| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Mechanical and Aerospace Engineering and Center for Biomedical Engineering, University of California, Irvine, 92697-3975; 2 Center for the Study of Health Effects of Exercise in Children and University of California, Children's Hospital, Irvine, California 92868-3201
Submitted 27 December 2002; accepted in final form 12 May 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
),
elbow flexion (Jansen-Osmann et al.
2002), and rapid goal-directed planar arm movements
(Yan et al. 2000). During
bi-manual unloading tasks, children exhibit less refined timing and more often
utilize muscle co-contraction strategies compared with adults
(Schmitz et al. 2002).
Skillful movement requires the ability of the motor control system to adapt
to a variety of external dynamic environments. Numerous studies have indicated
that the adult human nervous system uses "internal
models"—feedforward neural mappings between limb state and muscle
force—to adapt to altered dynamic environments (e.g.,
Brashers-Krug et al. 1996;
Scheidt et al. 2001;
Shadmehr and Holcomb 1997;
Shadmehr and Mussa-Ivaldi
1994; Thoroughman and Shadmehr
2000). Similarly, children as young as six were recently shown to
implement internal models of motor-applied viscous force fields during 1 df
elbow flexion movements (Jansen-Osmann et
al. 2002). This finding suggests that children use feedforward,
adaptive control strategies like adults and that these strategies are robust
to the increased internal neuromotor noise that is present in the developing
nervous system. It was also recently shown that adults are able to adapt and
compensate for the approximate mean of a noisy robot-applied force field by
using a dual strategy of internal model formation and impedance control
(Takahashi et al. 2001),
indicating that mature adaptive control systems are robust to environmental
noise that more than doubles their movement variability. If the developing
motor controller utilizes the same adaptive mechanisms as the adult
controller, it would be expected that motor adaptation in children should also
be robust to environmental noise as well as internal neuromotor noise,
although performance may still ultimately be limited by neuromotor noise. The
purpose of this study was to test this hypothesis by comparing the motor
performance of children and adults before and after they adapted to variable
force fields applied by a robot that more than doubled their movement
variability.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
|
Protocol
The seated subject held the end effector of a 3 df lightweight robot arm (PHANToM 3.0, SensAble Technologies) with the dominant hand (Fig. 1A). Each subject started with the reaching hand resting on the lap. A computer-controlled light-emitting diode (LED) prompted the subject to raise the hand to a physical "start" target—the tip of a small compliant plastic pointer—positioned two hand widths out from the center of the sternum. After attaining the start target, the computer sounded a tone, prompting the subject to reach out to a similar "finish" target, positioned just inside the boundary of the reaching workspace and aligned with the start target in the anterior direction (i.e., in front of the subject). After the subject attained the finish target, the computer sounded another tone, prompting the subject to return the hand to the lap, where the subject was allowed to rest for 1 s. After each movement the computer provided visual feedback on the reach speed (just right = desired reach time ±5%; too fast; or too slow). The desired reach time was determined from a test conducted at the beginning of the experiment in which the subject performed the same reaching exercise, only reaching as fast as possible to the finish target (20 trials or reaches). To scale the experimental conditions to each subject's "maximum" movement speed, the desired reach time was set to be 118% of the mean of the reach times of the fastest three trials in this test.
|
Subjects were exposed to five sequential dynamic environments, called
"stages" (Fig.
1B). In the first stage (null field 1), the robot did not
actively apply forces to the subject for 20 trials. Two distinct viscous curl
force fields, a mean field and a noise field, were applied in separate trial
blocks (stages 2 and 4) according to the equation
 | (1) |
In the mean field (50 trials), the force was applied according to Eq. 1, where the gain was constant (k = 1) for each reach. The force in the noise field (50 trials) was also applied according to Eq. 1, but the force gain varied across different trials randomly and unpredictably according to a normal distribution with a mean of 1.0 and a SD of 0.5. The gain values were truncated to ±1.0 about the mean. In addition, the gain of the first trial of the noise field was always set at 1.0 to facilitate comparisons with the reaching error in the first trial of the mean field. The effect of the noise field was to apply a different magnitude of force for each reach, but the average magnitude over many reaches differed by no more than 5% across all subject groups (Table 1).
The order of presentation of the noise and mean fields was randomized across subjects. Noise-then-mean (NM) subjects (n = 27) were exposed to the noise field in stage 2 and then to the mean field in stage 4. The order was reversed for mean-then-noise (MN) subjects (n = 28). The subject quantities were balanced between the NM and MN subject groups to control for possible ordering effects (e.g., exposure to the noise field first might alter performance in the mean field or vice versa). Statistical comparison (paired 2-sided t-test) of the MN and NM subject performances found no significant ordering effects for a variety of key measures including variability in the force fields, final error in the field, modeled gain, direct effect magnitude, and aftereffect magnitude.
The third and fifth stages were null fields (50 trials each), for which the robot did not actively apply forces to the subject, allowing measurement of the aftereffect and providing a "washout" of the previous force field. Subjects were given 1-min rests after trials 45 and 145 to avoid fatigue. Note that because the applied force field was velocity dependent and because the desired reach time was scaled to each subject's maximum movement speed, the forces applied by the robot were also scaled to each subject's maximum movement speed. Thus smaller children, who moved more slowly, experienced smaller applied forces.
An estimate of strength for each subject was obtained by measuring the
maximum isometric force generated in the vertical direction by each arm. The
subject lifted the arm upward with as much force as possible with the forearm
strapped to a six-axis force-torque transducer (Assurance Technologies, Theta
Model) that was positioned so that the hand was 
10 cm out from the torso
at the midline and 10 cm above the lap. The maximum voluntary strength was
taken to be the maximum of two attempts
(Table 1). Strength was
evaluated in the vertical, rather than horizontal, direction because the
hardware setup was simpler. Shoulder strength in the vertical and horizontal
directions differs by 20% in adult populations
(Hughes et al. 1999). Strength
was correlated with maximum movement speed (R2 = 0.54,
P < 0.001).
Data analysis
A computer sampled the three-dimensional position of the robot tip (and thus the subject's hand position) at 1,000 Hz as inferred from rotational sensors at the robot joints. Because the force field pushed the hand to the left or right, disturbances to the reaching trajectory were mainly in the horizontal plane. Statistical analysis indicated that trajectories were not significantly disturbed in the vertical direction on initial exposure to or removal of the field. Thus reaching errors were quantified as the area between the trial path and a reference path projected onto the horizontal plane (X-Z plane, Fig. 1A), divided by the distance between the start and finish targets (Fig. 1C). The resulting geometric measure of error is the spatial average lateral deviation away from the reference path and thus does not depend on reach length. For right-handed subjects, reach paths that were to the right of the reference path were given positive values, whereas those to the left were given negative values. The reference path was selected to be the average path of the trials in the last half of null field 1 (trials 11–20). The average was computed by aligning the path data to an initial velocity threshold (75 mm/s) and computing the mean across the corresponding sampling points. For these trials, the subjects had presumably acclimated to using the robot but still had no perturbing force field applied to them. The averaged hand paths across subject groups during the different exposure stages (Fig. 2) were computed in the same fashion over the applicable range of trials and subjects.
|
Several kinematic measures of reaching were used to quantify the subject
response to the force fields. The "direct effect" of a field was
quantified as the reaching error in the first trial of that force field
(trials 21 and 121). The "performance improvement" was quantified
as the percent difference in the reaching error between the first reach in the
field and the mean of the last 20 reaches in the field. The
"aftereffect" was quantified as the reaching error in the first
trial after a force field was removed (trials 71 and 171), normalized by the
direct effect size of the same field to correct for inter-subject variation in
arm impedance. A subject with greater arm impedance would be expected to
exhibit a smaller direct effect as well as aftereffect
(Takahashi et al. 2001).
Normalizing by direct effect would therefore correct for inter-subject
differences in arm endpoint impedance. Statistical analyses described
throughout this paper were repeated using nonnormalized aftereffects and
similar results were obtained. For all analyses, data points exceeding 1.5 SD
away from the group mean were defined as outliers and were removed in a single
iteration prior to statistical testing. The data analysis for the left-handed
subjects was mirror-symmetric about the sagittal plane so that application of
the field always produced a negative reaching error.
As subjects adapted to the force fields and de-adapted after removal of the
force fields, they exhibited a gradual reduction in error with practice. The
rate of adaptation and de-adaptation was determined by fitting a single
exponential curve with a constant offset to the trial series error using a
least squares fit by the Gauss-Jordan method
 | (2) |
is the time constant
of the fit exponential. Because the trial series errors of individual subjects
were highly variable and typically not amenable to curve fitting, the curves
were fit to the averaged trial series data within each age group, obtained by
ensemble averaging the trial series data across subjects in the respective age
group. The rate of adaptation was quantified as the "time
constant" () of the fit exponential.
Subjects adapted to the noise field with repetitive reaching practice. One
measure of adaptive ability in the noise field was the modeled gain. The
modeled gains were quantified by linearly regressing reaching error and field
gain (k) over the last 20 data points in the noise field. The modeled
gain was quantified as the "zero crossing" of the regression
line—that is, the field gain at which subjects minimized their reaching
error (Scheidt et al. 2001;
Takahashi et al. 2001).
The application of slightly different field strengths on each trial in the
noise field allowed an estimate of limb impedance to be made
(Takahashi et al. 2001).
Specifically, impedance was quantified as the slope of the regression line of
the spatial average lateral force magnitude and the spatial average lateral
deviation from baseline (i.e., the reaching error) over the last 20 trials in
the noise field. This slope indicated the relationship between the average
displacement of the hand and the average displacing force and is equivalent to
the stiffness of the limb if the limb behaves like a linear spring in the
perturbation direction. We also calculated limb impedance by regressing time
averaged force against time averaged lateral deviation and peak force against
peak deviation. The calculated impedance values were similar for each
technique.
For analysis purposes, children were grouped into three age categories: ages 6–8, ages 9–12, and ages 13–17 (Table 1). Adults were grouped as ages >17. Because the onset of puberty can vary between subjects, we also performed the data analysis with slightly redefined age groups (ages 6–8, ages 9–11, ages 12–17, and ages >17) and obtained similar results.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
The hand paths before, during, and after application of the noise and mean force fields were similar across groups (Fig. 2). The pattern of reaching errors, quantified as the spatial average of the lateral deviation, was also similar across age groups (Fig. 3). All age groups exhibited an increased reaching error (i.e., a "direct effect") when the forces were initially applied (trials 21 and 121), reduced their trajectory error with practice in the field, and exhibited a mirror-image trajectory error (i.e., an aftereffect) when the forces were unexpectedly removed (trials 71 and 171).
|
Statistical analysis of performance confirmed the use of internal models
and motor adaptation with repetitive reaching practice in all age groups
(Fig. 4). The initial,
unexpected application of the first force field significantly perturbed all
age groups away from baseline (P < 0.001, t-test, both
the noise and mean fields, 1st direct effect). The direct effect magnitudes
did not depend on the age grouping although there was a nearly significant
trend for the direct effect to be smaller for adults
(Fig. 4A, ANOVA linear
contrast, P = 0.09). With repeated reaching practice, all age groups
showed significant performance improvement (paired 1-sided t-test
across subjects; P < 0.001, both fields) that did not depend on
age grouping (Fig. 4B,
ANOVA linear contrast; P = 0.25 noise field; P = 0.26 mean
field). In addition, a linear regression of time constants of the ensemble
averaged trial series error across age groups revealed that adaptive rates in
the force fields (i.e., the rate of error reduction) also did not depend on
age grouping (Fig. 4C;
R2 = 0.44, P = 0.33 noise field;
R2 = 0.28, P = 0.47 mean field). The performance
improvements ultimately resulted in final error values (the average over the
last 20 trials of each field) that did not depend on age grouping
(Fig. 4D, ANOVA linear
contrast, P = 0.22 noise field, P = 0.30 mean field).
Finally, all age groups showed significant aftereffects away from baseline
(t-test; P 
0.001, both fields) on removal of the force
fields. However, the aftereffect magnitude did not depend on age grouping
(Fig. 4E, ANOVA linear
contrast, P = 0.33 noise field, P = 0.63 mean field).
|
Both children and adults formed a model of the approximate mean of the noise field, with the mean modeled gain between 0.71 and 0.84 (Fig. 5A, t-test compared with 0, P < 0.001, for each age group). The modeled gain (i.e., the 0-crossing of the linear regression of error vs. field gain at the end of the noise field) did not depend on age grouping although there was a nearly significant trend for the modeled gain to be smaller for adults (ANOVA linear contrast, P = 0.06). Power (at the 0.05 level of significance for a type I error) was >0.80 for values of modeled gain <0.70 and >0.92 (all children groups).
|
Adults exhibited higher levels of estimated limb impedance (quantified as the slope of the linear regression of the average spatial force strength against reaching error) compared with children both at the beginning (i.e., the 1st trial) and end (i.e., over the last 20 trials) of the noise field (Fig. 5B; ANOVA linear contrast, P < 0.001). Only the ages 9–12 group showed significant increases in limb impedance at the end of the noise field compared with that at the beginning of the field (paired 1-sided t-test, P = 0.001). Impedance increases for the ages 13–17 group approached significance (P = 0.08). Aftereffect size was not significantly smaller after noise field exposure compared with following mean field exposure (paired 1-sided t-test; P = 0.34, ages 6–8; P = 0.14, ages 9–12; P = 0.33, ages 13–17; P = 0.13, ages >17).
Children moved more slowly and exhibited greater movement variability than adults, although the youngest children reduced their variability with practice
Despite the ability of the children to adapt to the mean and noise fields like adults, several key differences in the children's motor performance were apparent. Children moved more slowly than adults as evidenced by a significant increase of reach time with decreasing age grouping (Fig. 6A, ANOVA linear contrast, P < 0.001). There was also a trend for children to de-adapt more slowly in the null fields than adults. A linear regression of the time constants of the ensemble averaged trial series error ("de-adaptation rates") across all age groups showed a significant linear trend (Fig. 4C, R2 = 0.90, P = 0.05, null fields 2 and 3 combined).
|
Children also showed more initial trial-to-trial spatial and temporal variability (SD of reaching error and reach time, respectively, across trials) in their arm movements in the null field. There was a significant linear trend in spatial (Fig. 6B, ANOVA linear contrast across children groups and across all age groups, P < 0.001) and temporal (Fig. 6C, P < 0.001) movement variability with age grouping near the beginning of the experiment (i.e., the last half of null field 1, trials 11–20). The youngest children (ages 6–8) had significantly higher initial movement variability compared with the ages 9–12 group (P < 0.001 spatial, P = 0.02 temporal, Bonferroni-adjusted), the ages 13–17 group (P < 0.001 spatial, P < 0.001 temporal), and to adults (P < 0.001 spatial; P = 0.001 temporal). Consistent with this increased variability, children scored the desired movement time (total number of "just rights") less often than adults (Fig. 6D, ANOVA linear contrast, P < 0.001).
Despite their more highly variable start, the youngest children (ages 6–8) were able to significantly reduce their trial-to-trial spatial (Fig. 6B, paired 1-sided t-test for ages 6–8, P = 0.006) and temporal (Fig. 6C, paired 1-sided t-test for ages 6–8, P = 0.05) variability by the end of the experiment (i.e., over the last 10 trials of null field 3). This reduction resulted in spatial variability levels that were not significantly different from the remaining children groups, although the adult group maintained significantly lower spatial variability compared with the children groups (ANOVA with a planned comparison, P = 0.002 spatial; P = 0.14 temporal). Furthermore, by the end of the experiment, there was no longer a significant linear trend in spatial (ANOVA linear contrast across all children groups, P = 0.99) and temporal (ANOVA linear contrast across all children groups, P = 0.38) variability with age grouping. Consistent with this decrease in variability, the youngest children scored the desired movement time more frequently by the end of the experiment (Fig. 6D, paired one-sided t-test, P = 0.03).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
One caveat in interpreting these results is that the children were
operating proportionally closer to their maximum strength during the force
field perturbation because the magnitude of the field was scaled to movement
speed rather than arm strength, and adults were only 1.5 times faster but 4.5
stronger than the youngest children (Table
1). However, it seems unlikely that this difference would affect
adaptation ability since the peak forces applied to the youngest children were
still <4% of their maximum shoulder strength
(Table 1). In addition, the
given field strengths resulted in an initial kinematic perturbation to
movement that was comparable between groups as measured by the area reaching
error and the maximum lateral deviation of the direct effect
(Fig. 2,
Table 1). If kinematic error is
the signal that drives adaptation (Goodbody
and Wolpert 1998; Scheidt
1998; Scheidt et al.
2000), then the driving signal for adaptation was about the same
magnitude across ages.
Another caveat is that the younger children may have exhibited greater
relative movement variability in the force field due to the relatively greater
force demand. Variability increases with applied force in isometric force
generation in adults (Jones et al.
2002). However, this effect would be expected to be small since,
as noted in the preceding text, the external forces applied to the youngest
subjects were <4% of their maximum strength. Moreover, the children still
exhibited greater variability even in the absence of the field, consistent
with previous reports (Jansen-Osmann et
al. 2002; Kuhtz-Buschbeck et
al. 1998; Yan et al.
2000).
Adaptive control by children: formation of internal models and impedance control
The results of the present study are consistent with recent findings
indicating that children ages 6–10 can form internal models during a
bimanual load-lifting task (Schmitz et al.
2002) and during single degree-of-freedom elbow movements
(Jansen-Osmann et al. 2002).
During bi-manual unloading tasks, the adult neuromotor system forms
anticipatory models of the effects of unloading while attempting to maintain
forearm posture (Massion and Dufosse
1988; Massion et al.
1999) and exhibits aftereffects when the external forces are
unexpectedly removed (Lum et al.
1992). Schmitz et al.
(2002) found that children can
also form such internal models, but they exhibit less refined timing and more
often utilize muscle co-contraction strategies compared with adults. Studying
viscous loading of single degree-of-freedom elbow movements, Jansen-Osmann et
al. (2002) found age-related
differences in aftereffect magnitude and longer de-adaptation rates in young
children and concluded that children formed models, albeit less
precisely—perhaps due to less precise tuning of dynamics parameters. In
the present study, younger children had longer de-adaptation rates, consistent
with results by Jansen-Osmann et al.
(2002), but the aftereffect
magnitude was comparable across all age groups. The apparent adult-like
performance of the children's adaptive control systems in the present study
may be due to differences in the type of movement practiced (i.e., free
reaching in three space versus constrained, 1 df elbow flexion movements), or
the type of perturbation applied (viscous curl field in gravity vs. viscous
load in a gravity-eliminated environment or vs. self-imposed postural
perturbation).
A simple computational process that is robust to noise may underlie both
adult and child motor learning. Adults are able to learn to compensate for the
approximate mean of substantially variable force fields
(Scheidt et al. 2001;
Takahashi et al. 2001). A
linear adaptive parametric model, using information from only a limited number
of previous practice trials, can account for the ability to achieve a sort of
moving average (Scheidt et al.
2001). Preliminary analysis of the data from the present
experiment indicates a similar computational process can adequately model the
performance of children (Takahashi et al.
2002).
Limb impedance, estimated from the differential trajectory errors produced
by the noise field, increased with age. This age-dependent increase is likely
accounted for by the relatively smaller forces experienced by the younger
children compared with adults because muscle stiffness increases with muscle
force (Hunter and Kearney
1982; Mirbagheri et al.
2000; Weiss et al.
1988; Zhang et al.
1998).
Limb impedance did not increase consistently across age groups in response
to the environmental variability of the noise field. Consistent with this
finding, aftereffects were not significantly smaller after exposure to the
noise field compared with after exposure to the mean field
(Fig. 4E). This is in
contrast to previous studies that suggest that the nervous system manages
variable or destabilizing dynamic environments not just by internal model
formation but also by impedance control
(Burdet et al. 2001;
Milner 2002;
Milner and Cloutier 1993;
Takahashi et al. 2001;
Wang et al. 2001). One
possible explanation for the lack of a clear impedance control effect in the
present study is related to the specific design of the reaching task. For
instance, in a previous study conducted in this laboratory
(Takahashi et al. 2001), adult
subjects reached alternately to two targets from a resting position on a
cantilever beam that extended across the lap and exhibited increased impedance
and diminished aftereffects after exposure to a noise field. In the current
study, subjects were required to locate a starting position before reaching
out. The increased muscle activation necessary to raise the arm against
gravity and stabilize the hand to accurately attain a relatively small point
in space may have increased the limb impedance to levels high enough at the
start of movement to mask or render unnecessary further impedance increases
during the movement.
Sources and mutability of movement variability in children
The increased spatial and temporal variability in the movements of young
children are consistent with the results of numerous studies on arm movement
in children (e.g., Jansen-Osmann et al.
2002; Kuhtz-Buschbeck et al.
1998; Yan et al.
2000). There are two general ways that this increased movement
variability might arise. First, it may reflect a fundamental physiological
constraint in the capability of the developing nervous system. For instance,
movement variability has been considered to be a manifestation of inherent
noise in the neuromotor system (Fitts
1954; Schmidt et al.
1979), which may fundamentally constrain motor planning
(Harris and Wolpert 1998).
Such inherent noise may arise for example, from variability in single-neuron
firing patterns, which may in turn be attributable to noise in membrane
biophysical properties (Azouz and Gray
1999; Shadlen and Newsome
1998). Motor neuron recruitment mechanisms may also affect motor
output variability (Jones et al.
2002).
The developmental constraints on such noise mechanisms are unclear.
Structural maturation of motor tracts, including myelination and axon diameter
changes, is an ongoing process through adolescence
(Fietzek et al. 2000;
Muller and Homberg 1992;
Paus et al. 1999). Immaturity
in neural transmission might increase motor variability by affecting the
integrity of neural signals. Alternately, we recently performed simulations of
a population-coding model of movement control that incorporates neural firing
rate variability and summation of responses from broadly tuned neurons
(Reinkensmeyer et al. 2003). These simulations indicate that trial-to-trial
movement variability increases as the population size decreases because the
magnitude of the population vector decreases more quickly than its SD for
decreasing cell populations, provided physiological levels of firing rate
noise (Lee et al. 1998) are
present. Thus in a population-coding framework that incorporates firing-rate
noise, a less-experienced nervous system with fewer directionally tuned cells
would be expected to exhibit greater variability. A third possibility is that
movement variability is greater because children have more difficulty
attending to their movements. For instance,
(Yan and Thomas 2002)
demonstrated that children with attention deficit hyperactivity disorder
exhibit more variable and slower movements compared with control subjects.
However, the youngest children decreased their variability later in
the experiment when presumably they would have more difficulty attending to
the repetitive task. The work of Todorov and Jordan
(2002) suggests that
variability may increase if children have not learned how to optimally
distribute variability in redundant dimensions. Identifying actual
physiological constraint mechanisms in the developing nervous system is an
important future direction.
Second, increased movement variability may reflect a systems-level process
implemented by the developing nervous system for functional benefit (cf.
Manoel Ede and Connolly 1995).
For example, because the motor control system of a child must perform in the
context of continuously changing system parameters (body mass, dimension,
neural properties, etc.), it may have to constantly perform
system-identification procedures to optimize its performance.
System-identification techniques often involve obtaining a rich experience
through a thorough investigation of the configuration space. Thus movement
variability in children may be an intentional feature of such a functional
optimization process. If so, the pediatric neuromotor controller may opt to
reduce intentional noise when presented repeatedly with the same task, trading
off system-identification processes for better performance.
The results of the present study are consistent with the combined presence of these two mechanisms. Young children (ages 6–8) quickly reduced their spatial and temporal variability with practice. This rapid reduction of variability with practice is consistent with a systems-level mode switching rather than alteration in a fundamental physiological constraint, which would not be expected to change appreciably over the short time period measured here. The physiological constraint mechanism would more likely express itself as age-related baseline variability that changes slowly through a developmental process. Consistent with this idea was the inability of children in the present study to reduce their variability to adult levels, even with practice.
Reduction of movement variability with practice has been observed before in
adults during rapid aiming movements
(Abrams and Pratt 1993;
Darling and Cooke 1987;
Gottlieb et al. 1988) and in
rhesus monkeys during planar reaching
(Georgopoulos et al. 1981). The
adult subjects in the present study did not exhibit a reduction of movement
variability during the adaptation portion of the experiment, possibly because
the time frame considered was insufficiently long to observe this effect. The
relatively greater reduction in movement variability in the youngest children
in this study is consistent with the results of Thomas et al.
(2000), who observed a
relatively greater increase in the duration of the primary submovement and
corresponding decrease in jerk for children practicing a rapid aiming
movement.
Implications and directions for future research
This study suggests or reinforces several key ideas for the understanding
of motor control in the developing nervous system. First, regardless of the
mechanisms, this study confirms that children's movement is inherently more
variable than adults even after motor adaptation. Thus increased movement
variability likely plays a key role in children's appearance of incoordination
and more frequent motor accidents even at well-learned dynamic tasks (e.g.,
spilling, tripping). Increased movement variability likely also constrained
the younger children to plan slower movements to consistently attain the
target with a fixed accuracy requirement
(Harris and Wolpert 1998).
Second, the study confirms that the computational processes that support
internal model formation are implemented by the nervous system early in
development and thus likely support not only motor learning of new tasks at a
young age but the continual control adjustments needed to compensate for the
morphological growth associated with development. From this perspective, the
effects of increasing limb size can be viewed as ongoing changes in the force
field induced by limb mechanics; children as young as age 6 can compute the
internal models needed to predictively compensate for this force field. Third,
young children's ability to reduce movement variability more rapidly than
other age groups provides a possible mechanism for the casual observation that
they appear to improve more rapidly than other age groups in specific motor
tasks. In particular, we hypothesize that they may appear to improve more
rapidly not because they form internal models more accurately or quickly than
adults, but because they more quickly reduce their movement variability after
starting from a higher level of variability. Finally, the paradigm and
measures developed in this study might ultimately prove useful in the clinical
setting as a minimally invasive, relatively simple tool to aid in the
diagnosis and treatment of children who have difficulty mastering motor tasks,
for example, due to neurological disorders, or as a predictor of
activity-related, orthopedic injuries that may be linked to increased motor
variability.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests: D. J. Reinkensmeyer, Dept. of Mechanical and Aerospace Engineering, 4200 Engineering Gateway, University of California, Irvine 92697-3975 (E-mail: dreinken{at}uci.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
Azouz R and
Gray CM. Cellular mechanisms contributing to response variability of
cortical neurons in vivo. J Neurosci
19: 2209–2223,
1999.
Brashers-Krug T, Shadmehr R, and Bizzi E. Consolidation in human motor memory. Nature 382: 252–255, 1996.[Medline]
Burdet E, Osu R, Franklin DW, Milner TE, and Kawato M. The central nervous system stabilizes unstable dynamics by learning optimal impedance. Nature 414: 446–449, 2001.[Medline]
Darling WG and Cooke JD. Changes in the variability of movement trajectories with practice. J Mot Behav 19: 291–309, 1987.[ISI][Medline]
Fietzek UM, Heinen F, Berweck S, Maute S, Hufschmidt A, Schulte-Monting J, Lucking CH, and Korinthenberg R. Development of the corticospinal system and hand motor function: central conduction times and motor performance tests. Dev Med Child Neurol 42: 220–227, 2000.[ISI][Medline]
Fitts PM. The information capacity of the human motor system in controlling the amplitude of movement. J Exp Psychol 47: 381–391, 1954.[Medline]
Georgopoulos AP, Kalaska JF, and Massey JT. Spatial
trajectories and reaction times of aimed movements: effects of practice,
uncertainty, and change in target location. J
Neurophysiol 46:
725–743, 1981.
Goodbody SJ and
Wolpert DM. Temporal and amplitude generalization in motor learning.
J Neurophysiol 79:
1825–1838, 1998.
Gottlieb GL, Corcos DM, Jaric S, and Agarwal GC. Practice improves even the simplest movements. Exp Brain Res 73: 436–440, 1988.[ISI][Medline]
Harris CM and Wolpert DM. Signal-dependent noise determines motor planning. Nature 394: 780–784, 1998.[Medline]
Hughes RE,
Johnson ME, O'Driscoll SW, and An KN. Age-related changes in normal
isometric shoulder strength. Am J Sports Med
27: 651–657,
1999.
Hunter IW and Kearney RE. Dynamics of human ankle stiffness: variation with mean ankle torque. J Biomech 15: 747–752, 1982.[ISI][Medline]
Jansen-Osmann P, Richter S, Konczak J, and Kalveram KT. Force adaptation transfers to untrained workspace regions in children: Evidence for developing inverse dynamic motor models. Exp Brain Res 143: 212–220, 2002.[ISI][Medline]
Jones KE,
Hamilton AF, and Wolpert DM. Sources of signal-dependent noise during
isometric force production. J Neurophysiol
88: 1533–1544,
2002.
Kuhtz-Buschbeck JP, Stolze H, Boczek-Funcke A, Johnk K, Heinrichs H, and Illert M. Kinematic analysis of prehension movements in children. Behav Brain Res 93: 131–141, 1998.[ISI][Medline]
Lee D, Port NL,
Kruse W, and Georgopoulos AP. Variability and correlated noise in the
discharge of neurons in motor and parietal areas of primate cortex.
J Neurosci 18:
1161–1170, 1998.
Lum PS, Reinkensmeyer DJ, Lehman SL, Li PY, and Stark LW. Feedforward stabilization in a bimanual unloading task. Exp Brain Res 89: 172–180, 1992.[ISI][Medline]
Manoel Ede J and Connolly KJ. Variability and the development of skilled actions. Int J Psychophysiol 19: 129–147, 1995.[ISI][Medline]
Massion J and
Dufosse M. Coordination between posture and movement— why and how.
News Physiol Sci 3:
88–93, 1988.
Massion J, Ioffe M, Schmitz C, Viallet F, and Gantcheva R. Acquisition of anticipatory postural adjustments in a bimanual load-lifting task: normal and pathological aspects. Exp Brain Res 128: 229–235, 1999.[ISI][Medline]
Milner TE. Adaptation to destabilizing dynamics by means of muscle cocontraction. Exp Brain Res 143: 406–416, 2002.[ISI][Medline]
Milner TE and Cloutier C. Compensation for mechanically unstable loading in voluntary wrist movement. Exp Brain Res 94: 522–532, 1993.[ISI][Medline]
Mirbagheri MM, Barbeau H, and Kearney RE. Intrinsic and reflex contributions to human ankle stiffness: variation with activation level and position. Exp Brain Res 135: 423–436, 2000.[ISI][Medline]
Muller K and Homberg V. Development of speed of repetitive movements in children is determined by structural changes in corticospinal efferents. Neurosci Lett 144: 57–60, 1992.[ISI][Medline]
Paus T,
Zijdenbos A, Worsley K, Collins DL, Blumenthal J, Giedd JN, Rapoport
JL, and Evans AC. Structural maturation of neural pathways in children and
adolescents: in vivo study. Science
283: 1908–1911,
1999.
Reinkensmeyer DJ, Iobbi M, Kahn LE, Kamper DG, and Takahashi CD. Modeling reaching impairment after stroke using a population vector model of movement control that incorporates neural firing rate variability. Neural Comput In press.
Scheidt RA. Effects of Mechanical and Visual Constraints on Motor Adaptation During Voluntary Movement of the Arm (PhD dissertation). Northwestern University, Evanston, IL: 1998.
Scheidt RA,
Dingwell JB, and Mussa-Ivaldi FA. Learning to move amid uncertainty.
J Neurophysiol 86:
971–985, 2001.
Scheidt RA,
Reinkensmeyer DJ, Conditt MA, Rymer WZ, and Mussa-Ivaldi FA. Persistence
of motor adaptation during constrained, multi-joint, arm movements.
J Neurophysiol 84:
853–862, 2000.
Schmidt RA, Zelaznik H, Hawkins B, Frank JS, and Quinn JT Jr. Motor-output variability: a theory for the accuracy of rapid motor acts. Psychol Rev 47: 415–451, 1979.[Medline]
Schmitz C, Martin N, and Assaiante C. Building anticipatory postural adjustment during childhood: a kinematic and electromyographic analysis of unloading in children from 4 to 8 years of age. Exp Brain Res 142: 354–364, 2002.[ISI][Medline]
Shadlen MN and
Newsome WT. The variable discharge of cortical neurons: implications for
connectivity, computation, and information coding. J
Neurosci 18:
3870–3896, 1998.
Shadmehr R and
Holcomb HH. Neural correlates of motor memory consolidation.
Science 277:
821–825, 1997.
Shadmehr R and Mussa-Ivaldi FA. Adaptive representation of dynamics during learning of a motor task. J Neurosci 14: 3208–3224, 1994.[Abstract]
Takahashi CD, Nemet D, Rose-Gottron CM, Larson JK, Cooper DM, and Reinkensmeyer DJ. Computational motor adaptation—a kindergarten skill. Proceedings of the Second Joint Meeting of the IEEE Engineering in Medicine and Biology Society and Biomedical Engineering Society, Oct. 23–26, Houston, TX. p. 2418–2419, 2002.
Takahashi CD,
Scheidt RA, and Reinkensmeyer DJ. Impedance control and internal model
formation when reaching in a randomly varying dynamical environment.
J Neurophysiol 86:
1047–1051, 2001.
Thomas JR, Yan JH, and Stelmach GE. Movement substructures change as a function of practice in children and adults. J Exp Child Psychol 75: 228–244, 2000.[ISI][Medline]
Thoroughman KA and Shadmehr R. Learning of action through adaptive combination of motor primitives [Comment. In: Nature 407: 682–683, 2000]. Nature 407: 742–747, 2000.[Medline]
Todorov E and Jordan MI. Optimal feedback control as a theory of motor coordination. Nat Neurosci 5: 1226–1235, 2002.[ISI][Medline]
Wang T, Dordevic GS, and Shadmehr R. Learning the dynamics of reaching movements results in the modification of arm impedance and long-latency perturbation responses. Biol Cybern 85: 437–448, 2001.[ISI][Medline]
Weiss PL, Hunter IW, and Kearney RE. Human ankle joint stiffness over the full range of muscle activation levels. J Biomech 21: 539–544, 1988.[ISI][Medline]
Yan JH and Thomas JR. Arm movement control: differences between children with and without attention deficit hyperactivity disorder. Res Q Exerc Sport 73: 10–18, 2002.[ISI][Medline]
Yan JH, Thomas JR, Stelmach GE, and Thomas KT. Developmental features of rapid aiming arm movements across the lifespan. J Mot Behav 32: 121–140, 2000.[ISI][Medline]
Zhang LQ, Nuber G, Butler J, Bowen M, and Rymer WZ. In vivo human knee joint dynamic properties as functions of muscle contraction and joint position. J Biomech 31: 71–76, 1998.[ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  M. Maalouf, C. D. Takahashi, D. J. Reinkensmeyer, D. M. Cooper, and J. M. Rho Impaired Motor Control in Patients With Benign Focal Epilepsy of Childhood J Child Neurol, February 1, 2006; 21(2): 157 - 160. [Abstract] [PDF]
|
|
|
|
 |
|
 C. D. Takahashi, D. Nemet, C. M. Rose-Gottron, J. K. Larson, D. M. Cooper, and D. J. Reinkensmeyer Effect of muscle fatigue on internal model formation and retention during reaching with the arm J Appl Physiol, February 1, 2006; 100(2): 695 - 706. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE |
