991, Chen et. al., 1995), internally initiated movements
(Kurata and Wise, 1988; Halsband et. al., 1993), the control of posture (Wiesendanger
et. al., 1987; Viallet et. al., 1992) and bimanual movements (Wisendanger et. al.,
1994,96; Uhl et. al., 1996, Tanji et. al., 1988, Donchin et. al., 1998, 2001 a&b).
The SMA is reciprocally connected with its contralateral homotopic area, and
bilaterally, with the upper parts of area 6 and with MI (Karol and Pandya, 1971;
Pandya and Vignolo, 1971). The SMA also receives dense heterotopic callosal
projections from the contralateral rostral and caudal cingulate motor areas, moderate
projections from the lateral premotor cortex, and sparse projections from the MI.
The involvement of subcortical structures in bilateral coordination may also be
important, since the SMA is connected with a number of subcortical structures
involved in motor functions. These connections include bilateral ones with the
striatum, ipsilateral connection with the thalamus, red nucleus, pontine nuclei, and
dorsal column nuclei, and direct bilateral corticospinal projections (Jones et. al., 1977;
Kuypers and Lawrence, 1967; Porter, 1990; Luppino et. al., 1994).
Parkinson disease (PD) is associated with a loss of nigral dopaminergic neurons
projecting to the striatum, particularly to the dorsal putamen (Brooks et. al., 1990).
The major output of the putamen is to the SMA (Alexander et. al., 1990), via the
pallidum and ventrolateral thalamus. Disruption of the nigrostriatal projections may
be responsible for the development of akinesia in PD, as it results in functional
deafferentation of the SMA (Dick et. al. 1986, DeLong et. al 1990).
Indeed,
compared with normal subjects, PD patients exhibit a relatively reduced signal in the
SMA, as was shown in PET (Playford et. al., 1992, Rascol et. al., 1992) and fMRI
studies (Sabatini et. al., 2000).
It is known that PD patients experience difficulty in performing two different tasks
simultaneously (Brown et. al., 1993). Schwab et. al. (1954) reported that the ability of
PD patients to perform two simultaneous tasks, squeezing a bulb ergograth with one
hand and drawing triangles with the other hand was impaired. The similarity of
7
“mirror movement” symptoms caused by SMA lesion and Parkinson disease
symptoms reflects the tight connections between the basal ganglia and the SMA, and
further support the notion that coordination requires interactions between different
structures. This issue deserves further research.
All this suggests that bimanual coordination may involve complex interactions
between many brain areas, rather than localized activation of a single area. Further
support to this view was provided by lesion studies as described in the next section.
Lesion studies.
Brinkman et. al. (1981,1984) studied short-term and long-term behavioral effects in
unilateral lesions of the SMA. Two SMA-lesioned monkeys showed a characteristic
deficit of bimanual coordination in which the two hands tended to behave in a similar
manner, instead of sharing the task between them. Callosal section immediately
abolished the bimanual deficit, although the clumsiness returned transiently.
It is known that lesions in the SMA in humans (Viallet 1992; Chan et. al., 1988,
Shimamura et. al., 2001) and in monkeys (Brinkman et. al., 1981,84) cause a tendency
toward mirror movements of the two hands. Brinkman demonstrated that callosotomy
abolished the mirror movement tendency, suggesting that most of the hand synergy
networks pass through the corpus callosium.
Chan and Ross (1988) described a patient with a right-sided infarction involving the
SMA that developed mirror movements and had a particular difficulty in performing
nonsymmetrical movements, like threading a needle or climbing a rope. When he was
asked to pretend to drive a car, instead of moving both hands in the same circular
direction in order to turn the steering wheel, he moved them in an arc either up or
down towards each other. He recognized the error but was unable to perform the task
correctly. When asked to imitate the examiner, however, he could reproduce the
correct movement after some hesitation. This person was right-handed in writing, and
8
when he was asked to write with his left hand, the writing was always mirror-imaged.
Bimanual coordination is also impaired in chronic schizophrenia probably due to
SMA disfunctioning or/and faulty callosal integration (Bellgrove et. al., 2001).
Conclusion
There are two main hypothesis relating to bimanual motor planning:
The first, mainly based on psychophysical research in humans and some
electrophysiological studies, suggests that there is a common movement plan. The
second, based on psychophysical studies and the “well-known” physiological fact that
motor areas control their contra lateral side of the body, suggests that there are two
independent motor plans for both hands. The well-known dogma of contralateral
control is based on work recording ONLY one side of the brain during movement of
contralateral hand ONLY. In recent years more and more groups have reported a
relatively high ipsilateral and bilateral activity in motor cortical areas during
unimanual movement.
To summarize this section, although electrophysiological and lesion studies have
allowed us to amass a great deal of knowledge regarding the role of different motor
areas participating in motor control, neither of the two possibilities has been
completely rejected or accepted. The current level of physiological research does not
enable us to see the whole picture of a multiple motor areas network functioning to
produce a certain movement.
Working hypothesis.
We hypothesize that there are two main neural networks participating in motor
planning. The first, codes “bimanual default” movements, the second - unimanual
dominant movements. These functional networks, which partly overlap, are able to
reorganize and adapt themselves according to behavioral demands. In absence of
other instructions, the “bimanual default” leads to bimanual symmetric movement.
9
Different motor areas are associated with these motor programs: MI is mostly engaged
in coding default (bimanual symmetric) movements, while SMA and the Basal
Ganglia (BG) play an important role in breaking the default symmetry.
Rationale
Evolutionary, physiological systems developed according to functional demands. In
this context, I suggest that symmetrical movements are more “natural” and common
(walking, climbing trees, grasping food, etc). This may explain why the motor system
was designed for symmetrical movements. On the other hand, from the early
childhood humans are forced to use one hand more frequently and more precisely
(eating, writing, etc). Therefore it is conceivable that during the development of motor
behavior, in addition to the “bimanual neural network,” the brain developed a
“unimanual dominant neural network” that partly overlaps with the previous one.
When the SMA is injured, in the absence of symmetry-breaking mechanisms, the
system tends to mirror movements. Mirror movements can be seen as the inability to
increase the system’s degrees of freedom for non-symmetric (independent)
movement. It has been found (Woolsey et. al., 1952; Penfield and Welch, 1951;
Welker et. al., 1957) that stimulation of the SMA in anesthetized animals elicits
complex movements, often involving more than one joint, and, only rarely, distal
extremity movement.
It should be stressed that our hypothesis and the mentioned above investigations are
distinguish between the proximal and the distal parts of the body. In the early
evolution of man, he learned to work with the distal arm (fingers) in a very
independent manner. It is known that the two MI are poorly connected through the
corpus callosium for the hand and fingers, and well connected for the upper arm
(Rouiller et. al., 1994). Functionally, there is a fundamental difference between the
control of the proximal and distal parts of the limb (Gazzaniga et. al., 1998, Jakobson
et. al., 1994). Therefore, “bimanual symmetrical default” is related more to the
proximal part of the arm.
10
If we accept for a moment that the role of the SMA is to increase the number of DF in
the motor system, this may explain the appearance of complex multijoint movements
after stimulation of the SMA. It is important to note that the SMA modulates the MI to
generate complex multijoint movement (without the MI this effect is not seen). It is
known that a BG lesion is followed by reduced activity in the SMA (Jenkins et. al.,
1992; Dick et. al., 1986; Deeke et. al., 1987). This is usually explained by the dense
neuron projection from BG to SMA. It is likely that in BG diseases like PD, some of
the motor deficits are not caused by damage to the BG itself, but by decreased SMA
activity. Therefore the behavioral deficits of coordination are similar in these two
kinds of lesions. An example of this similarity is shown in our SMA lesion study.
Experimental predictions of the working hypothesis:
1.
The hypothesis assumes that limbs should be programmed to move in a
coordinated fashion as a default choice. Interhand coupling reduces the FDF
(functional degrees of freedom) and, therefore, simplify the computation of the
movement.
2.
The hypothesis assumes that performance of nonsymmetric movements requires
a mechanism that breaks the default coupling. This suggests that such movements
must be acquired through special learning processes.
3.
The hypothesis assumes that unimanual movements resemble a particular case
of bimanual nonsymmetric movements, movements of the non-dominant hand
alone should be “more complicated” then in bimanual tasks. Namely, we expect
longer preparation time for unimanual then for bimanual symmetric movements.
4.
The hypothesis assumes that SMA (or BG) have a role in decoupling, therefore,
in the absence of SMA (or BG) function, neural activity in the MI should be more
related to bimanual symmetric movements then under normal conditions.
11
Section 2: The interconnection between the various parts of this work
This work began with my participarion in recording neuronal activity in the MI and
the SMA in monkeys during their performance of unimanual and bimanual task. In
this research, carried out together with Opher Donchin and Orna Steinberg, we
showed that bimanual specific neuronal activity exists in these brain areas (Donchin
et. al., 1998, Donchin et. al., 2001 a&b), and concluded that bimanual movement is a
different entity from unimanual movement.
In order to understand the nature of this specific representation of
bimanual
movements, I studied several of their behavioral aspects. First, unstructured natural
movements, which allowed us to test the natural tendencies of humans in producing
bimanual movements. Second, well structured, center-out unimanual and bimanual
movements, where we could control most of the movement parameters.
In addition to the behavioral study in monkeys and humans, we recorded neural
activity in monkeys during a center-out unimanual and bimanual task. Analysis of the
neural recordings showed some neural correlates with the behavioral phenomena we
observed.
In the previous recordings (Donchin et. al., 1998, 2001) we did not find any functional
difference between MI and SMA motor areas. Therefore to address the specific role of
the SMA in bimanual coordination, we made lesions in the arm-related area of the
SMA. Laterality analysis revealed the neural code meaning of hand preference. We
suggest that hand preference plays an important role in bimanual motor planning.
Cross-correlation analysis and its role in bimanual coordination were investigated with
my collaborator, Simone Cardoso de Olievera. In our paper (Cardoso et. al., 2001) we
describe how the neural network as a whole participates in motor coordination
planning.
To summarize, we integrated psychophysical, electrophysiological and lesion study
approaches in order to arrive at a more holistic understanding of the nature of
bimanual coordination.
12
Experiments and results
Chapter A: Scribbling psychophysics.
Introduction
It is very easy to perform the same movement with both hands, while it pretty
complicated to perform different movements simultaneously. In recent years, this
bimanual coupling phenomenon has been extensively studied in various labs (Franz
et. al., 1996, 1997, 2001, Eliassen et. al., 2000, Kelso et. al., 1984, Semjen et. al.,
1995, Byblow et. al., 1998, 2000, Walter & Swinnen 1990, Swinnen et. al., 1996,
1997 a&b). These experiments involved relatively simple, well-instructed and
structured tasks. People generally continuously draw circles, ovals or lines with two
hands, using a different frequency of drawing for each hand. In these kinds of tasks it
was shown that there is a tendency to temporal lock of the phase between the two
hands to in-phase or anti-phase. In addition to the temporal coupling, spatial coupling
was also found, where people tend to draw two ovals with both hands, although asked
to draw a circle and line (Franz, 1997). This spatial coupling exists also in amputee
patients who only imagine the drawing by the non-existent limb while drawing with
the other (Franz et. al., 1998). Temporal coupling is enhanced in split-brain patients
(Tuller et. al., 1989), whereas spatial coupling is reduced in split-brain patients (Franz
et. al., 1996, 2000, Eliasen et. al., 2000). These facts lead to the conclusion that the
nature of such coupling is not mechanical (some muscles in the periphery), but neural
(a common plan for both hands).
Knowing that such coupling exists in a well-instructed structured task, we posed the
question: How does person move without any instructed constraints. For this purpose,
we seated the person near a mechanical two-arm manipulandum and asked him/her to
13
perform a “chaotic movement”, where the only requirement was to move the
manipulandum with both hands. The duration of the experiment was 1-3 min.
Methods
Behavioral setup
The experimental setup is described in the papers attached to this dissertation
(Donchin et al., 1998, Donchin et. al., 2001 a,b). Subjects simultaneously moved two
separate manipulanda, one with each arm. Each manipulandum consisted of a lowweight, low-friction, two-joint mechanical arm, moveable only in the horizontal
plane. The X and Y positions of the manipulandum handles were continuously
recorded during performance of the task. The instructions were given verbally to each
subject before the experiment.
Two different groups participated in the experiments. The first, consisting of 9 healthy
students (6 males, 3 females), aged 24-40 years, were asked to perform a “chaotic
movement”, where the only requirement was to move the manipulandum with both
hands. The duration of the experiment was 1-3 min. The second group consisted of 10
healthy students (7 males, 3 females) was taken mostly from our lab, who were
strongly motivated to “prove us wrong.” This group was shown the results of the first
group and was told that our prediction was that by default arms are strongly coupled.
They then were asked to perform a “non-symmetric” movement, in order to show that
both arms can be consciously decoupled. The duration of this experiment was 1-3
min.
The experimental procedures were in accord with the Declaration of Helsinki
(“Ethical Principles for Medical Research Involving Human Subjects”, adopted by the
18th WMA General Assembly Helsinki, Finland, June 1964 and continuously
amended, last by the 52nd WMA General Assembly, Edinburgh, Scotland, October
2000).
14
Data analysis
The simplest way of evaluating arm interdependency was to plot the trajectory
components of both arms. As trajectory components, we took X and Y coordinates in
Cartesian space, and phase and amplitude as polar coordinates.
To quantify this interdependency, we calculated the correlation coefficients between
the two arms for each trajectory component. These correlation coefficients were
calculated for short (330-380 ms for X and Y coordinates, 80-120 ms for phase) nonoverlapping pieces of data. These lengths were chosen because this was the typical
time period range of the data. The distribution of the correlation coefficient for all the
data was then plotted separately for each trajectory component.
Results
The surprising finding from the first set of experiments was that although people tried
to move their two hands randomly, there was very high inter-arm coupling during the
movements. It is evident from the X and Y coordinate traces for each hand that both
hands tend to perform symmetric in-phase or anti-phase movements (Fig 1). The Y
inter-hand coordination was largely in anti-phase, whereas the X coordinate was
usually in-phase.
An additional interesting finding is that the X and Y coordinate are apparently
meaningful for humans, because the coupling pattern can be orthogonal in the X
compared with the Y coordinate. (X is right/left movement; Y is forward/toward the
body movement).
15
Fig. 1: Bimanual “chaotic” scribbling: This figure is depicts a typical example
of movement in an unstructured bimanual “chaotic” task by one subject.
A. Two-dimensional traces of both hands; red - right hand; blue – left hand.
B. X-projection of the movement; red - right hand; blue – left hand.
C. Y-projection of the movement; red - right hand; blue – left hand.
To obtain some independency from X and Y coordinate system, we used the polar
coordinate system. In these coordinates the phase measurement is of prime
importance, because r (amplitude) largely reflects the size of the movement (Fig 2). In
these coordinates the inter-hand coupling is even more evident.
Fig. 2: Bimanual “chaotic” scribbling: In here the same data as in Fig. 1
is plotted in polar coordinates. Red - right hand; blue – left hand.
16
Subjects, who performed the “non-symmetric” scribbling, reported that to do so they
used different tricks, like thinking about playing different music with each arm,
imagining Brownian motion or juggling. The subjects also reported that this task was
unexpectedly hard to perform and demanded high concentration. When I attempted to
prolong the duration of the experiment, most of the subjects asked me to terminate the
experiment, because they felt too tired to continue. Analysis of the last minute of the
experiment usually showed a greater tendency to inter-arm coupling than during the
first minute. This means that it was hard to perform something really non-symmetric
for even two minutes. Further analysis of the “non-symmetric” scribbling experiments
showed that on the average, the correlation between the hands was lower than in the
“chaotic scribbling”, but still quite high, indicating that in this condition as well the
two arm were moving in a coordinated, rather than independent fashion (Fig 3).
Fig. 3: This figure shows “chaotic” compared with “non-symmetric” scribbling
Partial X and Y projections of movement plotted according to the same format as before.
Normalized distribution of the correlation coefficients between the X and Y projections of the two
arms is plotted separately. Cross correlation coefficients were calculated in a 375 ms time
window, the average time period for all the data.
A. “Chaotic” scribbling
B. Nonsymmetric scribbling
17
Calculation of the distribution of phase correlation coefficients showed an even stronger
tendency of both hands to work in symmetric fashion in both “chaotic” and “nonsymmetric” scribbling. An example of phase analysis of “non-symmetric” scribbling is
shown in Fig. 4.
Fig. 4: Phase correlations in nonsymmetric scribbling.
A. Partial plot of data phases in polar coordinates for right and left hand.
B. Normalized distribution of phase correlation coefficients. The cross
correlation coefficients were calculated in a 100 ms time window, the
average time period for all data.
One person from our laboratory volunteered to learn to decouple the arms during
several sessions on different days. Each day after performing a non-symmetric task he
was shown the results. This allowed him to draw conclusions from his performance and
to prepare a new tactic for the following day. The experiment was carried out for 4
days. Upon termination of the experiment he successfully fulfilled most of our criteria
for non-symmetric performance most of the time, possibly because he know exactly
which type of analysis would be applied to the data. Even at the last session, he
reported that it was extremely hard to remain in non-symmetric fashion of movement
during the two-minute period.
18
Intermediate summary, discussion and conclusions
To summarize this section: we show that in a simple unstructured task, as close as
possible to a natural movement, both hands are strongly coupled by default. Every
person can consciously reduce this coupling, but significant decoupling demands
special training.
The main difference between our task and tasks reported in the literature is that our
assignment involved minimal instructional constraints and, therefore, was closer to
what humans tend to do naturally.
What is the neural basis of inter-arm coupling?
We know that our brain is a dynamic system that can change according to our daily
needs. If we look at the most frequently used types of movement we perform, we notice
that many of them are unimanual (using one hand, while the other is “paralyzed” in
place) or bimanual symmetric (where both hands move in-phase or anti-phase). For
example, hand movement during walking, opening a bottle, catching and talking are
bimanual symmetric, whereas writing, drawing, reaching and pointing are performed
using one hand only. However, bimanual nonsymmetric movements are much less
common, and usually require special skills.
We should bear in mind that our body is a multi-joint system with high degree of
freedom. The control of such a system entails a very high computational cost. We know
that our movements are relatively fast for such a high computational time. This leads us
to suggest that our system most likely uses some time efficient algorithms to reduce the
computational cost. One possible way to reduce the computational amount is by coding,
in addition to the coordinates of each joint, the correlation value with other joints. This
is reasonable in view of the fact that most of the time the joints do not move
independently and, the functional number of degrees of freedom is much lower then the
anatomical one. Therefore computation of the functional degrees of freedom is also
lower then would be expected on the basis of the anatomical number of joints. The
existence of inter-arm coupling, showed in our experiments, further supports the
hypothesis of inter-joint coupling in order to reduce the computational cost for each
movement. Assuming this notion is correct, we can readily understand why the
19
performance of really non-symmetric (high degree of freedom) movements demands
special learning
Going back to the experimental results, we showed that both hands are not independent
in the simple natural task. Therefore, the FDF (functional degree of freedom) of the two
hands system is lower then ADF (anatomical degree of freedom). However, in our
experiments we showed only inter-dependency of the movement of both manipulanda
by two hands. The next question is how this inter-hand coupling is reflected by the
coupling of the different joints within and between the hands. Another student in our
lab. investigates this issue now. He runs the similar series of experiments using the
OPTOTRACK system, which enables us to track the position of each joint of each arm
independently.
20
Chapter B: Psychophysiology article (in preparation)
Neuronal
correlates
of
temporal
bimanual
coordination:
a
combined
physiological and behavioral study in monkeys and humans.
Gribova A.1, Donchin O.1,3, Bergman H.1 , Vaadia E.1 , Cardoso de Oliveira, S.1,2
1: Department of Physiology and the Interdisciplinary Center for Neural
Computation, The Hebrew University, Hadassah Medical School, P.O.B. 12272,
Jerusalem 91120, Israel
2: present address: Institut for Arbeitsphysiologie an der Universit‫ה‬t Dortmund,
Ardeystr. 67, 44139 Dortmund, Germany
3: present address: Department of Biomedical Engineering, Johns Hopkins
University, 720 Rutland Ave. 416 Taylor Bldg., Baltimore, MD 21205, USA
Abbreviated Title: Neural correlates of bimanual coordination
Text:
34 Pages
Figures:
7
Tables:
1
Abstract:
205 Words
Introduction:
698 Words
Discussion:
1838 Words
21
Correspondence to:
Simone Cardoso de Oliveira
Institute for Arbeitsphysiologie
An der University Dortmund
Ardeystr. 67
D-44139 Dortmund
Tel:
++49-(0)231-1084-311
Fax:
++49-(0)231-1084-340
email: [email protected]
Acknowledgements:
The research was supported in part by the Israel Science Foundation, the Israel
Academy of Sciences and Humanities, the United States-Israel Binational Science
Foundation, the German-Israeli Foundation for Scientific Research and Development
(GIF) and the DFG (CA 245/1-1).
We thank the MINERVA foundation that
supported S.C.d.O. as a postdoctoral fellow, Mifal Hapise foundation for a fellowship
to A.G. and the Clore Foundation for a fellowship to O.D.. We also thank G. Goelman
for obtaining the MRI-pictures. Histological evaluation was performed in
collaboration with S. Haber (Univ. Rochester, USA).
22
Abstract:
This study investigates the programming/planning mechanisms of bimanual
coordination by analyzing unimanual and bimanual movements in a combined
behavioral and physiological approach. Human subjects and rhesus monkeys
performed the same bimanual task. While the monkeys were engaged in the task,
neuronal activity was recorded simultaneously in the two hemispheres of motor
cortex.
Both for monkeys and humans, the reaction times of bimanual movements never
significantly exceeded the one of unimanual movements of the slowest arm.
Consistent with this, the longest delay between neural activity onset and movement
initiation occurred with unimanual movements of the slower arm and not with
bimanual movements. Both results suggest that the programming of bimanual
movements does not require more processing time than unimanual ones.
In both species, movement initiation (before: reaction time) of the two arms was
highly correlated. However, once movements began, the temporal correlation between
the arms progressively declined. Movement decorrelation was preceded by a net
decorrelation of neuronal population activity, suggesting that neuronal interactions are
related to the level of bimanual coupling and decoupling.
Our results suggest that bimanual movements are jointly programmed and initiated.
We identified neuronal processing time and neuronal interactions as functional
correlates that may underlie bimanual programming and temporal coupling in
monkeys and humans.
23
Introduction:
Whenever we perform simultaneous actions of multiple body parts, their movements
have to be coordinated in some way to produce a joint movement pattern. How is this
coordination achieved? Are the individual movement components programmed
separately, and later combined into a coordinated complex movement, or is the whole
movement pattern planned in a joint effort? The most popular example of motor
coordination between multiple movements is the coordination between arms in
bimanual movements. On the basis of behavioral findings, at least 3 different
hypotheses about the mechanisms of bimanual coordination have been proposed.
One hypothesis holds that there may be an independent cognitive instance, a
"coordinating schema" (originally suggested for the interaction of reach and grasp,
Hoff & Arbib, 1993), which could coordinate between two independent motor plans
by rescaling the individual movements such that they are temporally aligned. In this
view, it is conceivable that the coordinative schema will require additional cognitive
resources, and probably also additional time to fulfill it's task. Finding increased
reaction times for bimanual as compared to unimanual movements could support this
idea. Indeed, some studies have reported that bimanual movements require longer
reaction times than unimanual ones (Kelso et al 1979, Garry at al. 2000). Other
studies, however, could not always confirm this hypothesis, since they found
prolonged reaction times only in certain bimanual tasks (Aglioti at al. 1993, Anson &
Bird, 1993).
However, there are also other possibilities how bimanual coordination may come
about. One of them is that the two independent motor plans interact directly with each
other, resulting in crosstalk-effects (Marteniuk et al., 1984, Heuer et al., 2001).
Crosstalk occurs probably on different levels, accounting both for high-level
interactions at the programming stage and execution-related interactions. Recent
studies have suggested that at least some components of crosstalk between bimanual
movements are not static, but rather decay with increasing processing time (Heuer et
al., 1998). The question arises whether these changes in crosstalk will be observable
24
during the execution of bimanual movements and will be reflected in changing levels
of temporal coupling between the arms. Previous psychophysical work suggests that
temporal coupling is strongest at the time when movements are started (Boessenkool
et al., 1999), and that coupling may deteriorate with time (Fowler et al., 1991). If this
decoupling is a result of crosstalk between the motor codes of the two arms, one may
expect to find signs of these interactions within the nervous system, at the level where
the motor codes are generated.
An alternative view takes the trend to produce spatially and temporally coupled
movements (Kelso et al., 1979, Franz et al., 1997, Eliassen et al., 2000, Kelso 1984,
Semjen et al., 1995, Byblow et al., 1998, 2000, Walter & Swinnen, 1990, Swinnen et
al., 1996, 1997 a&b) as support for the hypothesis that bimanual movements are not
generated on the basis of two independent motor plans, but rather by a single common
movement plan, a "generalized motor program" (GMP) (Schmidt 1975, Schmidt et
al., 1979). As a variant of this idea, the term "coordinative structure" was coined for
collectives (synergies) of muscles that can be controlled jointly as a single functional
unit (Easton, 1972, Turvey 1977). Like the GMP, the "coordinative structure"
enhances coding efficiency by reducing the number of separately controlled degrees
of freedom (Bernstein, 1967, Turvey, 1977), and explains the tendency for common
timing of bimanual movements. In this framework, a bimanual movement should be a
motor pattern just as a unimanual movement, and therefore, no differences between
bimanual and unimanual movements have to be expected.
In order to test the plausibility of the mentioned above models of bimanual
coordination, we addressed the hypotheses derived from them in a unimanual and
bimanual reaching task. To get access not only to the behavioral level, but also to the
underlying neuronal mechanisms, we combined the behavioral investigation with
recording neuronal activity from the brain of monkeys engaged in the same bimanual
task. To reveal interactions between the motor plans for both hands, we recorded in
both hemispheres of primary motor cortex.
25
Previous studies provided first evidence that primary motor cortex may not only
produce the motor programs of the contralateral arm, but may also be involved in
bimanual coordination (Donchin et al., 1998, Kermadi et al., 1998, Cardoso de
Oliveira et al., submitted, Dochin et al., submitted). A comparison of neuronal
activities during bimanual and unimanual movements had revealed that the neuronal
codes for both types of movements are different (Donchin et al., 1998 and submitted).
This finding renders first support to the idea that composed movements are not coded
by their components, but by a unified code for the whole motor pattern, as suggested
by the idea of the GMP model. On the other side, however, it is also evident that
hand-specific coding also exists in motor cortex (Evarts, 1979, Georgopoulos, 1982,
Kakei et al., 1999). Together with the idea that each motor cortical hemisphere mainly
controls the contralateral arm, one could expect that if the neural codes for both arms
interact, this interaction should show up in interhemispheric correlations of neuronal
activity. Indeed, studies on split-brain patients suggest that the corpus callosum is
involved in both spatial and temporal crosstalk (Franz et al., 1996, Eliassen et al.,
2000). By comparing behavioral interactions between the arms to the neuronal
interactions, we aimed at elucidating whether neuronal interactions really can be
involved in bimanual coordination. In addition, examining the time course of neuronal
activation will also help to infer whether preparation of bimanual movements is more
time-consuming than that of unimanual movements. This study deals with selected
aspects of neuronal activity that have direct bearing to the questions raised above.
Other details of the neurophysiological data are presented in separate studies
(Steinberg et al., 2001 submitted, Donchin et al., 1998, 2001 a&b, Cardoso de
Oliveira et al., in press).
26
Methods:
Behavioral paradigm
The same experimental setup was used for experiments with humans and monkeys.
The experimental setup and the principle of the task design have already been
described in Donchin et al., (1998 & in press). A sketch of a monkey engaged in
performance of the task is shown in Fig. 1A. Subjects moved simultaneously two
separate manipulanda, one with each arm. Each manipulandum was a low-weight,
low-friction, two-joint mechanical arm, moveable only in the horizontal plane.
Movement of each manipulandum caused movement of a corresponding cursor on a
vertically oriented 21” video screen located ~50 cm in front of the subject. The
movement of each cursor was mapped to its corresponding manipulandum movement
such that each millimeter of manipulandum movement caused one millimeter of
movement of the cursor on the video display.
The time course of typical unimanual and bimanual trials was as follows. A trial
began when the subject placed both cursors within 0.8 cm diameter ‘origins’ (Fig. 1A)
and held them still during a hold period (of 500 ms in monkeys G and F, or 1000 ms
in monkey P and humans). For each arm, a target (also 0.8 cm diameter) could appear
at a distance of 3 cm (all three monkeys) from the origin, and in one of eight different
directions. For humans, targets could occur at long (5 cm) or short (2.5 cm) distances
from the origins. If only one target appeared — signaling a unimanual trial — the
subject moved the appropriate arm and brought the corresponding cursor into the
target, but did not move the other arm. If two targets appeared — signaling a
bimanual trial — the subject moved both arms, such that the two cursors moved into
the targets on the screen. Three types of bimanual movements were studied: parallel,
opposite and perpendicular. Unimanual movements comprised movements in eight
different directions (in the four cardinal directions, plus those directions 45 degrees
from the cardinal directions, Fig. 1C). For humans, we also included long unimanual
movements, which were components of the bimanual movement types. Monkey F and
Monkey G performed parallel and opposite movements in eight directions. Monkey P
27
performed parallel, opposite and 90 degree movements in two different directions.
Human subjects performed parallel and 90 degrees movements with same or different
amplitudes and opposite movements.
Reaction time was not restricted, but targets had to be reached within 1 s (monkey F,
G and humans) or 1.5 s (monkey P) from target appearance. For bimanual trials, the
subjects were additionally required to begin movement of the arms within a maximal
inter-arm-interval (IAI) of 200 ms, and the targets had to be reached with an IAI of
400 ms. Following target acquisition, both arms had to be held still in the target circle
for at least 500 ms. No inter-trial interval was imposed, but successful trials were
rewarded with a liquid reward and were followed by a 2-second pause to allow for its
consumption. In experiments with human subjects, no reward was administered, but a
click signaled successful trials. In all sessions, the different movement types were
presented in pseudo-random order. During the behavior hand positions and trial
events were recorded continuously using in-house software built around data
acquisition boards (DAP 3200e, Microstar Laboratories, Bellevue, WA, USA).
28
Fig. 1: The behavioral task. A: The monkey moved two manipulanda in the horizontal plane. The
position of each manipulandum was displayed as a cross-shaped cursor on a vertical screen in front of
the monkey. Each trial began by presenting two origin circles in the middle of the display (circles with
crosses). After the monkey placed the cursors into the origins and held them immobile for a constant
delay, the origin circles went off and two target circles appeared at different locations. In unimanual
trials, one circle appeared in the same location as the origin (for the non-moved hand). The other
circle was displaced from the origin in one of the eight directions shown in C. B: Sketch of the raisin
boards used. Each well was supplied with a raisin and the monkey’s task was to retrieve them at its
own pace, and with either hand. The upper board tested simple reaching, while the lower one required
finger coordination to produce a precision grip. C: Unimanual movements could be in one of the eight
directions shown here. D: Schematic representation of the bimanual movements used in our task.
Upward arrows in C and D correspond to forward movements of the monkey, and downward
arrows to backward movements towards the chest of the monkey.
29
Monkeys
Three female rhesus monkeys (Macaca mulatta) (monkey F, 3.5 kg, monkey G, 3.5
kg, and monkey P, 4 kg) were trained in the task. They were familiarized with the task
in a step-by-step procedure lasting for 6-10 months. The behavioral data presented
here were collected after extensive training, and after the monkeys reached a stable
level of performance in all movements types. During the sessions used for this study,
neuronal activity was also recorded (see below). After completion of this set of
recordings, monkey G has been trained in a different set of movements. This second
set of recordings and recordings from monkey P have been analyzed also in another
study that deals in more detail with LFP dynamics (Cardoso de Oliveira et al.,
submitted).
In order to test the hand preference in monkeys, we used two ‘raisin board’ tasks (see
Fig. 1B). The first consisted of a 12.5 x 34 cm perspex board into which 9 wells of 4
cm diameter were drilled. This arrangement allowed fast retrieval of food reward out
of the wells by whole-arm reaching grasping. The second was a 10 x 20 cm perspex
board with 15 elongated slots (15 mm long, 6 mm deep and 6 mm wide) oriented
randomly at 0, 45, 90, 135 or 180 degrees. The spatial dimensions of the wells
required retrieval of food by opposing thumb and index finger in a precision grip
(Brinkman, 1984). Each well was loaded with a raisin by the experimenter. The
monkey sat in the primate chair and was presented the board, which was positioned in
the middle, ~ 30 cm in front of the monkey’s chest. The monkey rapidly collected all
raisins from the wells. The behavior of the monkey was videotaped, and hand
preference was determined by counting how often the monkey used either hand to
retrieve the raisins.
Human subjects
All human subjects (age 23-32, 2 female, 4 male, 5 right-handed, 2 left-handed,
according to a modified Edinburgh inventory questionnaire, Oldfield, 1971) gave their
informed consent and were naive to the purpose of the experiment. None of the
subjects was aware of any neurological or neuromuscular abnormalities. The
experimental procedures were in agreement with the Declaration of Helsinki ("Ethical
30
Principles for Medical Research Involving Human Subjects", adopted by the 18th
WMA General Assembly Helsinki, Finland, June 1964 and continuously amended.
last by 52nd WMA General Assembly, Edinburgh, Scotland, October 2000) . Each
human subject performed 3-6 sessions on different days. Only sessions with at least
60 % successfully performed trials were included in this study. In some subjects, the
first session did not conform to this criterion and were discarded.
Recording of neuronal activity in monkeys
After completing the training period, two recording chambers (exposing the primary
motor cortex) and a head holder were implanted on the skull. Surgery was performed
as described before (Donchin et al., 1998, and in press, Cardoso et al, submitted). The
animals’ care and all surgical procedures were in accordance with the NIH Guide for
the Care and Use of Laboratory Animals (rev. 1996) and all applicable Hebrew
University regulations.
During recording sessions, the monkey was seated in a primate chair placed in a dark
chamber with its head fixed. Single unit activity and LFPs were recorded by eight
glass-coated tungsten microelectrodes (impedance 0.2-0.8 MΩ at 1 kHz) in the two
hemispheres (four electrodes in each hemisphere).
Electrodes were individually
driven. Therefore the depth of electrodes could vary, but the perpendicular distance
between electrodes was approximately 500 μm. The electrode signals were amplified
and filtered by a multi-channel analog data processor (MCP, Alpha-Omega, Nazareth,
Israel). The raw wideband signal was subjected to two different ranges of band-pass
filters in order to yield spike activity (filtered between 300 Hz and 8 kHz) and LFP
signals (filtered between 1 and 150 Hz).
Neurons were selected for recording on the basis of the isolation quality of their spike
waveforms and stability of their firing rates. We used the MSD sorter (Alpha-Omega,
Nazareth, Israel) to isolate the spiking activity of up to three neurons per electrode
based on an eight-point template-matching algorithm. Spike occurrences and
behavioral events were recorded with a temporal resolution of 24 kHz, but were
down-sampled off-line to a resolution of 400 Hz. The waveforms of all detected
31
spikes and all unclassified threshold crossings were also sampled at 24 kHz allowing
off-line confirmation of spike sorting. The LFP data were sampled at 400 Hz.
Selection of recording sites
Four electrodes were advanced into the brain at each penetration coordinate. After
penetrating the dura mater, electrodes were separately advanced until well-isolated
units were detected. Recording sites were not especially selected for activity related to
movements, but for the occurrence of stable single unit recordings (as judged by
stability of spike amplitude and spike rate). At the end of each recording session, we
tested for neuronal responses to passive manipulation and tactile stimulation of the
limbs, tail, trunk and head.
Evoked activity was evaluated by listening to the
amplified spike signal passed directly into a loudspeaker. Finally, we applied
intracortical microstimulation (ICMS) with 50 ms trains of 200 μs cathodal pulses at
300 Hz with an intensity of 10-80 μA (BPG-2 and BSI-2, BAK Electronics,
Germantown, MD).
When ICMS evoked movements, we documented the
movements evoked and the threshold stimulation intensity. Stimulation and passive
manipulation were performed at the end of each recording session, as well as in
dedicated mapping penetrations. For this study, we only included recording sites that
were well within the arm representation, as determined by passive stimulation and
ICMS.
Confirmation of recordings sites:
After completion of recordings, monkey G and F were sacrificed and the brains
histologically processed and Nissl stained. Penetration positions were determined by
mapping them onto the surface of the brain using the chamber coordinates and
positions. Since monkey P is still participating in experiments, the locations of the
recording sites in this monkey were confirmed by magnetic resonance images. Most
of our recordings were in the exposed part of primary motor cortex, but some of the
more anterior sites in monkey G were located in the caudal portion of dorsal premotor
cortex.
32
Data analysis:
Behavioral data
All trials were aligned upon the beginning of movement of the hand that moved first.
Movement onsets (Mon) were determined by an off-line algorithm, and doublechecked manually. The algorithm (courtesy of A. Arieli) calculated the zero intercept
of a line passing through two points on the velocity profile. One point was at 2/3 of
the peak speed and the other was at 1/3 of the peak speed. It also included corrections
for different special cases of unusual velocity profiles. Alternatively, we also used an
online movement onset definition, which was defined using two velocity thresholds,
checked at different intervals. The more restrictive (15 mm/s) was averaged over a
greater time window (approximately 100 ms) and this prevented slow drift of the
arms. The less restrictive (30 mm/s) was averaged over a shorter time window.
The end of movement (Moff) was similarly determined offline, based on two
thresholds, which were set to 5 and 3 times the averaged velocity noise in a 100 ms
window before movement onset. Movement offset was detected at the time when the
more restrictive threshold was crossed, if after the crossing the average velocity
remained for 200 ms below the less restrictive threshold.
Reaction time (RT) and movement time (MT) were calculated separately for each
hand and each movement. RT was defined as the time interval between target onset
and Mon of the respective hand. MT was defined as the interval between Mon and Moff.
To elucidate differences in the execution of bimanual and unimanual movements, we
compared the RT, time to peak velocity (TTP), and MT of a given arm in bimanual
movements to the corresponding unimanual movements (non-parametric Mann
Whitney test, Siegel, 1956, p.126).
To assess the time course of bimanual correlations, we also determined the times of
additional movement landmarks of the velocity profile. These were the times when
movements reached 1/3 of their peak velocities (t1), 2/3 of their peak velocities (t2)
and peak velocity (t3) during the accelerating phase, and the times at which 2/3 (t4)
and 1/3 (t5) of the peak velocity were reached in the decelerating phase. Temporal
33
correlations between the movements of both arms were assessed by correlating these
temporal landmarks of both arms with each other, using the non-parametric Spearman
rank correlation coefficient (Siegel, p. 202 ff).
To compare the inter-arm synchronization at the beginning and the end of the
movement, we compared the time difference (inter-hand-interval, IHI) between the
temporal landmarks of each movement (RTs, t1-t5, MTs, the arrival times at the target
= target acquisition time, TAT).
Neuronal data
All cell recordings were assessed for inter-trial stability, and only stable portions were
analyzed.
To compare the onsets of movements to the beginning of significant neuronal
activation, we determined the onsets of neural activity changes for each cell using the
CUSUM algorithm (Ellaway 1977, Davey et al. 1986). This procedure was performed
on the basis of the average firing rate within a time window of 200 ms before to 300
ms after movement onset (as defined by the offline detection algorithm). We used two
thresholds, the stricter one being four times the confidence limit as defined in Ellaway
(1977) and the less strict one 1.5 times the confidence limit. Onsets were limited to
the time from target appearance to 400 ms after movement initiation. The distribution
of activity onsets of each bimanual movement type was compared to the one of the
corresponding unimanual movements. The significance of the difference was tested
using the non-parametric Mann Whitney test.
We next sought to compare the time course of movement correlations to the
correlations between neural activities within the cortical hemispheres. To that end, we
analyzed the LFP signals of each pair and movement type within several time
windows corresponding to behavioral landmarks. These windows contained the LFPs
from Mon to the averaged TTP, from the averaged TTP to the averaged movement
offset (Moff), and from Moff to 200ms after Moff. In addition, two windows of 200 ms
duration were analyzed before Mon (-400 to -200 ms before Mon , and -200 ms to Mon).
The average Mon, TTP and Moff were determined from the same movement type, and
were always taken from the first hand to move. Mean interhemispheric correlations
34
within these windows were calculated by averaging the LFP-correlations as revealed
by the JPETC-method (Cardoso de Oliveira et al., submitted). Briefly, the JPETC
method is an adaptation of the JPSTH method, developed by Aertsen and coworkers
for single neuron data (Aertsen et al. '89). The JPETC analyzes the correlations
between the trial-by-trial deviations of the LFP signal around their mean over trials.
Here, we restricted the analysis to synchronous correlations, corresponding to the
diagonal of the JPETC. After determining the average correlation in these windows
for each pair and movement type, we summarized the data by averaging over all pairs
and movement types.
35
Results:
Data:
Data were recorded during 11-26 recording sessions in monkeys and 3-6 sessions in
human subjects (for details see table 1). Between 1000 and 3000 (monkeys) and 300500 (humans) movements were executed in each session, resulting in 30-200
(monkeys) and 10-20 (humans) repetitions of each movement type. The neural
activity analysis includes data from 3 monkeys recorded during performance of
center-out task. Seventy-two units were recorded from the left and 57 from the right
hemisphere of monkey F, 130 from the left and 132 from the right hemisphere of
monkey G, and 249 from the left and 237 from the right hemisphere of monkey P.
The LFP signals from 48, 112 and 248
(monkeys, F, G, and P, respectively)
recording sites were chosen for analysis.
Hand preference
All but two human subjects were right-handed as judged by a modified Edinburgh
inventory questionnaire (Oldfield, 1971).
The hand preference of monkeys was determined by the raisin board tasks and by
performance in the bimanual task. Fig. 2 A demonstrates that all three monkeys
preferred to pick up raisins with the right hand. In bimanual movements, the right arm
was leading more often than the left arm. In addition, in two out of the three monkeys,
the reaction times of the right arm were significantly smaller than those of the left
arm, both during unimanual and bimanual movements (Fig. 2 B, monkey G and F,
Mann Whitney Test, p<0.05).
36
Fig. 2: Hand preference in monkeys. A.: The left column of the figure shows the
percentage of usage of the right (black bars) and left (white bars) hand in the two
raisin tasks (average values over 10 days for monkey G and 23 days for monkey P),
and the percentage of movements in which each hand began to move first in the
bimanual movements. In monkey F, the raisin task was not done. B.: The right column
of the figure shows the average RT for each hand in unimanual and bimanual
movements (white bars for the left hand, black bars for the right). Error bars display
standard deviations and the asterisks indicate a significant difference between the
hands (p<0.01, Mann-Whitney test).
Reaction times and movement times
Fig. 3 compares RTs of unimanual and bimanual movements (the latter being defined
by movement initiation of the first hand). In contrast to Fig. 2, we did not include all
unimanual and bimanual movement types, but specifically compare bimanual
movements to those unimanual movements that composed them. In all monkeys and
human subjects, the RTs of bimanual movements were not significantly longer than
those of the slower hand. RTs were usually shortest for the unimanual movements of
the dominant hand and longest for the unimanual non-dominant trials, while RT for
bimanual trials was in between.
In addition, we did the same comparison for each hand separately. Significance was
tested by a Mann-Whitney test, and significance thresholds were p<10-6 for monkeys,
and p<0.01 for humans (the reason for such a strict threshold in monkey was the large
number of observations). The results and significance values are shown in table 1.
37
Fig.3: Comparison of RTs in bimanual and unimanual movements. For each
subject, we show the average RT of the non-dominant hand during unimanual
movements (white bar), the RT of the leading hand during bimanual movements
(gray bar), and the RT of the dominant hand during unimanual movements (black
bar). In all monkeys and four human subject, the right hand was dominant. Bars
indicate standard deviations.
Also in this analysis, reaction times in bimanual movements were generally not longer
for bimanual movements than for unimanual movements of the slower arm. In all
monkeys and 4 out of six human subjects, the highest reaction times were found in
unimanual movements, mostly of the subdominant arm (only in case of the dominant
arm). Only in one single human subject, the RTs of the subdominant arm in bimanual
movements significantly exceeded the one of the same arm in unimanual movements.
In two monkeys (P and F) and two human subjects (R and E), bimanual RTs of the
subdominant arm were even faster than unimanual ones, indicating that, if anything,
bimanual performance accelerates reactions of the non-dominant arm. For the
dominant arm, however, there was no such clear trend. Monkey P and subject E
showed an acceleration of the dominant arm during bimanual movements, but the 2
other monkeys (F and G) and 2 subjects (M and A) showed a significant slowing of
responses during bimanual performance. We did not find any significant difference
between the different movement types. Usage of the online movement onset definition
gave similar results, with a tendency of RTs to be even shorter in bimanual
movements than in unimanual ones.
We also compared MT and TTP values in bimanual movements to the corresponding
unimanual movements (Table 1). MTs and TTPs showed opposite trends in humans
38
and monkeys. In monkeys, bimanual movement durations were generally smaller than
unimanual ones. In humans, in contrast, bimanual movement times were longer than
unimanual ones. Similarly, the length of the accelerating phase of the movement (the
TTP value) in monkeys was always shorter during bimanual than unimanual (both
hands).
In contrast, 4 humans showed significantly longer bimanual TTPs than
unimanual ones. The difference between monkeys and humans in the relative duration
of bimanual movements is the only clear difference between species noted in this
study. It may be related to the very different amount of training. While monkeys were
over-trained in the task, humans had not undergone training before data collection.
Monkey P, 22 sessions
Uniright
RT
MT
TTP
343(326,83)
729(728,18)
262(252,82)
Right in
bimanual
335(319,81)
612(594,173)
190(169,63)
Left in
bimanual
340(321,87)
632(621,151)
199(189,58)
Unileft
348(334,87)
677(673,16)
264(252,72)
Right
signif.
6.5 * 10-7
3 * 10-185
0
Left
signif.
5.6 * 10-8
2.1 * 10-60
0
Right
signif.
0
1 * 10-14
1 * 10-147
Left
signif.
1.6 * 10-6
1 * 10-121
0
Right
signif.
3 * 10-7
0.0177
1 * 10-37
Left
signif.
2 * 10-16
2 * 10-73
1 * 10-171
Right
signif.
0.4655
0.009
0.0071
Left
signif.
0.4589
2 * 10-6
2 * 10-8
Left
signif.
0.0122
1 * 10-18
1 * 10-14
Monkey G, 26 sessions
Uniright
RT
MT
TTP
234(227,48)
492(484,12)
231(217,68)
Right in
bimanual
254(247,49)
506(496,138)
211(197,65)
Left in
bimanual
279(269,63)
478(471,146)
222(204,72)
Unileft
280(264,78)
519(524,14)
262(247,81)
Monkey F, 11sessions
Uniright
RT
MT
TTP
238(234,46)
380(379,84)
274(267,66)
Right in
bimanual
241(239,45)
387(376,97)
263(249,66)
Left in
bimanual
271(259,64)
386(386,90)
227(217,62)
Right in
bimanual
292(284,73)
610(589,163)
249(244,67)
Left in
bimanual
295(292,68)
600(569,164)
238(232,66)
Unileft
280(267,72)
414(411,94)
257(244,70)
Subject I, 4 sessions
Uniright
RT
MT
TTP
291(280,64)
594(559,16)
240(237,62)
Unileft
297(282,79)
560(531,16)
219(212,52)
Subject D, 4 sessions(note: this subject was left-handed!)
Uniright
RT
MT
263(252,59)
583(534,14)
Right in
bimanual
267(259,59)
649(624,159)
TTP
257(247,55)
276(272,66)
Left in
bimanual
281 (272,56)
609(581,152)
Unileft
290(279,63)
539(511,14)
Right
signif.
0.0339
1 * 10-16
271(267,69)
245(234,65)
3 * 10-8
39
Subject A, 3 sessions
Uniright
RT
MT
TTP
293(272,96)
556(541,16)
275(264,68)
Right in
bimanual
302(287,86)
618(584,189)
291(283,70)
Left in
bimanual
313(299,90)
578(554,186)
266(254,76)
Right in
bimanual
351(338,92)
600(570,319)
325(326,86)
Left in
bimanual
322(314,87)
653(620,274)
309(309,79)
Unileft
301(277,93)
570(551,15)
267(247,79)
Right
signif.
0.0034
9 * 10-8
2 * 10-4
Left
signif.
8 * 10-4
0.2166
0.4455
Right
signif.
0.256
0.0337
0.0115
Left
signif.
5 * 10-4
0.3823
0.3015
Right
signif.
0.0747
0.0189
3 * 10-9
Left
signif.
5 * 10-4
0.0994
0.35
Right
signif.
7 * 10-18
1 * 10-45
1 * 10-12
Left
signif.
1 * 10-43
0.0303
8 * 10-10
Subject R, 3 sessions
Uniright
RT
MT
TTP
347(326,83)
649(566,23)
313(294,87)
Unileft
335(324,77)
653(619,19)
310(294,85)
Subject M, 3 sessions (note: this subject was left-handed!)
Uniright
RT
MT
TTP
221(227,32)
460(489,13)
189(179,54)
Right in
bimanual
230(224,51)
495(474,161)
222(217,55)
Left in
bimanual
240(233,65)
479(444,156)
205(198,45)
Right in
bimanual
282(277,57)
519(516,82)
263(257,63)
Left in
bimanual
307(302,64)
526(521,87)
266(257,65)
Unileft
219(217,39)
484(504,13)
204(197,45)
Subject E, 6 sessions
Uniright
RT
MT
TTP
295(289,58)
486(482,81)
251(242,61)
Unileft
333(324,70)
532(529,80)
277(269,67)
Table 1: Summary of RTs, MTs and TTPs of all monkeys and human subjects.
The first number always is the mean value; in brackets follow the median value and
the standard deviation. Uniright = mean values (for all sessions and all trials) during
unimanual right movement, etc. Prior to averaging, outliers have been eliminated.
They were defined as RTs and MTs <50 or >1500, and TTPs <50 and >500. The two
right most columns indicated the error probabilities of a Mann-Whitney U-test
comparing values during unimanual movements to the movements of the same arm in
bimanual movements.
Delays between onsets of neural activity and movement initiation
In the single-unit activity recorded from the MIs of the three monkeys, we determined
the onsets of evoked activity prior to movement initiation. Fig. 4 shows the delays
between activity onsets and movement initiation during bimanual and unimanual
movements. Since results did not differ between the individual bimanual and
unimanual movement types, we present average values of all unimanual and all
40
bimanual movement types. In all 3 monkeys, this delay was the same for unimanual
right and bimanual movements. Unimanual left movements, however, were associated
with a significantly longer delay between activity onset and movement initiation.
Fig. 4: Activity onsets. In this figure we plot the mean (over all repetitions, trial-types
and recording days) time delay between of activity onsets and movement initiation in
bimanual (open bars) and the corresponding unimanual movements (gray for
unimanual right, filled bars for unimanual left). Negative numbers indicate that activity
began before movement onset. In all three monkeys, the activity onset in bimanual
movements is not significantly different from the corresponding unimanual movements
of the right hand, but significantly later then the activity onset during the movements of
the left hand. Significance was tested in a Mann-Whitney test (p<0.01).
Temporal correlations between the movements of the two arms
To unravel temporal coordination between the arms, we analyzed the temporal
relations between the movements of the two arms, and how these relations changed
over time. These temporal relations were determined by calculating the crosscorrelation coefficients between five different landmarks of the movements of both
arms. These landmarks were: reaction time (RT), the times elapsed till velocity had
risen to 1/3 (t1) and 2/3 (t2) of the peak velocity, the time till the peak velocity was
reached (t3), the times till velocity dropped to 2/3 (t4) or 1/3 (t5) of the peak velocity,
and, finally, the time till the end of the movement (MT). Fig. 5 shows the mean
correlations of these landmarks for each subject separately. In all subjects,
correlations were decreasing over time. For the mean over all subjects (all humans
41
and all monkeys), we calculated the slope of a regression line fit to it and the
significance of the linear regression coefficients. Both slopes were negative and p
values of the linear regression were highly significant (p<0.01).
In order to test whether the decorrelation seen here was the result of an accumulative
effect of errors during the movement, we also correlated the lengths of the time
windows between two subsequent time steps of the above analysis (the time
differences between the time taken to reach first 1/3 of peak velocity from the
movement onset (w1), time of 1/3 of the peak velocity and tame taken first to reach
2/3 peak velocity (w2), and so on). Fig. 5 B shows the result of this analysis. The
correlations decreased also in this case, indicating that the decrease was not an
accumulating effect, but correlations indeed became markedly weaker during
movements. Again, the slopes of the linear fit were negative and the linear fit was
highly significant (p<0.01). The slopes of the linear fit did not show any consistent
relation to the different types of bimanual movements.
Fig. 5 also shows that the average level of correlations between the arms was higher
in monkeys than in humans, but this difference was not significant (Mann-Whitney
Test, p=0.21). Nevertheless, the negative slopes of regression lines were very similar.
This indicates that while a species difference may exist in the average strength of
inter-arm coupling, the phenomenon of progressive decorrelation is common for both
humans and monkeys.
42
Fig. 5: Temporal correlations during bimanual movements: A.: The left panel shows an
example of the mean velocity profile of all trials from the same movement type, and the time
points at which 1/3, 2/3, peak velocity, etc are reached. All times are measured from the
beginning of movement, besides RT, which is measured from target onset. The right panel
shows the mean correlations (of all trials, movement types and days) for each time point
and each subject separately (different line styles represent different subjects). The
significance of a linear regression of the mean value (p) and the slopes of the regression
lines (s) are shown as numbers in the upper right corner of each plot. Fig. 5 B shows the
correlations of the lengths of time intervals (w1-w6, w6 being the interval between t5 and
TAT) elapsing between the temporal landmarks analyzed in Fig. 5 A. The first values of
these plots correspond to the time interval between reaching 1/3 of the maximal velocity and
movement onset and therefore are the same values as the second values in fig. 5 A.
In order to address whether not only the correlations, but also the time differences
between the arms changed over time, we analyzed also the differences between RTs,
MTs, TATs of the two hands and their variability (Fig. 6A). In all subjects (humans
and monkeys), the standard deviations of the RT differences were significantly
smaller than the MT and AT differences (p<0.05, Wilcoxon signed rank test). This
result was very consistent through all movement types. In all monkeys and most
subjects, the mean values of RT, MT and TAT differences were not significantly
different (only in human E, MT differences were smaller than RT differences). In
43
order to assess the time course of the increase in variability and compare it to the time
course of movement decorrelation (Fig. 5), we also calculated the differences and
standard deviations for the behavioral landmarks along the velocity profile (t1-t5,).
Fig. 6 B shows that along the movement, there was a progressive increase in
variability of the time differences between corresponding times in the movements of
both arms. Fitting a regression line to the mean over all subjects yielded a positive
slope.
Fig. 6: Differences (inter-hand intervals, IHI) of RT, MT and TAT in monkeys and
humans. This figure shows the means (big bars) and standard deviations (thin bars)
of RT differences (left black bars), MT differences (gray bars) and TAT differences
(right black bars) between the arms. For all the subjects (monkeys and humans) the
standard deviations of RT difference were significantly lower than the MT or TAT
differences (p<0.05, Wilcoxon signed rank test). There were no consistent significant
differences between the mean values of RT, MT and TAT differences.
44
LFP correlations
In order to compare the behavioral correlations to the interactions between neruronal
populations within and between the motor cortical hemispheres, we calculated the
average correlations between them in several time windows before and during
movement (see methods). Because neuronal correlations were noisy, we chose to
average over longer time windows than those in Fig. 5, separating two windows of
200 ms before movement onset, an acceleration phase (from Mon to- TTP), a
deceleration phase (TTP to Moff ), and 200 ms after Moff. Averaging the mean
correlations in these windows over all recorded pairs yielded the average time course
of neuronal correlations for each movement type. Since all movement types showed a
consistent time course, we averaged over all movement types, which is displayed in
Fig. 7.
In two monkeys, (monkey G and P), and both within and between
hemispheres, there was a strong decrease in correlation over time. This decrease
started already before movement onset (the first time window is significantly different
from all the subsequent ones, Mann Whitney test, p<0.001). There is an further
decrease after movement was initiated (the second window is also significantly
different from all the subsequent ones, Mann Whitney test, p<0.001). A similar
decrease was also seen with unimanual movements (not shown). In one monkey
(monkey F), we observed no significant change in correlation. This monkey, however,
showed also only a very mild decrease in correlation of behavior (see Fig. 5,
uppermost line in the monkeys' plot. In addition, the amount of data available from
this monkey was much smaller than in the other two monkeys (only 6 recording days
as compared to 30 in monkey P and 14 in monkey G), which resulted in more data
variability and less reliability of statistical tests.
45
Fig. 7: Time course of interhemispheric LFP correlations during bilateral movements.
Synchronous LFP correlations were averaged within time windows corresponding to the
behavioral landmarks of movement onset (Mon), TTP, and movement offset (Moff), as determined
by the mean of these values of the first hand moving. Additionally, we present two 200 ms wide
time windows before movement onset and one window after movement onset. Correlations
between the hemispheres decrease during movements, starting already before movement onsets.
Discussion
This paper describes two behavioral phenomena in bimanual coordination that occur
both in human and non-human primates. For each of those, we identified
corresponding neuronal phenomena that may constitute their physiological basis.
Motor programming of bimanual movements does not require more processing
time than unimanual movements
First, we found that bimanual movements did not require longer reaction times than
unimanual movements of the slower arm. This is in contrast to a view assuming that
motor programming is a time-consuming process which can be prolonged by
concurrently planned movements of the other arm (Spijkers et al., 2000). Previous
studies on this question have produced ambiguous results. Despite the tendency for
bimanual RT to exceed unimanual RT reported by Kelso et al (1979), there are
46
several cases where this has not been found. Garry & Franks (2000) saw increased
reaction times for bilateral movements only when an accuracy constraint was imposed
on the left, non-dominant hand. Anson et al. (1993) reported longer bimanual RT only
for finger extension, but not for elbow flexion movements, and Aglioti et al. (1993) in
finger, but not shoulder movements (in patients with callosal agenesis or
callosotomy). In agreement with the latter studies, our results showed also no
consistent RT difference between bimanual and unimanual movements. The bimanual
RTs in all monkeys and human subjects were equal or lower than the unimanual RTs
of the slower hand. In a few cases, the slower hand was even a sped up in bimanual as
compared to unimanual movements. This advantage in bimanual movements may be
due to coupling to the faster, dominant arm.
Revisiting the studies cited above, it seems that the exact task can account for the
differences between the results. Variations in additional constraints, such as
movement accuracy, have been shown to lengthen bimanual responses (Garry &
Franks, 2000). There seems also to be a difference between proximal and distal
movements (Aglioti et al., 1993, Anson et al., 1993). The fact that our movements
entailed mainly proximal movements supports the notion that in proximal movements,
bimanual RTs are equally short as unimanual RTs, while in distal movements, this is
not the case. A possible explanation for this phenomenon may be related to the fact
that the proximal representations of both arms are much more densely interconnected
between the cortical hemispheres than the distal ones (Rouiller et al., 1994), and thus,
interlimb coupling may be more efficient between proximal joints. Another relevant
parameter could be a different amount of training allowed for the experimental
subjects. In our study, we discarded the very first sessions of human subjects, where
performance was still erratic, and our monkeys extensively over -trained in the
bimanual task. On the other hand, the studies that did show longer RTs for bimanual
movements, used a single experimental session, probably resulting in less efficient
learning. One more finding of our study supports the notion that temporal demands on
bimanual movements are affected by training. Bimanual movement times in humans
were significantly longer than unimanual movement times. In our over-trained
47
monkeys, however, bimanual movements were even faster than unimanual ones. This
suggests that the time needed for execution of bimanual movements is strongly related
to the amount of training, and that bimanual movements only need more time than
unimanual ones when they are not well trained. One possible explanation for this
would be that prolongation of movement times is caused by independent motor errors
in the two arms, which have to be corrected by additional small compensatory
movements. With increased training, the subjects may learn to make their movement
plans more accurate, resulting in smaller deviations of the desired trajectory. The
successful refinement of the two motor plans may lead to enhanced coupling of the
arms, and that may also explain why the level of temporal correlation between the
arms was higher in monkeys than in humans.
The approximately equal RTs for bimanual and unimanual movements may be
explained on the physiological basis by the fact that the delay between neuronal
activity onset and movement initiation was not longer for bimanual than for
unimanual movements. This indicates that the neural processes specifying the motor
commands were produced at comparable speeds. It has been previously shown that
the level of neuronal activity in motor cortex is strongly related to the point in time
when a movement will be initiated (Evarts, 1974, Lecas et al., 1986). Insofar, the
good correspondence of physiological and behavioral data could have been expected.
However, it was unclear before, whether this would also hold for composite bimanual
movements. Another result further confirms the notion that reaction times are strictly
related to neuronal activity in motor cortex. Bimanual movements were initiated at
smaller delays from activity onset than unimanual movements of the left arm, which
is consistent with the fact that the RTs of all monkeys were longest for unimanual
movements of the non-dominant hand.
In addition to processing time, it has also been shown that bimanual movements do
not require more average activation than unimanual ones (Donchin et al., in press).
This is in agreement with a recent fMRI study, showing also no difference between
48
the MI activation during unimanual as compared to bimanual movements (Jencke et
al., 2000).
It is noteworthy that we did not find any difference in RTs and MTs of different
bimanual movement types, although it is known that symmetric movements are more
easily performed than non-symmetric movements. One explanation for this finding
could be the fact that the main part of recording (including humans) was done in the
stable part of the learning curve. Any possible differences between the easier,
symmetric and more difficult, non-symmetric movements may have been eliminated
by then by learning processes.
From the behavioral and physiological results discussed above we conclude that
simultaneous and coordinated bimanual movements do neither require additional
processing time nor stronger activation than unimanual ones. The question now arises,
which other parameters can account for coordination between the arms. The second
part of the discussion gives a potential answer to this question, by relating temporal
coordination between arms to temporal correlations of neuronal activity.
Progressive
decorrelation
of
movements
is
accompanied
by
neuronal
decorrelation
Second, the temporal coupling between movements of the arm underwent a
continuous decrease during movement execution, with RTs being more strongly
correlated than successive temporal landmarks along the velocity profiles. Temporal
decorrelation was also expressed in the fact that the variability of RT differences was
significantly lower than that of MT and TAT differences. This finding of progressive
bimanual decorrelation is consistent with several previous studies. Boessenkool et al
(1999) found significant bimanual correlation only for reaction times. Fowler et al.
(1991) also reported a progressive decorrelation of bimanual movements.
49
Assuming that each arm is controlled mainly by the activity in the contrateral cortical
hemisphere, one can hypothesize that temporal coupling between the arms may be
accomplished by correlated activity in the two hemispheres. Confirming this
hypothesis, interhemispheric correlations, like behavioral correlations between the
arms, decreased during movements. The one monkey that showed almost no
behavioral decorrelation also did not show a significant neuronal decorrelation
(monkey F, Figs. 5 and 7).
The fact that decorrelation began already at movement onset or even before, and thus
preceded the behavioral decoupling, is consistent with the idea that the two
phenomena are causally related. Previous work from our lab (in partly different sets of
movement types and different data sets) supports the idea that inter-hemispheric local
field potential correlations are related to the mode of bimanual coordination. In a
subset of interhemispheric pairs, activity correlations are transiently increased around
movement onset. These increases were found to be strongest during bimanual
symmetric movements, and less strong during bimanual asymmetric and unimanual
movements (Cardoso de Oliveira et al., submitted).
The behavioral and neuronal decorrelation raises the question as to its behavioral
significance. Why do movements start in a more coupled way and become decoupled
over time? Closer observation of movement kinematics suggests that independent
noise and proprioceptive feedback may account for progressive decorrelation. In our
task, velocity profiles were not bell-shaped and symmetric, but rather skewed to the
left, such that the acceleration phase was steeper and less variable than the
deceleration phase. An example of this can be seen in the average velocity profile
shown in Fig. 6. Asymmetric velocity profiles are observed especially when the final
location of the movement must be precisely controlled. MacKenzie et al. (1987) found
that the smaller the target the more time is spent in deceleration. Milner et al. (1990)
further demonstrated that increased requirements for accuracy produce an increased
asymmetry in the velocity profile. The current notion of motor control explains these
findings by a combined feedforward - feedback model, in which a fast, ballistic
50
feedforward program is first issued and, in the later stages of movement,
complemented by a feedback-based mechanism (Desmurget & Grafton, 2000).
Independent errors in the trajectories of the two arms would lead to independent
feedback. Proprioceptive feedback is relayed to the motor cortex, explaining why
neuronal activity becomes decorrelated. On the other hand, noise will cause
deviations from the desired feedback. Deviations from the desired trajectory will
make the movement slower, because corrective movements have to be produced, and
more variable, because different corrective movements are produced by the two arms,
leading to a deterioration of bimanual coupling. All these arguments suggest that
behavioral and neuronal decorrelation occur because of unspecific, independent noise
related to movement execution. The fact that we observed similar decorrelations in
symmetric and non-symmetric bimanual movements supports the idea that
independent, execution related noise is responsible for neuronal and behavioral
decorrelation.
Progressive decoupling occurs also within the same hemisphere and for unimanual
movements
However, we found decorrelation also within the same hemisphere. This finding can
be explained taking into account that there is also a considerable ipsilateral
representation in primary motor cortex (Wasserman et al., 1994), and many single
cells are active during both ipsilateral and contralateral movements (Tanji et al.1988,
Donchin et al. 1998, Kermadi et al. 1998). Thus, it is possible that interactions
between the motor codes of the two arms also take place between the ipsilaterally and
contralaterally dominated neuronal populations within each hemisphere. The fact that
decorrelation also occurred during unimanual movements seems more difficult to
explain. It may be plausible, however, under the mentioned above hypothesis that
behavioral decorrelation is the result of independent feedback. Clearly, the
proprioceptive feedback from the two arms is independent, irrespective of whether the
arms move or not. Thus, it may be explained that also decorrelations also occur when
only one arm moves while the other is stationary.
51
We conclude that motor processing time and behavioral decorrelation are reflected in
primary motor cortex. Together with previous accounts of bimanual related activity
(Donchin et al., 1998, in press, submitted) and dynamic LFP correlations (Cardoso et
al., submitted), these results strongly support a view in which MI and its
interhemispheric connections are involved in bimanual coordination. Both behavioral
findings were equally true for monkeys and human subjects, suggesting that the same
mechanisms may operate in human and non-human primates.
52
References:
Aglioti S., Berlucchi G., Pallini R., Rossi G.F., Tassarini G. (1993): Hemispheric control of unilateral
and bilateral respnses to lateralized light stimuli after callosotomy and in callosal agenesis. Exp. Brain
Res. 95: 151-165
Anson J.G., Bird N. (1993): Neuromotor programmimg: Bilateral and Unilateral effects on simple
reaction time. Human Movement Science 12: 37-50
Bernstein N. (1967): The coordination and regulation of movements. Oxford, England: Pergamon Press
Brinkman C. (1984): Supplementary motor area of the monkey's cerebral cortex: short- and long-trm
deficits after unilateral ablation and the effecst of subsequent callosal section. J. Neurosci. 4: 918-929
Boessenkool J.J., Nijhof E.-J., Erkelens C.J. (1999): Variability and correlations in bi-manual pointing
movements. Hum. Movmnt Sci., 18, 525-552
Byblow W.D., Bysouth-Young D., Summers J.J., Carson R.G. (1998): Performance asymmetries and
coupling dynamics in the acquisition of multifrequency bimanual coordination. Psychological Research
61: 56-70
Byblow W.D., Lewis G.N., Stienar J.W., Austin N.J., Lynch M. (2000): The subdominant hand
increases in the efficacy of voluntary alterations in bimanual coordination. Experimental Brain
Research 131: 366-374
Cardoso de Oliveira S., Gribova A., Donchin O., Bergman H. and Vaadia E. (submitted): Neural
interactions between motor cortical hemispheres during bimanual and unimanual arm movements
Davey N.J., Ellaway P.H., Stein R.B. (1986): Statistical limits for detecting change in the cumulative
sum derivative of the peristimulus time histogram. J. Neurosci. Methods 17: 153-166
Desmurget M., Grafton S (2000): Forward modeling allows feedback control for fast reaching
movements. TICS 4: 423-431
Donchin O., Gribova A., Steinberg O., Bergman H., Vaadia E. (1998): Primary motor cortex is
involved in bimanual coordination. Nature, 395, 274-278.
Donchin O., Cardoso de Oliveira S., Vaadia E. (1999): Who tells one hand what the other is doing: the
neurophysiology of bimanual movements. Neuron, 23, 15-18
Donchin O., Gribova A., Steinberg O., Bergman H., Cardoso de Oliveira S., Vaadia E. (in press): Local
field potentials related to bimanual movements in the primary and supplementary motor cortices. Exp.
Brain Res.
Easton T. (1972): On the normal use of reflexes. American Scientist 60: 591-599
Eliassen J.C., Baynes K., Gazzaniga M.S. (2000): Anterior and posterior callosal contributions to
simultaneous bimanual movements of the hands and fingers. Brain, 123, 2501-2511
Ellaway P.H. (1977): An application of cumulative sum technique (cusums) to neurophysiology. J.
Physiol. Lond. 265: 1P-2P
Evarts E.V. (1974): Precentral and postcentral cortical activity in association with visually triggered
movement. J. Neurophysiol. 37: 373-381
Evarts E.V. (1979): Brain mechanisms of movement. Scientific American 241: 164-179
53
Fowler B., Duck T., Mosher M., Mathieson B. (1991): The coordination of bimanual aiming
movements: evidence for progressive desynchronization. Quart. J. Exp. Psychol., 43, 205-221
Franz E.A., Eliassen J.C., Ivry R.B., Gazzaniga M.S. (1996): Dissociation of spatial and temporal
coupling in the bimanual movements of callosotomy patients. Psychol.Sci., 7, 306-310.
Franz E. (1997): Spatial coupling in the coordination of complex actions. Quart. J. Exp. Psychol., 50A,
684-704
Garry M.I., Franks I.M. (2000): Reaction time differences in spatially constrained bilateral and
unilateral movements. Exp. Brain Res. 131: 236-243
Georgopoulos, A.P., Kalaska, J.F., Caminiti, R., and Massey, J.T. (1982): On the relations between the
direction of two-dimensional arm movements and cell discharge in primate motor cortex. Journal of
Neuroscience 2: 1527-1537
Heuer H., Spijkers W., Kleinsorge T., Van der Loo H., Steglich C. (1998): The time course of crosstalk during the simultaneous specification of bimanual movement amplitudes. Exp. Brain Res. 118:
381-392
Heuer H., Kleinsorge T., Spijkers W., Steglich C. (2001): Static and phasic cross-talk effects in discrete
bimanual reversal movements. J. Mot. Behav. 33: 67-85
Hoff B., Arbib M.A. (1993): Models of trajectory formation and temporal interaction of reach and
grasp. J. Motor Behav. 25: 175-192
J‫ה‬ncke L., Peters M., Himmelback M., N‫צ‬sselt T., Shah J., Steinmetz H. (2000): fMRI study of
bimanual coordination. Neuropsychologia 38: 164-174
Kakei S, Hoffman DS, Strick PL. (1999): Muscle and movement representations in the primary motor
cortex. Science 285: 2136-2139
Kelso J.A., Southard D.L., Goodman D. (1979): On the coordination of two-handed movements. J.
Exp. Psychol., 5, 229-238
Kelso J.A.S. (1984): Phase transitions and critical behavior in human bimanual coordination. Am. J.
Phys., 246, R1000-R1004
Kermadi I., Liu Y., Tempini A., Calciati E., Rouiller E.M. (1998): Neuronal activity in the primate
supplementary motor area and the primary motor cortex in relation to spatio-temporal bimanual
coordination. Somatosens. Mot. Res., 15, 287-308
MacKenzie C.L., Marteniuk R.G., Dugas C., Liske D. and Eickmeir B (1987) Three dimensional
movement trajectories in Fitts’ task: implications for control. Quart. J. Exp. Psychol. 39A: 629-647
Marteniuk R.G., MacKenzie C.L., Baba D,M, (1984): Bimanual movement control: Information
processing and interaction effects. Quarterly Journal of Experimental Psychology 36 A: 335-365
Milner T.E., Ijaz M.M. (1990): The effect of accuracy constraints on three-dimensional movement
kinematics. Neurosci. 35: 365-374
Oldfield R.C. (1971): The assessment and analysis of handedness: The Edinburgh inventory.
Neuropsychologia 9: 97-113
Preilowski B.F.B. (1972): Possible contribution of the anterior forebrain commissures to bimanual
motor coordination. Neuropsychologia, 10, 267-277
Preilowski B.F.B. (1975): Bimanual motor interaction: perceptual-motor performance of partial and
complete ‘split-brain’ patients. In: Zulch K.J., Creutzfeld O., Galbraith G.C. (Eds.) Cerebral
localization. Springer: New York , pp. 115-132
54
Rouiller EM, Babalian a, Kazennikov O, Moret V, Yu XH, Wiesendanger M (1994): Transcallosal
connections of the distal forelimb representations in the primary and supplementary motor cortical
areas in macaque monkeys. Exp. Brain Res. 1994, 102:227-243
Schmidt R.A. (1975): A schema theory of discrete motor skill learning. Psychological Review 82: 225260
Schmidt R.A., Zelaznik H.N., Hawkins B., Frank J.S., Quinn J.T. (1979): Motor output variability: A
theory for the accuracy of rapid motor acts. Psychological Review 86: 415-451
Semjen A., Summers J.J., Cattaert D. (1995): Hand Coordination in bimanual circle drawing. Journal
of Experimental Psychology: Human Performance and Perception. 21: 1139-1157
Siegel S. (1956): Nonparametric statistics for the behavioral sciences. McGraw-Hill International
Student editions: Tokyo Auckland Duesseldorf Johannesburg London Mexico New Delhi Panama Sao
Paulo Singapore Sydney
Spijkers W., Steglich C., Heuer H., Kleinsorge (2000): Specification of movement amplitudes for the
left and right hands: evidence for transient parametric coupling from overlapping-task performance. J.
Exp. Psychol.: Hum. Perc. & Perf. 26: 1091-1105
Swinnen S.P., Jardin K., Meulenbroek R., (1996): Between-limb asynchronies during bimanual
coordination: effects of manual dominance and attentional cueing. Neuropsychologia 34: 1203-1213
Swinnen S.P., Van Langendonk L., Verschueren S., Peeters G., Dom R., De Weerdt W. (1997a):
Interlimb coordination deficits in patients with parkinson's disease during the production of two-joint
oscillations in the sagittal plane
Swinnen S.P., Jardin K., Meulenbroek R., Dounskaia N., Hofkens-Van den Brandt M. (1997b):
Egocentric and allocentric constraints in the expression of patterns of interlimb coordination. Journal of
Cognitive Neuroscience 9: 348-377
Tanji J., Okano K., Sato K.C. (1988): Neuronal activity in cortical motor areas related to ipsilateral,
contralateral, and bimanual digit movements of the monkey. Journal of Neurophysiology. 60, 325-343
Turvey M.T. (1977): Preliminaries to a theory of action with reference to vision. In R. Shaw & J.
Bransford (Eds.): Perceiving, acting, and knowing. Hillsdale, N.J.: Erlbaum
Walter C.B., Swinnen S.P. (1990): Asymmetric interlimb interference during the performance of a
dynamic bimanual task. Brain and Cognition 14: 185-200
Wasserman E.M., Pascual-Leone A., Hallet M. (1994): Cortical motor representation of the ipsilateral
hand and arm. Experimental Brain Research 100: 121-132
55
Chapter C: SMA lesion
Introduction
Although, the SMA is often associated with “coordinating bimanual movements”, this
notion is still disputable. Human EEG and brain imaging studies demonstrated the
specific activation of SMA during bimanual movements (Lang et. al., 1988, Viviani
et. al., 1998, Toyokura et. al., 1999). SMA lesions caused deficits in bimanual
coordination and “mirror movements” in monkeys (Brinkman et. al., 1981,1984) and
humans (Viallet 1992; Chan et. al., 1988). However, efforts to replicate bimanual
deficits in SMA-lesioned monkeys did not show any effect on bimanual coordination
(Kazennikov et. al., 1998, Kermadi et. al., 1997). Analysis of recordings from the
behaving monkey in our laboratory did not show any difference between the MI and
the SMA in the neural activity related to the control of bimanual movements
(Donchin et. al., 1998, 2001 a&b, Steinberg et. al., 2001). Other neuronal recordings
from a monkey performing pull and grasp bimanual tasks showed similar results
(Kermadi et. al., 2000).
To investigate the role of SMA in bimanual coordination we made a small ibotenic
acid bilateral lesion in the arm-related area of the SMA. In contrast to previous SMA
lesion studies (Brinkman et. al., 1981, 1984), we analyzed behavior and neural
activity in the MI from monkeys in both the normal and in the post lesion state.
Methods:
Behavioral paradigm
Two female rhesus monkeys (Macaca mulatta) (monkey G, 3.5 kg, and monkey P, 4
kg) were trained to move two manipulanda, one with each arm. The experimental
setup is described in the papers attached to this dissertation (Donchin et al., 1998,
Donchin et. al., 2001 a,b). A sketch of the monkey engaged in the performance of the
56
task is shown in Fig. 1A. The movement of each manipulandum caused a
corresponding cursor to move on a vertically oriented 21” video screen located ~50
cm in front of the monkey.
The movement of each cursor was mapped to its
corresponding manipulandum movement, such that each millimeter of manipulandum
movement caused one millimeter of movement of the cursor on the video display.
The time course of typical unilateral and bilateral trials was as follows: a trial began
when the monkey placed both cursors within 0.8 cm diameter ”origins” and held them
still for 500 ms (monkey G) or 1000 ms (monkey P). In all cases where the monkey
had to keep its arms still, we defined onset of movements by using two velocity
thresholds, checked at different intervals.
The more restrictive (15 mm/s) was
averaged over a greater time window (~ 100 ms), preventing slow drift of the arms.
The less restrictive (30 mm/s) was averaged over a shorter time window (~ 10 ms),
allowing rapid detection of true movement initiation. For each arm, a target (also 0.8
cm diameter) appeared at a distance of 3 cm (monkey P) from the origin. For monkey
G, targets were placed at long (5 cm) or short (2.5 cm) distances from the origins. If
only one target appeared — signaling a unilateral trial — the monkey moved the
appropriate arm and brought the corresponding cursor into the target, but did not
move the other arm (Fig. 1B). If two targets appeared — signaling a bilateral trial —
the monkey moved both arms, such that the two cursors moved into the targets on the
screen (Fig. 1C). Three types of bilateral movement were studied: parallel, opposite
and perpendicular. The types of bilateral trials used during the recordings are shown
in Fig. 1C. Note that for monkey G, the bilateral movements also entailed movements
of different amplitudes, but opposite movements were not included. Unilateral
movements comprised movements in eight different directions (in all four cardinal
directions, plus those directions 45 degrees from the cardinal directions, Fig. 1B).
Monkey P performed movements only of the same amplitude, opposite movements
were included. For monkey G, we also included long unilateral movements, which
were components of the bilateral movement types.
The monkeys’ reaction time was not restricted, but targets had to be reached within 1s
(monkey G), 1.5 s (monkey P) from the time of target appearance. For bilateral trials,
57
the animal was additionally required to begin movement of the arms within a maximal
inter-arm-interval (IAI) of 200 ms and the targets had to be reached with an IAI of
400 ms. Following acquisition of the targets, the monkey held both arms still in the
target circle for at least 500 ms. No intertrial interval was imposed, but successful
trials were rewarded with a liquid reward and followed by a 2-second pause to allow
for its consumption. In all sessions, trials were presented pseudo-randomly without
separation into blocks.
Fig 1: Reaching task
A. The monkey is performing a reaching task.
B. Unimanual right and left ” short” movement in eight directions
and unimanual ” long” movements
C. Example of bimanual movements. Right arrow - right hand, left
arrow – left hand.
58
In addition to this behavioral paradigm, we tested the monkeys´ hand preferences in a
“raisin board” task. The first consisted of a 12.5 x 34 cm Perspex board in which 9
wells of 6 cm diameter were drilled. This arrangement allowed fast retrieval of a food
reward from the wells by whole-arm reaching grasping. The second was a 10 x 20 cm
Perspex board with 15 elongated slots (15 mm long, 6 mm deep and 6 mm wide)
oriented randomly at 0, 45, 90, 135 or 180 degrees (Fig. 3A). The spatial dimensions
of the wells required retrieval of food by opposing thumb and index finger in a
precision grip (Brinkman, 1984). Each well was loaded with a raisin by the
experimenter. The monkey sat in the primate chair and was presented with the board,
which was positioned in the middle, ~ 30 cm in front of the monkey’s chest. The
monkey rapidly collected all the raisins from the wells. The monkeys’ behavior was
taped on video, and the frequency with which the monkey used either hand to retrieve
the raisins was counted.
Surgery
Surgery was performed as described (Donchin et. al., 1998, 2001 a&b; Cardoso de
Oliveira 2001). Animal care and all surgical procedures were in accordance with the
NIH Guide for the Care and Use of Laboratory Animals (rev. 1996) and all applicable
Hebrew University regulations.
Recording
During recording sessions, the monkey was seated in a primate chair placed in a dark
chamber with its head fixed. Single unit activity and local field potentials (LFPs)
were recorded by eight glass-coated tungsten microelectrodes (impedance 0.2-0.8 MΩ
at 1 kHz) in the two hemispheres (four electrodes in each hemisphere). The electrodes
were individually driven, allowing the depth-distance between electrodes to vary from
one session to another, but the perpendicular distance between the electrodes was
~500 μm. Electrode signals were amplified and filtered by a multi-channel analog
data processor (MCP, Alpha-Omega, Nazareth, Israel). The raw wideband signal was
subjected to two different ranges of band-pass filters in order to yield spike activity
59
(filtered between 300 Hz and 8 kHz) and local field potentials (LFPs) signals (filtered
between 1 and 150 Hz). The LFP data were sampled at 400 Hz and stored on a disk.
Fifty-Hertz noise caused by the A/C power line was attenuated using a digital notch
filter applied off-line (49.8 to 50.2, 99.8 to 100.2 and 149.8 to 150.2 Hz 4-pole
butterworth filter, applied forward and backward to prevent phase shifts).
Selection of recording sites
At each penetration coordinate, four electrodes were advanced into the brain, with a
lateral distance of 350-700 micron. After penetrating the dura, the electrodes were
separately advanced until well-isolated units were detected. Recording sites were not
selected for activity related to movements, but for the occurrence of stable single unit
recordings (as judged by the stability of spike amplitude and spike rate). LFP channels
numbered 0 to 3 were always recorded by microelectrodes in the right hemisphere,
and channels numbered 4 to 7 were recorded in the left hemisphere. Upon termination
of each recording session, we tested for neuronal responses to passive manipulation
and tactile stimulation of the limbs, tail, trunk and head.
Evoked activity was
evaluated by listening to the amplified spike signal passed directly into a loudspeaker.
Finally, we applied intracortical microstimulation (ICMS) with 50 ms trains of 200 μs
cathodal pulses at 300 Hz with an intensity of 10-80 μA (BPG-2 and BSI-2, BAK
Electronics, Germantown, MD). When ICMS evoked movements, we documented
the movements evoked and the threshold stimulation intensity. Stimulation and
passive manipulation were performed at the end of each recording session, as well as
in dedicated mapping penetrations. For this study, we included only recording sites
that were well within the arm representation, as determined by passive stimulation
and microstimulation.
SMA lesion
The lesion was made by several microinjection of 1ul of ibotenic acid each into the
area 2mmX10mm of SMA areas mapped previously as arm-related area. A quantity
of 1 mg of ibotenic acid (Sigma 12765, p581) was dissolved in 0.1 ml sterile saline,
60
yielding 10mg/ml. Each 1ul of ibotenic acid was gradually (0.1-0.2 ul) applied in 6080 sec intervals between boluses. After each injection, the Hamilton syringe was left
in place for an additional 5 min.
Histology
In the end of experiment monkey Galileo was sacrificed and her brain was sliced and
stained for the reconstruction of the lesion areas. The example for such a
reconstruction is plotted in Fig2.
Fig 2: Reconstruction of SMA lesion sites.
Here we draw the middle part of SMA lesioned area in monkey G. The black blobs
indicate the centers of the ibotenic acid lesion.
Monkey P is still in life. Based on the last MRI session the positions of the SMA
lesion in this monkey are questioned.
61
Data analysis:
All recording sites were assessed for inter-trial stability of single units and LFP
signals and for the occurrence of excessive noise. Recordings with recurring artifacts,
and those with strong 50 Hz noise persisting after notch filtering were excluded from
analysis. We also excluded data in which the single units or LFP-recordings changed
considerably across trials (e.g., baseline shifts or changes in noise level). No selection
was made on the basis of task-related activity.
All single unit activity and LFP traces were aligned upon beginning of movement,
determined by an off-line algorithm, and double-checked manually. The movement
initiation detection algorithm (courtesy of A. Arieli, The Weizmann Institute)
calculated the zero intercept of a line passing through two points on the velocity
profile. One point was at 2/3 of the peak speed and the other was at 1/3 of peak
speed. It also included corrections for different special cases of unusual velocity
profiles. For purposes of alignment of the neural activity data, the beginning of
movement in the bilateral trials was determined by the first arm to begin moving.
Performance rate was calculated by the percentage of successful trials in the total
number of trials.
Reaction time (RT) and movement time (MT) were calculated separately for each
hand for each kind of movement. Off-line definition of movement onset was
calculated using the Amos Arieli algorithm. On-line movement onset definition was
based on two thresholds, as described above. All these parameters were compared
during the normal state and after SMA lesions in the same monkey.
The significance of the differences was tested using the non-parametric Mann
Whitney test (S. Siegal, Nonparametric statistics for the behavioral sciences, 1956,
International Student’s Edition).
The onset of neural activity changes was determined for each peri-stimulus time
histogram (PSTH), using the CUSUM algorithm (Ellaway 1977;Davey et. al., 1986).
Onsets were limited to the time from target appearance to 400 ms after movement
initiation. The trial-by-trial firing rate of the cell was averaged from activation onset
until 500ms after activation onset (termed the activation epoch). The firing rate
62
during this epoch is termed the evoked activity. This was compared with a baselinefiring rate taken 350ms before activation onset to 100ms before activation onset (the
baseline epoch). While this period could, in principle, overlap the reaction time, the
algorithm guarantees that the neural activity is unchanged prior to response onset and,
therefore, our results were insensitive to the precise timing of the baseline epoch.
Generally, this was a period during which the monkeys’ arms were motionless at the
origin position, and we averaged activity in this period for each neuron across the
different types of movement. In cases with no response onset, as might occur for
example in non-preferred movement directions, we arbitrarily selected a default 500
ms period from 100 ms before movement initiation (the average activation onset
across responsive units) to 400 ms after movement initiation. PSTH onset distribution
was compared in normal and SMA lesioned monkey. The significance of the
difference between the normal and the lesioned state was tested using the nonparametric Mann Whitney test.
The average firing rate was calculated by averaging the neural activity of each cell in
a time window of -200:300 ms around off-line movement onset. The significance of
the difference between the normal and the lesioned state was tested using the nonparametric Mann Whitney test. Directional tuning of the cells was calculated by eye
for all the data.
The Laterality index was calculated for the average firing rate during unimanual
movements in a time window –200:300 ms around movement onset according the
formula:
Laterality index = (ContraResponse-IpsiResponse)/(ContraResponse+IpsiResponse)
The distribution of the laterality index is –1 to1. Cells responding mainly to the
contralateral side were give numbers closer to 1, while cells related mostly to the
ipsilateral arm tended to have a laterality index close to -1. This measurement is not
related to the directional tuning, but to the magnitude of the response. The
63
significance of the differences in laterality index distributions between the two
hemispheres was tested according to the chi square method.
Cross-correlation analysis
As a first approach towards studying correlations between LFPs, we calculated
classical cross-correlograms between LFP channels, which depict the time average of
the correlation between the channels. We applied this analysis to two different epochs
of the trial. The first was the interval between 750 and 250 ms before movement
onset, during which the monkey held its hands stationary at the origins and waited for
the target/s to appear. We call this time window the hold period. The second window
contained the time from 250 ms before to 1000 ms after movement onset, thus
including the time for motor preparation and movement execution. We call this time
window the movement period. For all possible pair-wise combinations of
simultaneously recorded LFPs, we calculated the correlograms obtained during these
two time windows, using time delays between -200 and +200 ms, and a bin width of
2.5 ms, which was our sampling resolution time. Correlation strength was expressed
as the correlation coefficient. The correlograms obtained for all trials recorded under
the same conditions (during the same movement type or during the hold period) were
averaged to reduce noise. Examples of such correlograms are shown in Fig. 10. From
each correlogram, we determined the maximum value. Negative correlations
constituted only about one percent of our total sample and were not analyzed
separately. For correlation peaks > 0.1, we also determined the time lag of the
maximum correlation.
The conventional crosscorrelation is affected both by the similarities of the average
signal (here, the mEPs) in the two electrodes and by possible trial-wise interactions
between them.
A common way to distinguish between these two aspects is to
calculate a ‘shift-predictor’ to approximate the correlation between the averages, and
then subtract it from the correlograms to derive an estimate of the ”pure” trial-wise
correlation. We calculated the shift-predictor by correlating the i-th trial of one
electrode with the i+1-st trial of the other, and the last trial with the first trial.
64
The shift predictor was also used to define a confidence limit for oscillatory
components in crosscorrelograms. Whenever we found at least one satellite peak (on
each side of the main correlogram peak), which exceeded these values, we scored a
correlogram as having a significant oscillatory component. As a confidence limit for
oscillatory correlation, we chose the mean (over all delays) of the predictor+/- 3
standard deviations (again, over delays).
Results
Before describing the results, it is important to mention an experimental problem. For
one of the monkeys (Monkey P) it was quite clear that neural recordings from the area
we considered SMA, was problematic. The monkey is still in its home cage and thus
final conclusions are limited. Yet, we have several evidence suggesting that we did
not ablate a significant portion of SMA, and maybe did not even record in the
classical area called SMA-proper. The cells in the sampled area were rare, smaller
then in other monkeys, and their activity was unstable. This may explain the different
results we got after the „SMA lesions“ between the two monkeys. In the other
monkey, both neural activity and SMA lesion were confirmed to be as planned, and
therefore the major conclusion relating to SMA activity and SMA lesion must be
confined to Monkey G.
65
Behavior
After SMA lesion, transient deficits can be seen in the performance of raisins and
center-out tasks. Fig. 3B shows this result demonstrating that Monkey G took
significantly more time to complete the raisin task after SMA lesion.
Fig 3: Raisin task
A. Two kinds of raisin boards: Board I (big wells) controls proximal
reaching;
Board II (small wells) is more related to distal collecting
B. Performance of raisin task before and after SMA lesion. Black – is
normal, yellow – after left SMA lesion, red – after right SMA lesion.
There was a transient decrease in performance rate of the reaching task, especially for
the bimanual movements (Fig. 4). Although some transient effects are evident in the
unimanual movements as well, the strongest effect can be seen in the “bimanual long”
movements. Unfortunately, we did not test Monkey G for the performance of
66
Fig 4: Average success rate before and after SMA lesion. The success rate
was calculated for each day for each type of movement, according to the
formula:
(successful trials)/(all attempted trials), and then averaged for
similar types of trials. Black – is normal, yellow – after left SMA lesion,
red – after right SMA lesion, green – five months after lesion, called
recovery.
“unimanual long” movements. Therefore, we cannot claim that this effect is “purely”
bimanual.
This monkey showed a significant increase in RT and time to peak of velocity in
unimanual forward movements, especially of the dominant hand, and an increase in
RT in all types of bimanual movements (Fig. 5). In bimanual trials for some types of
movement, both hands were affected after unilateral SMA lesions. Significance was
tested by Mann-Whitney test: asterisks indicate a significant difference after-lesion
67
Fig 5: Reaction time in unimanual and bimanual “short movement” changes after SMA lesion.
Asterisks indicate significant changes in RT (p<0.01), tested by the non-parametric Mann-Whitney
test. Black – is normal, yellow – after left SMA lesion, red – after right SMA lesion, green - five months
after lesion, called recovery.
A. Unimanual movements. Each bar represents the mean RT for each period. As can be seen, RT
increased after SMA lesion, mostly for the forward movements.
B. Bimanual movements. Each bar represents the mean RT for each period; the first bar of each
color represents the RT for the right hand, the second bar color represents the RT for the left
hand.
state compared with the normal state. MT does not show any significant changes after
SMA lesions.
Significant changes were also observed after lesion in the velocity profiles (Fig. 6,7),
although there was no significant difference between the velocity profiles of bimanual
and unimanual movement in normal and lesion states (Fig. 6).
68
Fig 6: Velocity profiles of bimanual movements
A. Velocity profiles before and after SMA lesion. Colors as in the previous pictures:
black is normal, yellow - after left SMA lesion, red - after right SMA lesion,
green - recovery.
B. Comparison of bimanual velocity profiles to unimanual movements constructing
them.
Monkey P behavioral data fluctuated considerably, so we were unable to discern any
significant behavioral changes.
69
Fig 7: Velocity profiles in unimanual movements before and after SMA lesion.
The changes for the forward direction right hand were found significant in both time to peak and
peak velocity (p<0.01, Mann Whitney test). Black – is normal, yellow – after left SMA lesion, red
– after right SMA lesion, green - recovery.
A. Mean time to peak velocity for each period, plotted in star plot for each movement
direction.
B. Mean of velocity peak value for each period plotted in star plot.
Neural activity
Monkey G showed a significant increase in firing rate, especially in the left
(dominant) hemisphere, in MI after SMA lesion (Fig. 8). This rise persisted for
several months after the lesion was induced. Monkey P did not exhibit a significant
change in firing rate, probably because from the onset of the experiment it was twice
as high as normally found.
Monkey G preparatory activity before the bimanual movements was significantly
prolonged after SMA lesion. This effect was even stronger for non-symmetrical
bimanual movements (Fig. 9). The significance of this change using the MannWhitney test was very high: p<4e-25. In contrast, following the lesion of the cortical
tissue (probably in cingulated), Monkey P does not give the similar result. In both
70
Fig 8: Firing rate before and after SMA lesion.
The average firing rate for each hemisphere before and after lesion is shown.
Asterisks indicate a significant difference from normal (p<0.01,MannWhitney test)
A. Monkey G; black is normal, red – after bilateral SMA lesion, green recovery
B. Monkey P; black is normal, red – after bilateral SMA lesion.
Recovery period was not recorded in this monkey.
monkeys the duration of the preparatory activity before unimanual movement did not
revealed any consistent change following SMA lesion.
Fig 9: Preparatory activity onset before and after SMA lesion. Averaged onsets of
neural activity in bimanual symmetric and non-symmetric movements are plotted
relative to movement onset. Black bar is normal, red bar - after bilateral SMA
lesion. Asterisks indicate a the significant difference from normal (p<0.01, MannWhitney test)
A. Monkey Galileo
B. Monkey Poisson
71
Directional cell tuning in MI changed significantly after SMA lesion in Monkey G
(Table1). The percentage of cells tuned to both ipsilateral and contralateral arms
increased significantly after SMA lesion. Monkey P did not exhibit similar changes.
Monkey G normal
Contra
Ipsi
Bilateral
Task related
Total
In % task related
Contra
Ipsi
Bilateral
Monkey G after SMA lesion
Contra
Ipsi
Bilateral
Task related
Total
In % task related
Contra
Ipsi
bilateral
Monkey P normal
Contra
Ipsi
Bilateral
Task related
total
In % of task related
Contra
Ipsi
bilateral
Monkey P after lesion
Contra
Ipsi
Bilateral
Task related
Total
In % of task related
Contra
Ipsi
bilateral
Right hemisphere
18
10
7
56
80
Left hemisphere
18
2
5
65
89
32.14 %
17.87 %
12.5 %
27.69 %
3.08 %
7.69 %
8
3-4
19
48
53
8
3
5
31
38
16.67 %
8.33 %
39.58 %
25.81 %
9.68 %
16.13 %
Right hemisphere
95
9
38
157
164
Left hemisphere
91
7
40
170
188
60.51 %
5.73 %
21.2 %
53.53 %
4.12 %
23.53 %
9
4
11
39
45
25
2
7
37
49
23.08 %
10.26 %
28.21 %
67.57 %
5.41 %
18.92 %
Table 1:Lateralization of tuning in MI before and after lesion
This tuning was calculated by eye on the whole data (before the division it to the stable
parts). The cells counted as bilateral were excluded from ipsilateral and contralateral.
In addition to the directional tuning, we measured the relative strength of contralateral
versus ipsilateral responses during unimanual center-out movements. We called this
measurement the Laterality index. In both monkeys the laterality index in the MI was
72
significantly lower after SMA lesion (Fig. 10). This finding was consistent in both
hemispheres. This means that after SMA lesion both sides of the brain respond with a
similar magnitude to each side of the body.
Fig 10: Laterality index distribution before and after SMA
lesion.
In both monkeys the decrease in the laterality index after the
lesion was significant according to chi-square (p<0.01)
A. Laterality index distribution in Galileo; black is
normal, red - after bilateral SMA lesion, green –
recovery.
B. Laterality index distribution in Poisson; black is
normal, red - after bilateral SMA lesion. The
recovery period was not recorded.
After SMA lesion, Monkey G showed slightly more bilateral somato-sensory
responses (66.6% of responses were bilateral after the lesion compared to 56.2% in
normal monkey). While in Monkey P the lesion had a much smaller effect (47% of
responses were bilateral after the lesion compared to 37.7% in normal monkey). It
important to mention that in the normal and SMA-lesioned monkeys, the passive
sensory-motor responses were frequently bilateral.
These electrophysiological findings, together with the directional tuning changes can
probably account for the common “mirror movement” symptom found in SMA lesion
patients.
This raises the question: where does this MI bilaterality, which is mostly exposed in
the absence of SMA, originate? There are two likely explanations:
73
1. Bilaterality always existed in the MI, but in normal monkey it obscured by the
SMA.
2. After SMA lesion, MI activity is modified to take some of the SMA functions.
One way to address these issues is to check the changes in inter- and intrahemispheric network connectivity before and after SMA lesion. Both monkeys
showed a significant increase in LFP correlations between the hemispheres, while
inter-hemispheric LFP correlations did not change significantly. (Fig 11, 12, 13 ).
Fig 11: Example of LFP correlation matrix for one normal day
(Monkey G)
74
Fig 12: Example of LFP correlation matrix for one day after SMA lesion
(Monkey G).
The increased intra-hemispheric connection observed in Monkey G, returned to
normal within several months of SMA lesion.
75
Fig 13: LFP max cross-correlation coefficients before and after SMA lesion.
Maximal LFP correlation coefficients were plotted before (black) and after lesion
(red). As can be seen, both monkeys showed significant increase in intrahemispheric correlation after the lesion, whereas the correlations within the same
hemisphere did not change.
A. Monkey G
B. Monkey P
Intermediate summary and discussion
For purpose of the discussion, we will use the results we obtained from Monkey G,
since the lesion in Monkey P, did not hit SMA
It is known that the various motor areas are interconnected anatomically in reciprocal
connections. Therefore it is reasonable to assume that even in simple motor tasks
several motor areas communicate and operate in parallel during preparation of
movement. Therefore, even if we neutralize (by a lesion) one part of the brain, this
may not fully explain its function, since other areas remain intact. Further, it is known
that the brain most important quality its plasticity, namely – the ability to adapt and
compensate for tissue loss. Our SMA lesion experiment, while still preliminary,
demonstrates this fact in several ways and also demonstrates how the remaining areas
(the two MIs) adapt after the lesion. Yet, as any lesion study it can only provide hints
about the real way MI and SMA operate together.
76
It should be noted that in contrast to previous SMA lesion research in monkeys
(Brinkman et. al., 1981,84) and stroke studies in humans (Viallet et. al., 1992; Chan
et. al., 1988), the ablations we made in the SMA were minimal. The small size of the
lesion may also explain how the monkey recovered completely from the lesion, and
due to the relative fast recovery process allowed us to study neural mechanisms that
may be involved in this recovery process.
After SMA lesion, there were some transient behavioral difficulties, although most of
them reverted to normal 5 months after the injury. Indeed, we found several changes
in neural activity, which partly rebound to the normal 5 months after SMA lesion. To
summarize the neural activity changes, observed in MI after SMA lesion:
a. Higher in single cell firing rate
b. Longer time of preparatory activity
c. Greater number of bilaterally tuned cells
d. Lower laterality index of single unit activity
e. Greater LFP correlation between hemispheres
Let us assume that the MI, as a lower region of the brain, is related largely to coding
“bimanual default,” that is symmetrical movements. If so, it should be similarly active
when it controls the ipsilateral and the contralateral sides of the body. However, we
know that this is not the case! The results of figure 10 and table 1, suggest that in fact
it is still possible that MI is potentially could function in a more symmetric way, but
normally, the SMA veils this tendency. Our results, showing that MI cells become
more symmetric after lesion suggest that SMA breaks the symmetry, possibly by
inhibiting activity that could activate the ipsilateral arm. An increase in an intrahemispheric connectivity further enhances this trend. This may also explain why after
SMA lesion in humans they tend to to perform “mirror movements.”
Some of the behavioral deficits and changes in neural activity after SMA lesion were
similar to those in MPTP-treated monkeys (a monkey model of Parkinson’s disease).
The increase in RT for forward directions after SMA lesion was similar to the RT
changes in MPTP-treated monkey (Camarata et. al., 1992).
77
Brain imaging studies in humans showed that in PD patients, compared with normal
controls, there is a relatively reduced signal in the rostral part of the SMA (Playford
et. al., 1992, Rascol et. al., 1992, Sabatini et. al., 2000), together with a significant
bilateral increase in the activity of the primary sensory-motor cortex, lateral premotor
cortex, caudal part of the SMA and anterior cingulated cortex (Sabatini et. al., 2000).
All these findings suggest that some of the symptoms observed in Parkinson patients
are a result or reduction of excitatory inputs from the BG to the cortex, thereby
decreasing firing in SMA, resembling the partial SMA lesion we introduced in our
experiment.
Most of the behavioral deficits we observed were relatively small and transient. This
finding is not surprising, given our knowledge of brain plasticity, and in particular
given the studies in humans who show the remarkable neural recovery after SMA
lesion. The most interesting finding in our study was the neuronal activity changes
reflecting the plasticity of the neural system and, probably, some of the ways to which
the system resorts to restore the function of the lesioned parts. For example, the
augmented correlation between the hemispheres may reflect neural network
reformation processes. In turn, the increased inter-hemispheric connectivity, may
contribute to the greater control of both sides of the body by each hemisphere.
78
Chapter D: Comparison of right and left arm-related MI.
Introduction.
Hand preference is perhaps the most evident behavioral asymmetry observed in
humans (Bryden et. al., 1977, 2000, Sainburg and Kalakanis, 2000) and monkeys.
Anatomic brain asymmetries found in humans (Amunts et. al., 1996,1997, 2000) and
monkeys (Nudo et. al., 1992) may be associated with hand preference, but a
functional relationship between asymmetries of the motor system and hand preference
has not been established (for review see Haaland and Harrington, 1996).
It has been shown that several motor areas participate in planning the movement of
one (Alexander et. al., 1990, Gourgopolous et. al., 1982, 1995, Scott et. al., 1997, Fu
et. al., 1995) and both arms (Toyokura et. al., 1999, Tanji et. al., 1988, Donchin et. al.,
1998, Kermadi et. al., 2000). In the early stages of this project we provided evidence
that each of the two hemispheres contain and may be able to contribute to the control
of movements of both arms (Donchin et. al., 1998, 2001 a&b, Steinberg et. al., 2001,
Cardoso et. al., 2001). Further, our psychophysical experiments suggest a common
mechanism for initiation of movements of the two arms, which may be responsible for
coupling between the arms at movement initiation. These mechanisms allow for the
non- dominant arm to move faster in bimanual, compared with unimanual trials.
There have been other psychophysical studies of inter-hand coupling (Boessenkool
et. al., 1999, Swinnen et. al., 1996, 1997, Franz et. al., 1996, 1997, 2001, Bogaerts et.
al., 2001) and dominant hand leading (Byblow et. al., 2000). In recent years have
witnessed a surge of human brain imaging studies comparing the activity of both
hemispheres during the performance of motor tasks (Kawashima et. al., 1998, van
Mier et. al., 1998, Jenkins et. al., 1997, Sabatini et. al., 1993, Yahagi and Kasai, 1999,
Civari et. al., 2000). However, the role of inter-hemisphere dominance in motor
planning is still obscure.
In the present study we compared MI neural activity in each hemisphere while
79
monkeys perform unimanual and bimanual reaching movements, in effort to:
(1) Elucidate the nature of functional differences between hemispheres and
(2) Study the behavioral correlates of these differences.
Methods
The behavioral tasks
The behavioral tasks were the same as described above. Three macaque rhesus
monkeys were trained to perform proximal center-out tasks, where we also recorded
neuronal activity. One monkey (monkey G) participated in two different behavioral
and recording sessions, which were analyzed separately.
Two of the three monkeys also performed a “more natural” raisin task were they were
required to pick up raisins from the big and small holes of plastic boards (Brinkman
et. al., 1981). The reason for using the different boards was that boards with big holes
were used for proximal reaching task, whereas small holes in different orientations
were used as a task involving distal precise grasping.
Hand Preference
The hand preference of each monkey was evaluated using several methods.
1. The reaction time (RT) was compared between the two hands during the
performance of unimanual right and left center-out reaching movements.
2. RT was compared between the two hands in the performance of bimanual
center-out task.
3. The percentage of bimanual trials, which were initiated by the left and right
hand, was calculated.
4. The percentage of raisin reaching and grasping, which was initiated by the left
and right hand, was calculated.
80
Activation of neuronal population
Average firing rate was calculated in a time window of -200:300 ms around the
movement onset, defined according to the Amos Arieli movement onset algorithm
described previously (Donchin et. al., 1998,2001; Cardoso et. al., 2001, Gribova et.
al., 2001). The significance of differences in firing rates between the hemispheres
was tested using Mann Whitney non-rank analysis. (Sidney Siegel, Nonparametric
statistics for the behavioral science, 1956). A compound peri-stimulus time histogram
(PSTH) was calculated by averaging the PSTH of all the cells for the different kinds
of trials.
Directional tuning in different movement types
Directional tuning of single cells was defined on the basis of neural activity during
the performance of unimanual center-out movements. Cell PSTHs were plotted for
unimanual left and unimanual right movements in eight directions. Tuning preferred
direction was determined both by eye and by fitting the data to cosine (using R2 as
described by Gourgopolous et. al., 1982). To compute the R2 we averaged the firing
rate in a 750 ms response window of -250:500 ms before and after movement onset
(defined as off-line). Cells that were directionally tuned only to the contralateral arm
were defined as contralateral; cells that were directionally tuned only to the ipsilateral
arm were defined as ipsilateral; cells that were tuned to both the ipsilateral and
contralateral arms were defined as bilateral. Cells counted as bilateral were not
included in the ipsilateral and contralateral groups.
LFP Directional tuning
LFP directional tuning was evaluated using two parameters: the depth of modulation
(max(LFP)-min(LFP)) in the response window or the sum of the RMS signal in the
response window. The response window was defined as -200:300 ms around
movement onset. Finally, the data was fitted to cosine using the R2 as for single units.
81
Laterality index
To assess the degree of laterality of neuronal activity we computed a “laterality
index” based on for the average firing rate (for single cells) and the LFP signal during
the unimanual movements in a window –200:300 ms around movement onset. The
laterality index is given by:
Laterality index = (ContraResponse-IpsiResponse)/(ContraResponse+IpsiResponse)
As the equation shows, the values of the laterality index range from –1 to1. For
Cells/LFP that mainly respond to the contralateral side, the value approaches 1, while
for cells/LFP that relate mostly to the ipsilateral arm the values reach -1. This measure
is not related to the directional tuning, but to the magnitude of the response. The
significance of the differences in laterality index distributions between the two
hemispheres was tested according to chi-square method. The calculation of chisquare takes into account the interdependency of the response of the same cell to eight
different directions of unimanual movement.
Onset time of neuronal activity
To compare onsets of movement to the beginning of significant neuronal activation,
we determined the onset of neural activity changes for each cell, using the CUSUM
algorithm (Ellaway, 1977, Davey et. al., 1986). This calculation was made on the
basis of the average firing rate within a time window of 200 ms before to 300 ms after
movement onset (as defined by the offline detection algorithm). We used two
thresholds, the stricter one being 4 x the confidence limit, as defined in Ellaway
(1977) and a less strict one, 1.5 x the confidence limit. Onsets were limited to time
from target appearance to 400 ms after movement initiation. Distribution of activity
onsets in the MI in one side of the brain was compared with that at the MI in the other
side. The significance of the difference was tested, using the non-parametric Mann
Whitney test.
82
Crosscorrelation analysis
Cross-correlation analysis was applied to the LFP signal, as described above for SMA
lesions. (See page 37)
83
Results
1. Hand preference – behavioral results
According to the hand preference test, all three monkeys were defined as right-handed
(Fig. 1), the magnitude of hand dominance of the individual monkey varied from task
to task. For example, according to raisins task, monkey P appeared to be strongly
right-handed, while in the center-out task she seemed to be a near balance between the
left and right side. This monkey exhibited the smallest inter-hand interval (IHI) in the
bimanual trials.
Fig. 1: Hand preference in monkeys. A.: The plots in left column show the percentage of usage of the
right (black bars) and left (white bars) hands in the two raisin tasks (average values over 10 days for
monkey G and 23 days for monkey P), and the percentage of movements in which each hand was first to
move in the bimanual movement. Monkey F, did not perform the raisin task. B.: The right column shows
the average RT for each hand in unimanual and bimanual movements (white - left hand, black bars –
right hand). Error bars display the standard deviations; the asterisks indicate a significant between-hand
difference (p<0.01, Mann-Whitney test).
2. Firing rate and movement related activation of single unit and LFP
A comparison of average firing rates showed a slight non-significant difference
between the two MI in the behaving monkey. All three monkeys revealed a higher
firing rate for the dominant (left) hemisphere. The compound PSTH consistently
shows that in the bimanual movements the left (dominant) hemisphere is significantly
84
more active then the right hemisphere in all three monkeys (Fig.2). This finding
supports the above-described one, which was weaker mostly because it was averaged
for all kinds of trials.
Another unexpected result, seen in the compound PSTH, is that the left (dominant)
hemisphere was active both during the contralateral and ipsilateral unimanual
movements, while the right (non-dominant) hemisphere was active mostly during the
contralateral unimanual movements (Fig. 2).
Fig. 2: Compound PSTH in monkeys. Average PSTH, relative to the
movement onset, is plotted for each hemisphere activity in unimanual right,
left and bimanual trials separately. Red is right hemisphere, green is left
hemisphere.
The comparison of medians of LFP modulation in the right and left hemisphere
showed that the modulation was significantly (p<<0.01) higher in the left (dominant)
hemisphere in all 3 monkeys, Fig. 3.
Fig. 3: Median of LFP modulation
The median of LFP modulation is plotted in the
percentage of the maximal value between the
left (black bar) and right (white bar)
hemisphere for each monkey.
All monkeys showed significantly (p<<0.01,
Mann Whitney test) higher modulation for the
left (dominant) hemisphere.
85
Two of the three monkeys showed a higher presentation of the dominant (right) hand.
The percentage of ipsilaterally tuned cells in the right (non-dominant) hemisphere was
higher then in the left. Monkey P showed a more similar distribution of tuned cell
percentage in each hemisphere, probably due to weak hand dominance in the centerout task. (Table 1)
Monkey F normal
In % of task related
Contra
ipsi
bilateral
Monkey G (session2)
In % task related
Contra
Ipsi
Bilateral
Monkey G (session 1)
In % task related
Contra
Ipsi
Bilateral
Monkey P normal
In % of task related
Contra
Ipsi
bilateral
Right hemisphere
42.65 %
14.71 %
17.65 %
Left hemisphere
(27 %)
(18%)
(18%)
Right hemisphere
32.14 %
17.87 %
12.5 %
(25%)
(15.87%)
(8%)
27.69 %
3.08 %
7.69 %
(50%)
(15.63%)
(6%)
Left hemisphere
(38%)
(22%)
(9%)
Right hemisphere
60.51 %
5.73 %
21.2 %
(40%)
(9%)
(3%)
Left hemisphere
Right hemisphere
24.32 %
18.92 %
39.19 %
44.09 %
8.6 %
17.2 %
47.06 %
22.06 %
11.76 %
(27%)
(14%)
(9%)
Left hemisphere
(31%)
(14%)
(15%)
53.53 %
4.12 %
23.53 %
(31%)
(16%)
(11%)
Table 1: Lateralization of tuning in MI.
The first number in the table is calculated using eye measurement, while the number in
brackets is calculated by R2 measurement with threshold 0.7 (window of response is [250:500] msec). The both methods of measurements give similar results. The cells, which
were counted in the bilateral category, were not counted as ipsi and contra.
Both methods of measurement: by eye and by R2 produced similar results. The
difference between the numbers cells obtained using the two measurements is due to
the fact that the R-square method was applied to the data after selecting it for stable
parts, so that the data set was smaller.
Calculation of the tuning of the mean modulation of LFP gave results similar to those
for cells tuning. An even stronger tendency toward high cortical representation of the
dominant (right) arm is evident: the percentage of cells tuned to the contra lateral side
was high in the left hemisphere, whereas the percentage for ipsilateral presentation
was higher in the right hemisphere. Since there was not sufficient stable data for
monkey F, it is not presented in the table. (Table 2) Calculation of LFP tuning, using
86
RMS function instead of depth of LFP modulation, produced very similar results.
Monkey G (2 session)
Contra
Ipsi
Bilateral
Left hemisphere
(22-7)/52 = 28.85%
(10-7)/52 = 5.77%
7/52 = 13.46%
Right hemisphere
(3-1)/52 = 3.85%
(10-1)/52 = 17.31%
1/52 = 1.92%
Monkey G (1 session)
Contra
Ipsi
Bilateral
Left hemisphere
(8-3)/39 = 12.82%
(9-3)/39 = 15.38%
3/39 = 7.69%
Right hemisphere
(3-3)/42 = 0%
(14-3)/42 = 26.19%
3/42 = 7.14%
Monkey P
Contra
Ipsi
Bilateral
Left Hemisphere
(24-5)/116 = 16.38%
(18-5)/116 = 11.21%
5/116 = 4.31%
Right hemisphere
(8-3)/116 = 4.31%
(20-3)/116 = 14.66%
3/116 = 2.59%
Table 2: Mean modulation LFP tuning by R2 method
Threshold is 0.7, window of response taken is [-250:300]msec. The cells, which were counted
in the bilateral category, were not counted as ipsi and contra.
Laterality index calculation of LFP RMS function revealed a similar tendency of the
non-dominant hemisphere to respond with more similar strength to the contralateral
and ipsilateral sides of the body then the dominant hemisphere. Distribution of the
Laterality index in most of the monkeys showed significantly lower indexes for the
right hemisphere than for the left (dominant) one (Fig. 4). Again, monkey P exhibited
a non-significant difference between the hemispheres, probably due to weak hand
dominance.
Fig. 4: Laterality indices distribution of LFP.
Distribution of the laterality index of LFP RMS
function for each hemisphere is displayed
separately. Each line represents a different
monkey; left column - left hemisphere MI, right
column - right MI. The red line plotted on each
distribution indicates the zero indexes where the
response to the ipsilateral and contralateral side
of the body was equal. For each monkey, the p
value of the difference between the right and left
hemisphere is indicated in red letters (Mann
Whitney non-parametric test).
The same result can
87
be seen in the cell laterality index, while in the cells themselves the significance of the
inter-hemispheric difference was lower. (Fig. 5
Fig. 5: Laterality indices distribution of single units.
Distribution of the laterality index of cells firing rate
for each hemisphere is displayed separately. Each line
represents a different monkey; left column - left
hemisphere MI, right column - right MI. The red line
plotted on each distribution indicates the zero index
where the response to the ipsilateral and contralateral
side of the body was equal. For each monkey, the p
value of the difference between the right and left
hemisphere is indicated in red letters (Mann Whitney
non-parametric test).
PSTH onset distribution did not show any consistently significant difference between
the right and left MI (Table 3).
Monkey F
All
Less then 0
Right hemisphere
-80.29 (-60.32)
-140.45 (-92.35)
Left hemisphere
-112.83 (-146.9)
-152.76 (-150.66)
significance
0.0278
0.0642
Monkey G (1st session)
All
Less then 0
Right hemisphere
-58.86 (-57.53)
-124.56 (-107.95)
Left hemisphere
-40.74 (-56.16)
-102.76 (-98.84)
significance
0.3712
0.0596
Monkey P
All
Less then 0
Right hemisphere
-117.24 (-108.57)
-116.53 (-124.64)
Left hemisphere
-114.49 (-69.28)
-156.38 (-107.96)
significance
0.2744
0.1204
Table 3: Mean (and median) of neural activity (cells) onsets in left and right
hemispheres MI.
Neural activity onsets were taken relative to the beginning of movement calculated off-line.
First line in the table indicates the mean (median) of the all PSTH onsets, while the second
line take into account only the cells, which were significantly activated before the movement
onset.
(The significance was tested by Mann Whitney test between the means through all trial-types
of the cells)
88
3. Intra-hemispheric and inter-hemispheric interactions
Cross-correlation analysis in 2 monkeys (Monkey G 2sess., Monkey P) showed that
intra-hemispheric LFP cross-correlations were slightly, but significantly, higher in the
left hemisphere (Fig. 6).
Fig6: Mean of LFP cross-correlations
The means LFP cross-correlations peak
values calculated separately for each
hemisphere of two monkeys are plotted.
The black bar is for the right MI, gray –
left MI and white for the correlations
between the right and left hemispheres.
Intermediate summary discussion and conclusions
Our work shows that there is a functional difference between the arm-related areas of
the left and right MI. This difference most probably originates in the behavioral hand
difference observed in the same monkeys. However, it bears mention that the
neuronal asymmetry between the right and left MI was much weaker than the
behavioral hand preference. This weak expression of hand preference in neuronal
activity could stem from a real physiological difference or might be a byproduct of
insufficient neuronal data. Further works with neuronal recordings from behaving
monkeys should resolve this issue.
We found that the tuning as well as the laterality index of both the cells and the LFP
reflects control of both sides of the body by the right (non-dominant) hemisphere. In
89
the LFP signal the result are more prominent. For example, it is clear from comparing
figure 4 to figure 5 that the distribution of laterality indices for LFP is narrower as
compared to the distribution for cells. This suggests that the inputs to a given cortical
cell are generally more bilateral (small laterality indices), and these inputs are
“filtered” by different MI cells – the majority of which sends as output – stronger
contra lateral signals. The result may also partly account for the better performance
of the non-dominant hand in bimanual movements than in unimanual ones.
We did not find any difference in neuronal activity related to different types of
bimanual movements. This is probably due to the fact that at the recording stage, all
the monkeys were over-trained. It is known that over-training makes movements more
automatic. Conceivably motor-learning experiments now in progress in our laboratory
will produce different results.
90
General conclusion and discussion:
Section 1: Overall summary and conclusions.
This study, along with other experiments conducted in our laboratory, aimed at
elucidating the nature of bimanual coordination and its underling neuronal substrate.
The major significance of this work is in demonstrating that bimanual movements are
generated and represented in the brain as a distinct entity, rather then a composition of
two unimanual movements. This notion is very different from the classical
“contralateral” approach, which is widely taught and used among neural scientists.
The idea that bimanual movement is planned as a whole may explain the “bimanual
coupling” paradigm, which is now being studied very intensively. The hypothesis that
planning of bimanual movement is one entity also leads to a more holistic view of the
motor planning problem, it preferences and restrictions.
Using the non-structured scribbling task in humans, we further demonstrated the
phenomenon of bimanual coupling. In most previous works the subjects were tested
while making highly structured movements. In the scribbling task the movements of
each arm seem to be very chaotic. Yet, we demonstrated that here as well, the two
arms naturally couple, even no restrictions were imposed. We advanced one step
forward by showing that this non-symmetric bimanual movement is unexpectedly
hard to perform and requires special learning. These simple experiments reveal the
existence of a common plan for both hands, which satisfies the performance of most
kinds of movements in our daily life. A possible reason for this common plan is its
simplification of real time movement computation.
In the center-out task in humans and monkeys, we continued to investigate the
dynamics of inter-arm coupling and found that this coupling is high even before
movement starts, gradually decreasing toward the end of the movement. This result
provides experimental evidence that movements of the two arms are planned initially
by a common central mechanism and this supports the hypothesis that bimanual
movements are represented in the brain as a distinct entity. To achieve appropriate
91
motor behavior, decoupling mechanisms must come into action as the movement is
executed, even in bimanual movements. This is necessary to adjust the movements to
different loads, obstacles and to differences in muscular strength of the two arms. Our
results support the notion that decoupling is mediated by somato-sensory feedback
from the limbs, and possibly also by visual feedback. The neural level at which these
decoupling correction are performed is a subject for additional research. However, a
possible neural correlate at the cortical level is demonstrated in our study as well;
While the LFP cross-correlations high before the movement onset, they decreased as
the movement progressed. Interestingly, the time scales for the decrease in inter-arm
coupling and the LFP cross-correlation were very similar, further supporting the
notion that modulation of interhemisphric correlation is modulated by the feedback to
achieve the necessary corrections for each arm by decoupling the centrally generated
commands.
In the same center-out task we compared bimanual performance with the unimanual
movement constructing it, and showed that the bimanual movement does not have a
higher RT then the unimanual one. Furthermore, we found that preparatory neural
activity in monkeys is significantly longer for unimanual non-dominant movements
and equal for bimanual and unimanual dominant movements. These findings provide
additional support to the claim that bimanual movements are not a combination of
unimanual ones, and that it is the unimanual movement, which requires additional
processing – most likely by inhibition imposed on the non-moving arm.
The study of SMA lesions revealed transient changes in behavior, followed by
compensatory changes in MI. In the absence of a functioning SMA:
a. Firing rate of MI cells was increased, supporting the hypothesis that that SMA
is a source of some kind of inhibition, which may play a role in decoupling the
arms.
b. The number of bilaterally tuned cells in the MI increased and the magnitude of
neural response to both sides of the body become more similar. These changes
reveal the existence or development of more symmetrical control of both sides
of the body by the MI.
92
c. MI prolongs preparatory activity before movement onset. This may be due to
compensation for the missing functional areas.
d. The inter-hemispheric connectivity between the two spared motor areas (MI)
is increased - This may both enhance neural network reformation and
contribute to the bilateral control of the body by each hemisphere. The result
may also support the hypothesis that SMA induces de-correlation and thus
decoupling.
Cumulatively, our results suggest that in the absence of SMA, MI is even more
involved in the control of both sides of the body. Adding to this the discovery
“bimanual related activity” in the normal MI, our findings further support the notion
that motor planning starts with, a
“default plan” of a “bimanual symmetric”
movement, which is modulated to produce other types of movements (where
symmetry may break) by interactions between SMA and MI which includes inhibitory
decoupling influences from SMA to MI.
A comparison of the right and left MI exposed, as expected, various functional
differences between the two sides of the brain. However, at our hands, these were
small, and call for further research.
In sum, we showed here that that the two arms tend to be coupled in scribbling as well
as and center-out task, in both humans and monkeys and demonstrated neural
correlates to these phenomena. We also investigated behaviorally and physiologically
the decoupling mechanisms and suggest that it evolve during movement execution by
sensory feedback and mediated by inter-hemispheric interactions and inhibitory
influence of SMA on MI. As a whole, these results support my working hypothesis
which holds that the primary motor cortex attempts to initiate by default, a symmetric
movements of the two arms (using common plan for the two sides), while sensory
inputs together or via SMA influences can modulate and modify this plan during its
execution.
93
Section 2: Significance and possible clinical implementations
I would like to outline some practical outcomes, which might emerge from the results,
the conclusions and the philosophy of this work:
1. Parkinson’s’ disease: Our study of SMA lesions in monkeys shows that there
are some similar behavioral deficits in monkeys after SMA lesion and after
MPTP treatment (animal model of Parkinson’s disease). This finding, together
with the similar coordination pattern found in PD patients and SMA lesion
patients (difficulty to perform different movements by both arms
simultaneously), support the notion that some of PD symptoms originate in
weaker activation of the SMA by BG and not the BG themselves. These
findings may change present approach to the treatment of PD patients. For
example, a “long-shot” suggestion may be to monitor SMA activity and
stimulate when necessary.
2. Hemiplegia and CVA induced paralysis: Hemiplegia is motor deficit where
some limbs on one side of the body are either weak or completely paralyzed.
In many cases this was caused by damages to one side of the brain (mainly in
the cortex and white matter of motor areas, depriving the motor system of
cortical contribution to motor control). There are many stroke patients with
damage to one side of the motor cortical area. As a result, they suffer from a
weakness or the inability to move an arm/leg in the contralateral side.
Assuming that “bimanual symmetric default” exists, it should be possible to
more efficiently improve the control over the weak part of the body by
training subjects to attempt to make, or even just imagine they make bimanual
movements. This may facilitate ipsilateral activation in the spared hemisphere
and help restoring motor functions to the impaired side of the body. As an
anecdote, I would like to bring here the results of some preliminary
experiments. Using this approach, I taught my left hand to write in three
94
languages (Hebrew, Russian, English). I used the following algorithm: first
write with the left (non-dominant) hand, then with both hands together, and
then again with the left hand. I found that following a week or two of such
exercises it was possible to bring the non-dominant hand to a better writing
level. However, I also found that the improvement could only last by
continuing the training. A short protocol of this experiment is enclosed in
Appendix 0. For some people, who were born left-handed and who were
forced to use the right hand, this algorithm may be useful for reversing the
habit. Psychologists attach a great deal of importance to this possibility,
because they contend, that the original changing of hand preference may lead
to mental complications.
An additional advantage in enhancing the use of the non-dominant hand is derived
from Oriental medicine and Yoga, where the idea is to bring the body to the most
balanced form. The Feldenkrais method of therapy, taken from Yoga, is based on
the same notion. There is another kind of mental therapy, called Brain Gym (Carla
Hannaford, Smart Movies: Why learning is not all in your head, Great Ocean
Publishers, 1995), which is based on enhancement of the inter-hemispheric
connections in order to improve mental and intellectual capabilities. Training of
the non-dominant hand, partly using the bimanual network will no doubt produce
a stronger connectivity between the two hemispheres. Therefore, if we relate to
the brain as a whole system, higher order abilities may be gained, in addition to
the improvement of some motor abilities.
95
Reference List
Aglioti S., Berlucchi G., Pallini R., Rossi G.F., Tassarini G. (1993): Hemispheric
control of unilateral and bilateral respnses to lateralized light stimuli after callosotomy
and in callosal agenesis. Exp. Brain Res. 95: 151-165
Amunts K., Schlaug G., Schleicher A., Steinmetz H., Dabringhaus A., Roland P.E.,
Zilles K. (1996) Asymmetry in human motor cortex and handedness. Neuroimage 4(3
Pt 1): 216-22
Amunts K., Schmidt-Passos F., Schleicher A., Zilles K. (1997): Postnatal
development of interhemispheric asymmetry in the cytoarchitecture of human area 4.
Anat. Embriol. (Berl) 196(5): 393-402
Amunts K., Jancke L., Mohleberg H., Steinmetz H., Zilles K. (2000):
Interhemispheric asymmetry of the human motor cortex related to handedness and
gender. Neuropsychologia 38(3): 304-12
Anson J.G., Bird N. (1993): Neuromotor programmimg: Bilateral and Unilateral
effects on simple reaction time Human Movement Science 12: 37-50
Alexander G. E. and Crutcher M. D.. (1990): Neural representations of the target
(goal) of visually guided arm movements in three motor areas of the monkey.
J.Neurophysiol. 64 (1): 164-178
Bellgrove M.A., Bradshaw J.L., Velakoulis D., Johnson K.A., Rogers M.A., Smith
D., Pantelis C. (2001): bimanual coordination in chronic schizophrenia. Brain Cogn
45(3): 325-341
1
Bergman H., Feingold A., Nini A., Raz A., Slovin H., Abeles M., and Vaadia E.
(1998) Physiological aspects of information processing in the basal ganglia of normal
and parkinsonian primates. Trends in Neurosciences 21 (1): 32-38
Bernstein N. (1967): The coordination and regulation of movements. Oxford,
England: Pergamon Press
Bernshtein N. (1990) Physiology of movement. Moskow, Nayka press
Boessenkool J.J., Nijhof E.-J., Erkelens C.J. (1999): Variability and correlations in bimanual pointing movements. Hum. Movmnt Sci., 18, 525-552
Bogaerts H., Swinnen S.P. (2001): Spatial interactions during bimanual coordination
patterns: the effect of directional compatibility. Motor Control 5(2): 183-99
Bresson, F., Maury, L., Pieraut-Le Bonniex, G., & de Shonen, S. (1977): Organization
and lateralization of reaching in infants: An instance of asymmetric functions in hands
collaboration. Neuropsychologia, 15: 311-320
Brinkman C. and Porter R. (1979): Supplementary motor area in the monkey: activity
of neurons during performance of a learned motor task. J.Neurophysiol. 42(3): 681709
Brinkman C. (1981): Lesions in supplementary motor area interfere with a monkey's
performance of a bimanual coordination task. Neurosci.Lett. 27(3): 267-270
Brinkman C. (1984): Supplementary motor area of the monkey's cerebral cortex:
short- and long-term deficits after unilateral ablation and the effects of subsequent
callosal section. J.Neurosci. 4 (4): 918-929
Brooks D. J., Salmon E. P., Mathias C. J., Quinn N., Leenders K. L., Bannister R.,
Marsden C. D., and Frackowiak R. S. (1990): The relationship between locomotor
2
disability, autonomic dysfunction, and the integrity of the striatal dopaminergic
system in patients with multiple system atrophy, pure autonomic failure, and
Parkinson's disease, studied with PET. Brain 113 (Pt 5): 1539-1552
Brown R. G., Jahanshahi M., and Marsden C. D. (1993): The execution of bimanual
movements in patients with Parkinson's, Huntington's and cerebellar disease [see
comments]. J.Neurol.Neurosurg.Psychiatry 56(3): 295-297
Bryden P.J., Pryde K.M., Roy E.A. (2000): A performance measure of the degree of
hand preference. Brain Cogn 44(3): 402-14
Buchanan J.J., Kelso J.A., Guzman G.C. (1997): Self-organization of trajectory
formation. 1. Experimental evidence. Biol. Cybern. 76: 257-273
Byblow W.D., Bysouth-Young D., Summers J.J., Carson R.G. (1998): Performance
asymmetries and coupling dynamics in the acquisition of multifrequency bimanual
coordination. Psychological Research 61: 56-70
Byblow W.D., Lewis G.N., Stienar J.W., Austin N.J., Lynch M. (2000): The
subdominant hand increases in the efficacy of voluntary alterations in bimanual
coordination. Experimental Brain Research 131: 366-374
Camarata P.J., Parker R. G., Park S.K., Haines S.J., Turner D.A., Chae H., Ebner T.J.
(1992): Effect of MPTP induced hemiparkinsonism on the kinematics of a twodimensional, multijoint arm movement in the rhesus monkey Neuroscience 48(3):
607-619
Cardoso de Oliveira S., Gribova A., Donchin O., Bergman H. and Vaadia E.
(submitted): Neural interactions between motor cortical hemispheres during bimanual
and unimanual arm movements
3
Chan J. L. and Ross E. D. (1988): Left-handed mirror writing following right anterior
cerebral artery infarction: evidence for nonmirror transformation of motor programs
by right supplementary motor area. Neurology 38(1): 59-63
Chen C., Thaler D., Nixon P.D., Stern C.E., Passingham E. (1995): The function of
the medial premotor cortex. Brain Res. 102: 461-473
Civari C., Cavalli A., Naldi P., Varrasi C., Cantello R. (2000): Hemispheric
asymmetries of cortici-cortical connections in human hand motor areas. Clin.
Neurophysiol. 111(4): 624-9
Corbetta, D., Thelen, E. (1996): The developmental origins of bimanual coordination:
A dynamics perspective. Journal of Experimental Psychology: Human Perception
and Performance 22(2): 502-522
Crutcher M. D. and Alexander G. E. (1990): Movement-related neuronal activity
selectively coding either direction or muscle pattern in three motor areas of the
monkey. J.Neurophysiol. 64 (1): 151-163
Cunningham H. A. (1989): Aiming error under transformed spatial mappings suggests
a structure for visual-motor maps. J.Exp.Psychol.Hum.Percept.Perform. 15(3): 493506
Davey N.J., Ellaway P.H., Stein R.B. (1986): Statistical limits for detecting change in
the cumulative sum derivative of the peristimulus time histogram. J. Neurosci.
Methods 17: 153-166
Deecke L., Lang W., Heller H. J., Hufnagl M., and Kornhuber H. H. (1987):
Bereitschaftspotential in patients with unilateral lesions of the supplementary motor
area. J.Neurol.Neurosurg.Psychiatry 50(11): 1430-1434
De Long M.R. (1990): Primate models of movement disorders of basal ganglia origin.
Trends Neeurosci. 13: 281-285
4
Dick J. P., Benecke R., Rothwell J. C., Day B. L., and Marsden C. D. (1986): Simple
and complex movements in a patient with infarction of the right supplementary motor
area. Mov.Disord. 1(4): 255-266
Donchin O., Gribova A., Steinberg O., Bergman H., Vaadia E. (1998): Primary motor
cortex is involved in bimanual coordination. Nature 395: 274-278.
Donchin O., Cardoso de Oliveira S., Vaadia E. (1999): Who tells one hand what the
other is doing: the neurophysiology of bimanual movements. Neuron 23: 15-18
Donchin O., Gribova A., Steinberg O., Bergman H., Cardoso de Oliveira S., Vaadia
E. (2001): Local field potentials related to bimanual movements in the primary and
supplementary motor cortices. Exp. Brain Res. 140: 45-55
Donchin O., Gribova A., Steinberg O., Mitz R.A., Bergman H. and Vaadia E.
(submitted): Single-Unit Activity Related to Bimanual Arm Movements in the
Primary and Supplementary Motor Cortices
De Guzman G. C. and Kelso J. A. (1991): Multifrequency behavioral patterns and the
phase attractive circle map. Biol.Cybern. 64(6): 485-495
Easton T. (1972): On the normal use of reflexes. American Scientist 60: 591-599
Eliassen J.C., Baynes K., Gazzaniga M.S. (2000): Anterior and posterior callosal
contributions to simultaneous bimanual movements of the hands and fingers. Brain
123: 2501-2511
Ellaway P.H. (1977): An application of cumulative sum technique (cusums) to
neurophysiology. J. Physiol. Lond. 265: 1P-2P
5
Edward V. Evarts, Christoph Fromm, Jurgen Kroller, and Von A. Jennings. (1983):
Motor cortex control of finely graded forces. J. Neurophysiol. 49: 1199-1215
Fagard J. and Peze A. (1997): Age changes in interlimb coupling and the development
of bimanual coordination. Journal of Motor Behavior 29(3): 199-208
Fogassi L., Gallese V., Gentilucci M., Luppino G., Matelli M., and Rizzolatti G.
(1994): The fronto-parietal cortex of the prosimian Galago: patterns of cytochrome
oxidase activity and motor maps. Behav.Brain Res. 60(1): 91-113
Franz E.A., Eliassen J.C., Ivry R.B., Gazzaniga M.S. (1996): Dissociation of spatial
and temporal coupling in the bimanual movements of callosotomy patients.
Psychol.Sci. 7: 306-310.
Franz E. (1997): Spatial coupling in the coordination of complex actions. Quart. J.
Exp. Psychol. 50(3): 684-704
Franz E.A., Ramachandran V.S. (1998): Bimanual coupling in amputees with
phantom limbs. Nat. Neuroscience 1(6): 443-444
Franz E.A., Waldie K.E., Smith M.J. (2000): The effect of callosotomy on novel
versus familiar bimanual actions: a neural dissociation between controlled and
automatic processes? Psychol. Sci. 11(1): 82-85
Franz E.A., Zelaznik H.N., Swinnen S.S., Walter C. (2001): Spatial conceptual
influences on the coordination of bimanual actions: when a dual task becomes a single
task. J.Mot. Behav. 33(1): 103-112
Fu Q. G., Flament D., Coltz J. D., and Ebner T. J. (1995): Temporal encoding of
movement kinematics in the discharge of primate primary motor and premotor
neurons. J.Neurophysiol. 73(2): 836-854
6
Garry M.I., Franks I.M. (2000): Reaction time differences in spatially constrained
bilateral and unilateral movements. Exp. Brain Res. 131: 236-243
Gazzaniga M.S. (1998): The split brain revisited. Sci. Am. 279: 50-55
Gazzaniga M.S. (2000): Cerebral specilization and interhemispheric communication.
Does the corpus callosum enable the human condition? Brain 123: 1293-1326
Georgopoulos A. P., Kalaska J. F., Caminiti R., and Massey J. T. (1982): On the
relations between the direction of two-dimensional arm movements and cell discharge
in primate motor cortex. J.Neurosci. 2(11): 1527-1537
Georgopoulos A. P. (1995): Current issues in directional motor control [see
comments]. Trends.Neurosci. 18(11): 506-510
Georgopoulos A. P. and Pellizzer G. (1995): The mental and the neural: psychological
and neural studies of mental rotation and memory scanning. Neuropsychologia.
33(11): 1531-1547
Gessel A., Ames L.B. (1947): The development of handedness in infant interlimb
coordination. The Journal of Genetic Psychology. 70: 155-175
Gribova A., Donchin O., Bergman H., Vaadia E., Cardoso de Oliveira S. (in
preparation): Temporal aspects of bimanual coordination: Neuronal substrates
reflecting behavioral parameters.
Haaland K.H., Harrington D.L. (1996): Hemispheric asymmetry of movement.
Current Opinion in Neurobiology 6: 796-800
Halsband U., Ito N., Tanji J., and Freund H. J. (1993): The role of premotor cortex
and the supplementary motor area in the temporal control of movement in man. Brain
116(Pt 1): 243-266
7
Hannaford C. (1995): Smart movies: Why learning is not all in your head. Great
Ocean Publishers.
Helmuth L. L.and Ivry R. B. (1996): When two hands are better than one: reduced
timing variability during bimanual movements. J.Exp.Psychol.Hum.Percept.Perform.
22(2): 278-293
Heuer H., Spijkers W., Kleinsorge T., van der Loo H. (1998): Period duration of
physical and imaginary movement sequences affects contralateral amplitude
modulation. Q.J. Exp. Psychol. 51(4): 755-779
Hoff B., Arbib M.A. (1993): Models of trajectory formation and temporal interaction
of reach and grasp. J. Motor Behav. 25: 175-192
Jeka J. J., Kelso J. A., and Kiemel T. (1993): Pattern switching in human multilimb
coordination dynamics. Bull.Math.Biol. 55(4): 829-845
Jacobson L.S., Servos P., Goodale M.A., Lassonde M. (1994): Control of proximal
and distal components of prehension in callosal agenesis. Brain 117: 1107-1113
Jeka J. J. and Kelso J. A. (1995): Manipulating symmetry in the coordination
dynamics of human movement. J.Exp.Psychol.Hum.Percept.Perform. 21(2): 360-374
Jenkins I. H., Fernandez W., Playford E. D., Lees A. J., Frackowiak R.S., Passingham
R.E. (1992): Impaired activation of the supplementary motor area in Parkinson's
disease is reversed when akinesia is treated with apomorphine. Ann Neurol 32: 377590
Jenkins I.H., Passingham R.E., Brooks D.J. (1997): The effect of movement
frequency on cerebral actiovation: a positron emission tomography study. J. Neurol.
Sci. 151(2): 195-205
8
Jones E. G., Coulter J. D., Burton H., and Porter R. (1977): Cells of origin and
terminal distribution of corticostriatal fibers arising in the sensory-motor cortex of
monkeys. J.Comp.Neurol. 173(1): 53-80
Kaluzny P., Palmeri A., and Wiesendanger M. (1994): The problem of bimanual
coupling: a reaction time study of simple unimanual and bimanual finger responses.
Electroencephalogr.Clin.Neurophysiol. 93(6): 450-458
Karol E. A. and Pandya D. N. (1971): The distribution of the corpus callosum in the
Rhesus monkey. Brain 94(3): 471-486
Kawashima R., Matsumura m., Sadato N., Neito E., Waki E., Nakamura S.,
Matsunami K., Fukuda H., Yonekura Y. (1998): Regional cerebral blood flow
changes in human brain related to ipsilateral and contralateral complex hand
movements – a PET study. European Journal of Neuroscience 10(7): 2254-60
Kazennikov O., Wicki U., Corboz M., Hyland B., Palmeri A., Rouiller E. M., and
Wiesendanger M. (1994): Temporal structure of a bimanual goal-directed movement
sequence in monkeys. Eur.J.Neurosci. 6(2): 203-210
Kazennikov O., Hyland B., Wicki U., Perrig S., Rouiller E.M., Wiesendanger M.
(1998): Effects of lesions in the mesial frontal cortex on bimanual co- ordination in
monkeys. Neuroscience 85: 703-716
Kazennikov O., Hyland B., Corboz M., Babalian A., Rouiller E.M., Wiesendanger M.
(1999): Neural activity of supplementary and primary motor areas in monkeys and its
relation to bimanual and unimanual movement sequences. Neuroscience 89: 661-674
Kelso J. A., Southard D. L., and Goodman D. (1979): On the coordination of twohanded movements. J.Exp.Psychol.Hum.Percept. 5(2): 229-238
Kelso J. A., Putnam C. A., and Goodman D. (1983): On the space-time structure of
human interlimb co-ordination. Q.J.Exp.Psychol.A. 35(Pt 2): 347-375
9
Kelso J.A. (1984): Phase transitions and critical behavior in human bimanual
coordination. Am. J. Phys., 246, R1000-R1004
Kelso J. A., Buchanan J. J., and Wallace S. A. (1991): Order parameters for the neural
organization of single, multijoint limb movement patterns. Exp.Brain Res. 85(2): 432444
Kelso J. A. and Jeka J. J. (1992): Symmetry breaking dynamics of human multilimb
coordination. J.Exp.Psychol.Hum.Percept.Perform. 18(3): 645-668
Kermadi I., Liu Y., Tempini A., Rouiller E.M. (1997): Effects of reversible
inactivation of the supplementary motor area (SMA) on unimanual grasp and
bimanual pull and grasp performance in monkeys. Somatosens Mot Res 14: 268-280
Kermadi I., Liu Y., Rouiller E.M. (2000): Do bimanual motor actions involve the
dorsal premotor (PMd), cingulated (CMA) and posterior parietal (PC) cortices?
Comparison with primary and supplementary motor cortical areas. Somatosens. Mot.
Res. 17(3): 255-71
Kurata K. and Wise S. P. (1988): Premotor and supplementary motor cortex in rhesus
monkeys: neuronal activity during externally- and internally-instructed motor tasks.
Exp.Brain Res. 72(2): 237-248
Kuypers H. G. and Lawrence D. G. (1967): Cortical projections to the red nucleus and
the brain stem in the Rhesus monkey. Brain Res. 4(2): 151-188
Lang W., Lang M., Podreka I., Steiner M., Uhl F., Suess E., Muller C., and Deecke L.
(1988): DC-potential shifts and regional cerebral blood flow reveal frontal cortex
involvement in human visuomotor learning. Exp.Brain Res. 71(2): 353-364
Lang W., Obrig H., Lindinger G., Cheyne D., and Deecke L. (1990): Supplementary
motor area activation while tapping bimanually different rhythms in musicians.
Exp.Brain Res. 79(3): 504-514
10
Lawrence D. G. and Kuypers H. G. (1968): The functional organization of the motor
system in the monkey. I. The effects of bilateral pyramidal lesions. Brain 91(1): 1-14
Lawrence D. G. and Kuypers H. G. (1968): The functional organization of the motor
system in the monkey. II. The effects of lesions of the descending brain-stem
pathways. Brain 91(1): 15-36
Luppino G., Matelli M., Camarda R. M., Gallese V., and Rizzolatti G. (1991):
Multiple representations of body movements in mesial area 6 and the adjacent
cingulate cortex: an intracortical microstimulation study in the macaque monkey.
J.Comp.Neurol. 311(4): 463-482
Luppino G., Matelli M., Camarda R., and Rizzolatti G. (1994): Corticospinal
projections from mesial frontal and cingulate areas in the monkey. Neuroreport 5(18):
2545-2548
Van Mier H., Tempel L.W., Permutter J.S., Raiche M.E., Petersen S.E. (1998):
Changes in brain activity during motor learning measured with PET: effects of hand
of performance and practice. J.Neurophysiol. 80(4): 2177-99
Murthy V. N. and Fetz E. E. (1992): Coherent 25- to 35-Hz oscillations in the
sensorimotor cortex of awake behaving monkeys. Proc.Natl.Acad.Sci.U.S.A. 89(12):
5670-5674
Murthy V. N. and Fetz E. E. (1996): Synchronization of neurons during local field
potential oscillations in sensorimotor cortex of awake monkeys. J.Neurophysiol.
76(6): 3968-3982
Murthy V. N. and Fetz E. E. (1996): Oscillatory activity in sensorimotor cortex of
awake monkeys: synchronization of local field potentials and relation to behavior.
J.Neurophysiol. 76(6): 3949-3967
11
Nudo R.J., Jenkins W.M., Merzenich M.M., Prejean T., Grenda R. (1992):
Neurophysiological correlates of hand preference in primary motor cortex of adult
squirrel monkeys. J.Neuroscience 12(8): 2918-47
Pandya D. N. and Vignolo L. A. (1971): Intra- and interhemispheric projections of the
precentral, premotor and arcuate areas in the rhesus monkey. Brain Res. 26(2): 217233
Penfield W. and Welch K. (1951): The supplementary motor area of the cerebral
cortex. Arch. Neurol. Psychiat. 66: 289-317
Playford E.D., Jenkins I.H., Passingham R.E., Nutt J., Frackowiak R.S., Brooks D.J.
(1992): Impaired mesial frontal and putamen activation in Parkinson’s desease: a PET
study. Ann Neurol 32: 161-161
Porter R. (1990): The Kugelberg lecture. Brain mechanisms of voluntary motor
commands - a review. Electroencephalogr.Clin.Neurophysiol. 76(4): 282-293
Rascol O., Sabatini U., Chollet F., Celsis P., Montastruc J.L., Marc-Vergnes J.P.
(1992): Supplementary and primary sensory motor area activity in Parkinson’s
desease. Regional cerebral blood flow changes during finger movements and effects
of apomorphine. Arch Neurol 49: 144-148
Rouiller E. M., Liang F., Babalian A., Moret V., and Wiesendanger M. (1994):
Cerebellothalamocortical and pallidothalamocortical projections to the primary and
supplementary motor cortical areas: a multiple tracing study in macaque monkeys.
J.Comp.Neurol. 345(2): 185-213
Rouiller E. M., Babalian A., Kazennikov O., Moret V., Yu X. H., and Wiesendanger
M. (1994): Transcallosal connections of the distal forelimb representations of the
primary and supplementary motor cortical areas in macaque monkeys. Exp.Brain Res.
102(2): 227-243
12
Sabatini U., Chollet E., Rascol O., Celsis P., Rascol A., Lenzi G.L., Marc-Vergnes
J.P. (1993): Effect of side and rate of stimulation on cerebral blood flow changes in
motor areas durng finger movements in humans. J. Cereb. Blood Flow Metab. 13(4):
639-45
Sabatini U., Boulanouar K., Fabre N., Martin F., Carel C., Colonnese C., Bozzao L.,
Berry I., Montastruc J.L., Chollet F., Rascol O. (2000): Cortical motor reorganization
in akinetic patients with Parkinson’s desease. A functional MRI study. Brain 123:
394-403
Sainburg R.L. and Kalakanis D. (2000): Differences in control of limb dynamics
during dominant and nondominant arm reaching. J.Neurophysiol. 83: 2661-2675
Scott S. H., Sergio L. E., and Kalaska J. F. (1997): Reaching movements with similar
hand paths but different arm orientations. II. Activity of individual cells in dorsal
premotor cortex and parietal area 5. J.Neurophysiol. 78(5): 2413-2426
Schmidt R.A. (1975): A schema theory of discrete motor skill learning. Psychological
Review 82: 225-260
Schmidt R.A., Zelaznik H.N., Hawkins B., Frank J.S., Quinn J.T. (1979): Motor
output variability: A theory for the accuracy of rapid motor acts. Psychological
Review 86: 415-451
Schwartz A. B., Kettner R. E., and Georgopoulos A. P. (1988): Primate motor cortex
and free arm movements to visual targets in three-dimensional space. I. Relations
between single cell discharge and direction of movement. J.Neurosci. 8(8): 29132927
Schwhab R.S., Chafetz M.E., Walker S. (1954): Control of two simultaneous
voluntary motor acts in normal and parkinsonism. Arch. Neurol. Psychiatry. 72: 591598
13
Siegal S. (1956): Nonparametric statistics for the behavioral science International
Student’s Edition.
Semjen A., Summers J.J., Cattaert D. (1995): Hand Coordination in bimanual circle
drawing. Journal of Experimental Psychology: Human Performance and Perception
21: 1139-1157
Shimamura N., Sekia T., Ohkuma H., Tabata H., Yagihashi A., Suzuki S. (2001): A
case report of mirror writing with low perfusion of bilateral anterior cerebral arteries.
No To Shinkei 53(6): 567-570
Steinberg O., Donchin O., Gribova A., Cardoso de Oliveira S., Bergman H., Vaadia
E. (submitted): Neuronal Populations in Primary Motor Cortex Encode Bimanual
Arm Movements
Stelmach G. E. and Worringham C. J. (1988): The control of bimanual aiming
movements in Parkinson's disease. J.Neurol.Neurosurg.Psychiatry 51(2): 223-231
Swinnen S. P., Young D. E., Walter C. B., and Serrien D. J. (1991): Control of
asymmetrical bimanual movements. Exp.Brain Res. 85(1): 163-173
Swinnen S. P., Walter C. B., Lee T. D., and Serrien D. J. (1993): Acquiring bimanual
skills: contrasting forms of information feedback for interlimb decoupling.
J.Exp.Psychol.Learn.Mem.Cogn. 19(6): 1328-1344
Swinnen S.P., Jardin K., Meulenbroek R. (1996): Between-limb asynchronies during
bimanual coordination: effects of manual dominance and attentional cueing.
Neuropsychologia 34: 1203-1213
Swinnen S.P., Van Langendonk L., Verschueren S., Peeters G., Dom R., De Weerdt
W. (1997a): Interlimb coordination deficits in patients with Parkinson's disease during
14
the production of two-joint oscillations in the sagittal plane. Mov Disord. 12(6): 95868
Swinnen S.P., Jardin K., Meulenbroek R., Dounskaia N., Hofkens-Van den Brandt M.
(1997b): Egocentric and allocentric constraints in the expression of patterns of
interlimb coordination. Journal of Cognitive Neuroscience 9: 348-377
Tanji J., Kazuhiko O, Kazuko C.S. (1988): Neuronal activity in cortical motor areas
related to ipsilateral, contralateral, and bilateral digit movements of the monkey.
J.Neurophysiol. 60: 325-343
Toyokura M., Muro I., Komiya T., Obara M. (1999): Relation of bimanual
coordination to activation in the sensorimotor cortex and supplementary motor area:
analysis using functional magnetic resonance imaging. Brain Res.Bull. 48: 211-217
Tuller B. and Kelso J. A. (1989): Environmentally-specified patterns of movement
coordination in normal and split-brain subjects. Exp.Brain Res. 75(2): 306-316
Turvey M.T. (1977): Preliminaries to a theory of action with reference to vision. In R.
Shaw & J. Bransford (Eds.): Perceiving, acting, and knowing. Hillsdale, N.J.:
Erlbaum
Uhl F., Kornhuber A. W., Wartberger P., Lindinger G., Lang W., and Deecke L.
(1996): Supplementary motor area in spatial coordination of bilateral movements: a
new aspect to “the SMA debate”? Electroencephalogr.Clin.Neurophysiol. 101(6):
469-477
Vaadia E., Aertsen A., and Nelken I. (1995): “Dynamics of neuronal interactions”
cannot be explained by “neuronal transients”. Proc.R.Soc.Lond.B.Biol.Sci. 261(1362):
407-410
15
Viviani P., Perani D., Grassi F., Bettinardi V., Fazio F. (1998): Hemispheric
asymmetries and bimanual asynchrony in left- and right-handers. Exp.Brain Res. 120:
531-536
Viallet F., Massion J., Massarino R., and Khalil R. (1992): Coordination between
posture and movement in a bimanual load lifting task: putative role of a medial frontal
region including the supplementary motor area. Exp.Brain Res. 88(3): 674-684
Walter C.B., Swinnen S.P. (1990): Asymmetric interlimb interference during the
performance of a dynamic bimanual task. Brain and Cognition 14: 185-200
Welker W.I., Benjamin R.M., Miles R.C. and Woolsey C.N. (1957): Motor effects of
stimulation of cerebral cortex squirrel monkey. J.Neurophysiol. 20: 347-364
Wiesendanger M., Hummelsheim H., Bianchetti M., Chen D. F., Hyland B., Maier V.,
and Wiesendanger R. (1987): Input and output organization of the supplementary
motor area. Ciba.Found.Symp. 132: 40-62
Wiesendanger M., Kaluzny P., Kazennikov O., Palmeri A., and Perrig S. (1994):
Temporal coordination in bimanual actions. Can.J.Physiol.Pharmacol. 72(5): 591-594
Wiesendanger M., Rouiller E. M., Kazennikov O., and Perrig S. (1996): Is the
supplementary motor area a bilaterally organized system? Adv.Neurol. 70: 85-93
Woolsey C.N., Settlage P.N., Meyer D.R., Spencer W., Hamuy T.P. and Travis A.M.
(1952): Pattern localization in precentral and “supplementary” motor areas and their
relation to the concept of premotor area. Res. Publ. Assoc. Res. Nerv. Ment. Dis. 30:
238-264
Yahagi S., Kasai T. (1999): Motor evoked potentials induced by motor imagery reveal
a functional asymmetry of cortical motor control in left- and right-handed human
subjects. Neurosci. Let. 276(3): 185-8
16
Yamanishi J., Kawato M., Suzuki R. (1980): Two coupled oscillators as a model for
coordinated finger tapping by both hands. Biol. Cybern. 37: 219-225
17