Discussion location in the muscle [14]. They proposed that central
136 S
average vector is the same as that of the corresponding However, the Extensor carpi radialis muscles are larger.
sum vector, the intermediate vector corresponded to an It is less likely that other muscle activities could be picked
averaging of the extension and abduction vectors. up by the same surface electrodes. Herrmann and Flanders
showed that in the Biceps brachii and deltoid muscles, motor units had best directions that changed gradually with
4. Discussion location in the muscle [14]. They proposed that central
mechanisms would recruit or derecruit those units more 4.1. Motor responses correlate with the vibration-
suited for the production of force in the required direction. induced illusions of movement and with the
It is thus possible that in the Extensor carpi radialis biomechanical actions of the muscles
muscles, motor units whose best direction is oblique are located in parts of the muscle slightly out of range of the
The parameters of an AVR have been shown to correlate microelectrode.
with the velocity of the concomitant kinesthetic illusion It thus appears that during the AVR, the activation of the
[3]. The aim of this study was to analyze the relations Abductor pollicis longus and Extensor carpi radialis mus-
between the direction of the illusory movement perceived cles is directly related to the direction of the kinesthetic
and the direction and other parameters of the AVR EMG illusion and to their respective biomechanical actions.
amplitude, motor unit type, recruitment, and discharge frequency and to determine whether vectorial models
4.2. Motor unit activation according to their slow or could account for the integration of proprioceptive inputs
fast type from several muscles and for the organization of the
elementary motor commands leading to one unified motor Motor units in the two muscles could be divided into
response. two groups: the non-directional units that responded in all
Overall, the results showed that during illusions of conditions, and the directional units that responded only in
movements in three directions the Extensor carpi radialis one or sometimes two conditions. This suggests the
and Abductor pollicis longus muscles were differentially existence of functional groups of motor units within these
activated. The activation of the two muscles correlated muscles. Such groups of motor units, whose activity
with the direction of the kinesthetic illusion and with the depends specifically on the motor task, have been evi-
biomechanical properties of the muscles. denced in muscles in the arm and forearm [10,30]. In the
The Extensor carpi radialis muscles are prime movers Biceps brachii muscle, groups of motor units in different
for wrist extension and for wrist abduction [2]. However, parts of the muscle are recruited in different orders
they contribute relatively more to wrist extension than to depending on the motor tasks [28].
wrist abduction [21]. This is compatible with the fact that Nevertheless, such a functional organization has never
motor units in the Extensor carpi radialis muscles are been demonstrated in the Extensor carpi radialis muscles
gradually fewer to be recruited and fire with gradually or in the Abductor pollicis longus muscle. In fact, Riek and
lower frequencies during illusory extensions, intermediate Bawa [21] showed that all motor units in the Extensor
movements and abduction of the wrist, successively. The carpi radialis muscles are activated, and in the same order,
Abductor pollicis longus muscle is involved in wrist during wrist contractions in extension and abduction
abduction [2] but not in wrist extension. This is compatible directions.
with the fact that motor units in the Abductor pollicis The discrepancy between their results and ours could be
longus muscle are activated mainly during illusions of due to the difference in experimental conditions. In a
wrist abduction and very little during illusions of wrist review article, Enoka reported that the motor unit popula-
movement in the other two directions. tions in hand muscles were active from half maximal
In contrast, during kinesthetic illusions of intermediate voluntary contraction [5]. In our study, motor unit dis-
direction, the surface EMGs recorded from the two mus- charges were recorded during very low force contractions
cles were as high as during kinesthetic illusions in their between 0 and 2.5 N, whereas in Riek and Bawa’s study
own preferred direction. However, motor unit discharge the muscular contractions during which motor unit activity
parameters were very low in the Abductor pollicis longus was recorded ranged more widely. Could we induce
muscle and intermediate in the Extensor carpi radialis stronger contractions, one may suppose that our results
muscles. This discrepancy between surface and in- would converge towards theirs: most motor units would be
tramuscular EMGs may be due to the recording, through recruited, probably in the same order, but with longer
surface electrodes, of activities of muscle fibers within the latencies and probably lower firing frequencies during
recorded muscle that are further away from the microelec- illusory sensations of abduction.
trode tip, or of activities from neighboring muscles. As for the Abductor pollicis longus muscle, Thomas et
Indeed, the Abductor pollicis longus muscle is very thin, al. [29] showed the existence of subgroups of motor units
so we cannot rule out that some activity from the Extensor in the Abductor pollicis brevis muscle. The very similar
pollicis longus was recorded. biomechanical functions of the Abductor pollicis longus
S . Calvin-Figuiere et al. Brain Research 881 2000 128 –138
137
and Abductor pollicis brevis muscles indicates that one and parietal cortex, related to the direction of arm move-
may extrapolate these results to the Abductor pollicis ments, have been described by vector sum [9,15,16,27].
longus muscle. However, these results could also be described by vector
Finally, our results showed that the nondirectional motor averaging [see also 19].
units were slow motor units, whereas the directional ones As we said earlier, we hypothesized that similar vectori-
were fast motor units. Similar results were found in the cat. al models could be proposed to represent kinesthetic
Schieber et al. showed that in multitendoned muscles in cat illusions and AVRs. We expected the motor vector, which
forepaw, subgroups of motor units were specifically acti- represents the motor response, to be broken down into as
vated depending on the direction of the contraction [26]. many subvectors corresponding to the motor commands
These direction-specific motor units were identified as fast directed to the corresponding motoneuron pools. This
motor units [26]. expectation is supported by our results. However, kines-
This result is compatible with the functional properties thetic illusions are best described by vector sum, whereas
of fast and slow motor units [20,25]. Slow motor units are motor responses are best described by vector averaging,
recruited first. Their contractions produce very little force. that is a sum normalized to the number of active muscles.
They are thought to be involved in processes of adjustment So, both computations share the same arithmetic bases.
of muscular viscosity, which could explain that they are Biomechanical constraints may make the averaging step
active in all conditions. Fast motor units are recruited later necessary in motor response production. Indeed, wrist
and their contractions produce more force. They are extension and abduction have a maximal range of 70
8 and thought to be responsible for a more modulated and
20 8, respectively [17]. During circumduction of the wrist,
dynamic production of muscular force, which agrees with the hand trajectory resembles a distorted ellipse rather than
their being recruited during contractions in a specific a circle. Movements intermediate between full extension
direction. and full abduction are therefore best described by vector
averaging than by vector sum. On the other hand, a simpler 4.3. Vectorial model
explanation may be that normalizing the motor output to the number of active muscles produces smoother hand
This study confirms the results by Roll and Gilhodes movements.
[23] showing that separate vibration of two adjacent muscle groups evoked illusory sensations of orthogonal
movements, while their covibration evoked illusory sensa-