The synergetic coupling of mechanically joined limbs such as the lower and the upper arm clearly simplifies motor control. A single motor command may suffice to realize motion towards a new posture of the whole kinematic chain. The coordination be- tween the left and the right hand or both feet, appears more complicated at first sight. They are not directly connected, but only through the whole body. The states of sev- eral intermediate joints, each with multiple degrees of freedom, affect their spatial relations. Moreover, the left and the right side of the body are controlled by differ- ent brain hemispheres. In case of symmetric activities, both sides could be provided with closely related commands to assume spatially congruent or mirrored postures. However, such a parallel control structure would not enable synergistic behavior like reciprocal error compensation, demonstrated in bimanual pointing tasks [77, 238]. Guiard demonstrated the predominance of asymmetric cooperation in mundane bi- manual actions. He argued that both hands are coordinated according to a kinematic chain. Despite the lack of a direct physical link, both hands are considered to operate like two serially assembled motors [115]. Following this model, the non-dominant hand initiates a bimanual action by providing a spatial reference frame for successive
Bilateral Coordination 51 actions of the dominant hand. An often quoted example of this behavior is the active involvement of both hands in writing. One hand holds and occasionally moves a sheet of paper while the other one is writing on it [115].
The kinematic-chain model redefines the concept of human handedness. Instead of a general preference for one of both hands it assumes the prevalence of bimanual in- teraction with asymmetric roles. Moreover, it embeds the actions of both hands in a longer chain of actions that includes the whole body. The non-dominant hand is con- sidered to be specialized in postural support and initial reach, while the dominant one manipulates objects with higher force, rapidity, and precision. Hinckley et al. demonstrated increased accuracy and faster performance in a bimanual target acqui- sition task if the roles were distributed according to Guiard’s model [138]. They also showed that the distribution of roles is less relevant if the task requires less accuracy and haptic guidance is provided.
The different roles are not inseparably tied to the left or right hand. Most people con- sider themselves to be either left- or right-handers, some assign the roles differently depending on the task, and ambidextrous people use both hands interchangeably. In principle, we all can switch the roles of our hands in everyday routines. Most of us will experience a lack of proficiency of both hands in the unfamiliar roles, but with additional effort, we can exchange roles even in asymmetric bilateral actions. When arriving at home with a heavy shopping bag in our stronger hand, for example, we may find ourselves opening the door lock with the non-dominant hand – although this is more difficult. Highly trained tasks or those that do not require high precision can be easily swapped between hands. Proficient tennis players, as another exam- ple, swap hands for a more balanced match against less trained opponents. In this sense, we may also consider antiphase symmetric actions like climbing and walking a special case of asymmetric cooperation with continuously alternating roles of both actors. While one foot is standing on the ground it provides a stable platform for a step of the other one. This step establishes the next intermediate point of support for the same action with reversed roles. If it comes to more difficult steps or those requir- ing more force (e.g. jumping), however, we generally revert to a clear preference of a dominant foot [258, 264].
4.3.1
Bilateral Movement Synchronization
Just as the synergies between directly connected limbs, also bilateral coordination implies temporal synchronization. Several studies have shown that it is difficult to perform repetitive tasks like finger tapping with both hands in different rhythms (e.g. [178, 263]). Instead, people tend to move their limbs synchronously, either in phase or antiphase. Coordinating simultaneous bilateral actions that are truly out of phase is difficult. Musicians train extensively for playing conflicting rhythms concur- rently (i.e. polyrhythms, irrational rhythms). The complex performance can be facil-
itated by restructuring the rhythms and melodies in chunks and phrases that better relate with each other. The accentuations of a three-four time and a four-four time, for example, coincide every 12 quarter notes. Presumably, the beginning, ending, and the adaptation of trained movement sequences involves more cognitive load than its continuation and repetition. Evidence has been found to show that memorized mo- tor programs can be performed without the intervention of sensory feedback [168], while switching tasks and task sequences involves a cognitive effort that becomes apparent in measures of rapidity and accuracy [304].
Besides implicit movement synchronization, also a tendency towards movements in phase at higher frequencies has been repeatedly shown (see [124, 169]). The partici- pants in a study by Kelso, for example, were asked to move both hands in antiphase to the beat of a metronome [169]. When the frequency increased beyond the indi- vidual preference (generally between 1 and 2 Hz), the coordination of both hands fell apart for a few cycles and then stabilized again, but, with movements of both hands in phase. When asked to perform symmetric hand motions, the coordina- tion remained stable across frequencies. One can reproduce these results easily with finger tapping on a table. Kelso argued that these observations could explain how animals automatically adapt their gait to changing speeds.
A number of similar experimental results have indicated that human motor behav- ior may be ruled by the principles of dynamic systems, rather than clearly specified cognitive programs. For coordinated movements of both hands, for example, math- ematical models of coupled oscillators were suggested [124, 307, 371]. Haken et al. demonstrated that such a model also predicts the observed phase transitions from asymmetric to symmetric bimanual movements [124].
An earlier study by Kelso et al. showed that temporal synchronization also occurs in non-repetitive aimed movements [171]. The experiment required two independent aimed movement tasks to be performed with both hands in parallel, but at targets of different size and distance. Fitts’s law predicts that the easier tasks will be performed faster, but instead, both hands performed in synchrony with almost identical move- ment times. Peak velocity and acceleration differed correspondingly to reach targets at different distances, but their profiles over time were almost identical. Kelso et al. suggested that the observed temporal covariation could be explained with shared afferent signals, i.e., shared control over both movements.
In a series of studies with humans and monkeys Wiesendanger et al. demonstrated that also asymmetric bimanual tasks are highly synchronized [364]. One experiment demonstrated how phasic synchronization facilitates posture stability in bi-manual unloading tasks. Corresponding to their observations, a waiter can hold a tray stable while unloading plates and glasses because changes in the muscle tensions of both hands are activated simultaneously (Figure 4.3). Another experiment required the opening of a drawer with one hand and the subsequent removal of an object with the other. For all human participants, motion onset and peak velocity of both hands were
Bilateral Coordination 53 almost in phase with 56 ms delay on average, although the individual goals (drawer, object) were necessarily reached sequentially (about 300 ms difference on average). On the one hand, these observations demonstrate largely parallel execution and tight temporal synchronization of asymmetric bimanual actions. On the other hand, they also show that the initiation and major events (e.g. reaching the individual targets) of the different actions occur one after another. The time differences were small, but, the grasping hand always followed the opening hand. The same synchronization pattern was observed in monkeys.
Figure 4.3: Unloading glasses from a hand-held tray requires tight synchroniza-
tion of both hands to maintain a stable posture despite the changing weight distribution. Wiesendanger et al. demonstrated this bimanual synchroniza- tion on the level of muscle activations as measured with an electromyograph (EMG) [364]. A waiter must achieve a similar quality of synchronization with the actions of another person to allow her taking a glass from the tray.
4.3.2
Synchronization through Perception-Action Coupling
The tendency for the synchronization between different limbs has long been ex- plained with models of shared afferent signals that co-activate linkages of muscles (e.g. [171]). More recently, it has been shown that the coordination, in particu- lar bilateral coordination, may instead occur on the perceptual level (e.g. [96, 228]). Mechsner et al. showed that bilateral movement synchronization is more strongly influenced by visual perception, but also proprioceptive feedback may guide the co- ordination [228]. A particularly convincing observation came from an experiment in which participants operated the rotation of two visible flags (or levers) through cranks invisibly mounted under a table. A gear system applied a 4:3 transmission
rate for one of both mechanisms (Figure 4.4). The participants were instructed to cir- cle both visible flags either in phase or antiphase at the given rhythm of a metronome. While drawing circles at 4:3 frequency ratios is extremely difficult when directly ob- serving the actions of both hands, the participants had no difficulties in performing this motor action with the manipulated visual feedback in this setting. Even a ten- dency of falling into in-phase rotation (of the two visible flags) as known from earlier studies [54, 169] was observed. If the coordination of both hands were governed solely on the basis of shared afferent signals, this behavior could not be explained. As an alternative, Mechsner et al. suggest that perception and action share a common cognitive representation which facilitates movement coordination through percep- tual goals.
Figure 4.4: Bilateral movement synchronization with different transfer func-
tions. Mechsner et al. showed that turning cranks with both hands at different velocities (e.g. in a 4:3 ratio) can be performed without difficulties, if the result- ing visual feedback compensates for the difference [228]. The participants of their study were asked to operate cranks that were mounted under a table. The movements were applied to visible flags (marked disks in this illustration) via gearboxes with different transmission ratios.
Kelso et al. studied pattern formation in the coupling of perception and action us- ing the example of synchronizing movements of a single finger to the beat of a metronome [170]. They found similar coordination dynamics as shown earlier for bilateral limb coordination and a systematic phase shift relative to the difference be- tween the metronome frequency and the participants’ preferred rate of finger flex- ions. As a consequence, they suggested an extended model incorporating the notion of the difference between requested and preferred frequency. The model predicts the synchronization of two systems (i.e. frequency locking) with a phase shift pro- portional to the difference of their eigenfrequencies. If the preferred frequency of the reacting entity (e.g. the moving finger) is smaller than the given one (e.g. the metronome beat), it will lag behind, and otherwise it will lead the sequence. If the difference is too large, synchronization will not occur.
Interpersonal Coordination 55