Journal of Sports Sciences, 200 1, 19, 507 520

E ects of manipulating relative and absolute motion

information during observational learning
of an aiming task
SALEH A. AL-ABO OD ,1* K EITH DAVID S,1 SIM ON J. BENN ETT,1
DEREK ASHFORD 1 and M AN UEL M ARTIN EZ M ARIN 2
1

Psychology Research G roup, Departm ent of E xercise and Sport Science, The M anchester M etropolitan U niversity,
2
Hassall R oad, Alsager ST7 2H L, U K and Facultad de C iencias del Deporte, U niversity of G ranada, G ranada, Spain

Accepted 22 M arch 2001

In the visual perception perspective of observational learning, the manipulation of relative and absolute m otion
inform ation in visual dem onstrations optim ally directs learners search towards appropriate task solutions.
We assessed the e ect of em phasizing transform ational information and removal of structural inform ation
using point-light kinematic displays in approximating the m odel s relative m otion patterns. Participants viewed
computer-sim ulated point-light demonstrations or normal video dem onstrations before and interm ittently
throughout 100 acquisition trials with knowledge of results on an underarm modi ed-dart aim ing task. On the
next day, all participants performed 20 retention trials without demonstrations. The kinem atics of spatial and
temporal coordination and control variables were exam ined relative to the model s action, as well as performance
scores. The results indicated that approxim ation of the m odel s spatial and tem poral coordination and control
patterns was achieved after observation of either type of demonstrations. No di erences were found in movement outcom es. In a second experiment, the e ects of manipulating absolute motion inform ation by slowmotion dem onstrations were exam ined relative to real-time dem onstrations. Real-tim e demonstrations led to
a closer approximation to the model s spatial and tem poral coordination patterns and better outcome scores,
contradicting predictions that slow-motion displays convey intact relative m otion information. We speculate that
the e ect of visual dem onstration speed on action perception and reproduction is a function of task constraints
that is, novelty or familiarity of relative motion of demonstrated activities.
K eywords : control, coordination, inform ational constraints, skill acquisition, visual dem onstrations.

Introduction
An im portant task for sport scientists interested in
m otor skill acquisition is to evaluate the e ectiveness
of di erent types of visual dem onstrations during sports
coaching, training and practice. Scully and N ewell
(1985) conceptually integrated evidence from research
on the perception of biological m otion w ith N ewell s
(1985) fram ework of learning stages (i.e. coordination,
control, skill) and proposed a visual perception perspective on observational learning as an alternative to
traditional theories. Central to this visual perspective
is the tenet that visual dem onstrations should be used
* Author to whom all correspondence should be addressed. e-m ail:
s.a.al-abood@m m u.ac.uk

prim arily to transm it relative m otion inform ation

essential to the task being learned or perform ed. From
this standpoint, it has been argued that one im portant
role of visual demonstrations during skill acquisition
is the m anipulation of this source of m ovem ent
inform ation, so that the search by learners to assem ble
e ective coordination patterns is optim ally directed
towards appropriate task solutions (Scully and N ewell,
1985; W illiam s et al., 1999).
O n the basis of this new theoretical rationale for
observational learning e ects, Scully (1988) suggested
som e practical im plications for the use of visual
dem onstrations in teaching and coaching m otor skills.
We address som e of the im portant questions related
to the argum ents proposed by Scully, in an attem pt to
exam ine the em pirical support for these suggestions.

Journal of Sports Sciences ISSN 0264-041 4 print/ISSN 1466-447 X online

http://ww w.tandf.co.uk/journals

2001 Taylor & Francis Ltd

508
Speci cally, we assessed whether the use of pointlight displays can facilitate further the acquisition of
m ovem ent coordination com pared to norm al videotaped dem onstrations. We also exam ined w hether
slow-m otion videotaped dem onstrations interfere w ith
the pick-up of relative m otion inform ation and the
perception of tem poral param eters of a m odelled action
com pared to real-tim e dem onstrations, as hypothesized
by Scully (1988).
In the rst study reported in this paper, we exam ined
the hypothesis that the use of point-light or kinem atic
displays in visual dem onstrations can aid the pick-up of
relative m otion inform ation, facilitating the acquisition
of m ovem ent coordination underlying the task being
learned (Scully, 1988). T his prediction is based on the
argum ent that kinem atic displays convey only transformational information essential to perceiving biological
m otion and m inim izing structural inform ation.
Previous research on visual perception of hum an
m otion has consistently dem onstrated that transform ational inform ation (i.e. inform ation about the m ovem ent of an individual) conveyed by point-light displays
is su cient for observers to discrim inate between
di erent classes of physical activities, such as walking,
running, cycling, dancing (Johansson, 1973, 1975),
throwing (W illiam s, 1985) and throwing and bow ling
(Scully, 1987). M oreover, the rem oval of structural
infor mation inform ation about the shape, colour,
size and other characteristics of a perform er did not
hinder obser vers identi cation and recognition of
activities. Research has also show n that kinematic displays a ord the perception of dynam ic param eters
of obser ved activities, such as force (Runeson and
Fr ykholm , 1981), speed (Scully, 1987; W illiam s, 1989)
and even technical execution and aesthetic quality in
gym nastics (Scully, 1986). Accordingly, the results of
these perception studies are su ciently com pelling to
consider w hether kinematic point-light displays can aid
perception in the m odelling process (W illiam s et al.,
1999).
To date, few studies have exam ined whether pointlight displays can be at least as e ective as norm al videotaped demonstrations in supporting the perform ance
and learning of m otor skills. The lim ited ndings
suggest that further research is required before practical
im plications can be advised. For exam ple, W illiam s
(1985) exam ined the e ect of presenting point-light
displays and norm al videotaped demonstrations on
observers perform ance of a throwing action consisting
of a sequence of 2 8 sub-m ovem ents. T he results
indicated no di erences in the production of lim b displacem ent and tim ing (phasing) between observers of
point-light displays and norm al videotapes. However,
as there was no retention or transfer test, W illiam s did
not com pare the e ects on lear ning of using di erent

Al-Abood et al.
types of display. T here were also several inherent
m ethodological lim itations, the m ost im portant of
which was that the point-light displays dem onstrated
only the throwing arm without the torso, which m ay
have m ediated observers perception of m ovem ent.
W hen point-light displays are used in perception
research, researchers typically use points representing
the key joints of the w hole body rather than just those
directly involved in the m ovem ent. Furtherm ore,
W illiam s (1985) did not consider how kinem atic
displays in uence m ovem ent outcom e scores. Consequently, in the rst study reported here, we tried to
establish the e ect of point-light displays, com pared
to norm al videotaped dem onstrations, on the learning
of an aim ing task by including m ore practice trials and
a retention test in the design. M oreover, we wished
to determ ine the e ect of these displays on m ovem ent
outcom es as well as m ovem ent kinem atics, which
would allow us to investigate the acquisition of m ovem ent coordination and control, as operationalized by
Scully and N ewell (1985).
In a second study, we exam ined w hether the observation of slow -m otion videotaped dem onstrations has
facilitative or detrim ental e ects on the acquisition
of m ovem ent coordination and control, as well as task
outcom es, com pared to videotaped dem onstrations presented in real-tim e. T his is an issue of som e signi cance
to sport scientists, since it has practical as well as
theoretical im plications for those involved in sports
coaching and training. Previous research on this
issue has been equivocal. For exam ple, N elson (1958)
found no signi cant di erences in outcom e scores when
observing a golf swing between a slow-m otion group
and a real-tim e videotape group. H owever, the slow m otion group did show a greater gain in scores later in
practice, while the real-tim e m odelling group showed a
greater gain earlier. M cG uire (1961) studied the e ect
of dem onstrating som e steps of perform ance on a
pursuit-rotor tracking task in slow m otion while keeping
other steps at real-tim e speed. He found enhanced
perform ance for the item s presented in slow m otion,
whereas perform ance on other item s was im paired.
W illiam s (1985, 1989) assessed the e ects of slowing
down or speeding up the presentation of a throwing
action in norm al videotaped demonstrations or
point-light displays and found that slow -m otion
dem onstrations embedded correct reproduction of
tim ing param eters of an action but had little e ect on
lim b displacement. M ore recently, and using a m ore
com plex action a ballet dance sequence Scully and
C arnegie (1998) found that the observation of slow m otion demonstrations, com pared to norm al-speed
dem onstrations, slightly facilitated the pick-up and
replication of the m odel s coordination function (e.g.
relative tim ing). H owever, slow m otion was found to

509

Visual dem onstrations and skill acquisition

have im paired the pick-up and replication of m ovem ent
control param eters such as m ovem ent tim e and force.
V isual perception research has also revealed m ixed
ndings on the relative e cac y of slow -m otion and realtim e video demonstrations. For exam ple, Barclay et al.
(1978) found that slow m otion hindered action recognition, while Scully (1987) observed little in uence on
action identi cation but an e ect for speed discrim ination. Although C utting and Pro tt (1982) indicated
that real-tim e demonstrations are essential to the perception of a fram e of reference for each activity, Scully
(1988) theorized that slow-m otion displays provide the
sam e relative m otion dem onstrated by norm al videotapes but exagg erate the tem poral param eters of a
m odelled action. In other words, slow-m otion displays
o er intact relative m otion inform ation of an action
but convey distorted `unreal-tim e features of the action,
thus destroying the perception of absolute m otion.
Relating the visual perception perspective on observational learning to N ewell s (1985) m odel of m otor
learning results in two m ajor predictions on the use
of slow -m otion dem onstrations. F irst, it is expected
that slow -m otion will be as e ective as real-tim e
dem onstrations in supporting the reproduction of an
observed coordination pattern. Secondly, slow m otion
is predicted to im pair the approxim ation of control
param eters of the dem onstrated action.
In short, although few perception and action studies
have exam ined the e ectiveness of viewing slow-m otion
as opposed to real-tim e dem onstrations, the evidence is
still inconclusive and further research is required before
practical im plications can be proposed. Accordingly, the
aim of the second experim ent was to address this issue
from a visual perception perspective.

Experim ent 1
The m ain aim of Experim ent 1 was to establish whether
dem onstrating an action with a point-light kinematic
display, com pared to a norm al videotaped display,
would aid further the pick-up of relative m otion
inform ation, facilitating the acquisition of a m ovem ent
coordination pattern. We hypothesized that, if pointlight displays do support the pick-up of relative m otion
inform ation, observers of point-light displays should
show a greater, or at least an equal, approxim ation to
the tem poral and spatial relative m otion patterns
dem onstrated by a m odel than observers of norm al
videotapes of the sam e action. F urtherm ore, in relation
to N ewell s (1985) m odel of m otor learning, we
also w ished to determ ine whether point-light displays
have the potential to support the pick-up of m ovem ent control variables such as m ovem ent tim e and
velocity.

M ethods
Ten m ale participants (m ean age = 24.2
Participants.
years) volunteered for the experim ent. They were all
right-handed and had norm al or corrected-to-norm al
vision. All participants were naive to the exp erim ental
task, provided inform ed consent before participation
and were told that they were free to withdraw at any time.
T he task involved an underarm
Apparatus and task.
throw of a m odi ed dart originally used by Al-Abood
et al. (in press). T he aim of the task was to score as m any
points as possible by aim ing the m odi ed dart w ith the
dom inant arm towards a target dartboard. The target
was a standard dartboard (Unicorn) m odi ed for the
outcom e scoring system of the experim ent. It contained
10 concentric circles. The bullseye had a diam eter of
2.25 cm , w ith each other circle increasing by 2.25 cm
in radius. To provide outcom e scores as a dependent
variable, the bullseye was awarded 10 points, with
each concentric circle radiating out from the bullseye
decreasing by one point so that the outerm ost circle was
worth only one point. T he target was placed on the oor
3 m away from a throwing line. A regular (U nicorn) dart
was also m odi ed by attaching an additional shaft to
the free end of the ight. T his allowed the participants
to hold the dart from its end, perm itting an underhand
aim ing m ovem ent. T he m odi ed dart had a m ass of 40 g
and was 20 cm long (see Fig. 1).
T he task was unfam iliar to all participants and
involved m ultiple biom echanical degrees of freedom ,
perm itting the exam ination of m odelling e ects on
m ovem ent coordination and control (M cD onald et al.,
1989).
An
Data collection and dem onstration preparation.
ELITE on-line m otion analysis system (see Pedotti and
Ferrigno, 1995) was used to collect and analyse m ovem ent kinem atics. Three m arkers were positioned on
three joints of the dom inant arm (i.e. the right upper
lim b) of the m odel and all participants: the acrom ion
process of the shoulder and the lateral condyles of
the elbow and wrist. A fourth m arker was attached to
the m odi ed dart at the additional shaft to determ ine
the release tim e of the dart. T he two-dim ensional
coordinate data were recorded on-line at a sam pling frequency of 100 Hz. After testing, the three-dim ensional
coordinates were reconstructed from the transform ed
two-dim ensional coordinate data of the re ective
m arkers recorded from the two cam eras. Then, the
raw displacement data were ltered w ith a recursive
second-order Butterworth lter w ith a cut-o frequency
of 5 Hz, which was applied twice to negate the phase
shift (Wood, 1982). T he ltered displacement data were
then di erentiated o -line to derive velocity data.

510

Al-Abood et al.

F ig. 1. The underhand modi ed-dart aim ing task.

Finally, the resultant displacement and velocity data

were calculated.
A video cam era (Panasonic, F -15) and recorder
(Sam sung, SV 821K) were used to videotape the m odel
for later dem onstrations. Before being videotaped, the
m odel practised 3000 trials on the exp erim ental task
over 30 days at a rate of 100 trials per day. The m odel
was videotaped during the nal 20 trials on the last
day of practice. T hen, a videotaped colour recording
was edited by selecting six representative trials to be
presented as visual dem onstrations to participants. T he
m ean outcom e score for these selected trials was 8.8 out
of a m axim um possible score of 10 points. T he viewing
tim e of the videotape recording was approxim ately 3
m in. T he view of the demonstrations in this videotape
recording contained the m odel s whole body m ovem ent,
the target and the trajector y of the dart. The videotape
recording also provided auditory knowledge of results
after ever y trial dem onstrated by the m odel. To present
the videotape dem onstrations to the participants, the
videotape recorder was connected to a colour television
(H antarex, screen dim ensions = 65 50 cm).
W hile the m odel was being videotaped, a record of
his m ovem ent kinem atics was also taken via the EL IT E
system following the aforem entioned procedures. T he
m odel s kinematic data were collected for later com parative analyses. Speci cally, for all trials dem onstrated
by the m odel, angular displacement and velocity of the
aim ing arm were calculated from the instant of m ovem ent initiation to the release of the dart.
A com puter-sim ulated point-light demonstration was
created based on the digitized kinem atic data of the six
trials selected for presentation in the norm al videotaped
recording for the m odel. T herefore, the point-light

display provided the sam e transform ational m ovem ent

inform ation dem onstrated by the norm al videotaped
presentation. H owever, no structural inform ation was
available in the point-light display. In the point-light
dem onstration, 21 points represented the key joints
of the m odel s body and two points represented the
dart. Furtherm ore, one segm ent-light and one pointlight represented the target and its centre, respectively.
A com puter-based, rather than videotape, display was
used in the study to elim inate the detrim ental e ect of
`bloom ing that m ay occur as a result of using bright
light sources against a dark background (K ozlowski and
C utting, 1977).
T he participants were assigned at random
Procedure.
to one of two exp erim ental treatm ent groups (n = 5):
norm al-videotape group and the point-light group.
All participants followed the sam e experim ental procedures but di ered with respect to whether they
observed norm al videotaped dem onstrations or pointlight dem onstrations. All participants observed their
respective dem onstrations before practice and interm ittently every 10 consecutive practice trials through out the acquisition session. Before and through out
the retention trials, all participants perform ed without observing any demonstrations. All participants
were instructed to use the inform ation available in the
dem onstrations to help them to perform the task and
to im prove their perform ance. T hey also received prelim inary instructions from the exp erim enter on how to
hold the dart.
Each participant was tested individually in the
presence of the experim enter only. T he study consisted
of two sessions, acquisition and retention, held on two

Visual dem onstrations and skill acquisition

consecutive days; these sessions consisted of 100 and
20 trials, respectively. T he participants were allowed
2 m in rest after every 20 acquisition trials. T he tim e
between each set of 10 consecutive acquisition trials
was the sam e for participants in both groups. Each trial
was initiated by a `ready com m and given by the
experim enter approxim ately 2 s before the `go com m and to start the m ovem ent. All participants received
knowledge of results based on their perform ance on
each trial. Knowledge of results was provided for two
reasons: rst, to determ ine the potency of m odelling
in m otor skill acquisition w hen such knowledge is available; secondly, to generalize the ndings of the present
experim ent to learning and perform ance settings in
which such knowledge is provided.
D ata on m ovem ent outcom es
Dependent measures.
and kinem atics were collected for all acquisition and
retention trials following the data collection procedures
described above. However, for m ovem ent kinem atics
only, the rst six and last six acquisition trials as
well as the rst six retention trials were analysed.
One m ovem ent outcom e m easure and several m ovem ent kinem atic m easures were selected as appropriate
variables to exam ine the predictions advanced in this
study about the e ects of visual dem onstrations on
m ovem ent outcom es, coordination and control.
To determ ine the e ects of
M ovement outcomes.
the type of dem onstrations on m ovem ent outcom es,
m eans and standard deviations (s) of perform ance
scores were calculated to establish the aim ing accuracy
of each participant for each block of 10 trials. This
procedure resulted in 10 acquisition and two retention blocks. T he resultant data were subm itted to
separate (2 groups 12 trial blocks) two-factor analyses
of variance (AN OVA) w ith repeated m easures on the
trial blocks.
To exam ine
M ovement coordination (relative motion).
the e ects of the type of demonstrations on m ovem ent
coordination, we used the relative m otion of upperand lower-arm segm ents of the aim ing arm . Speci cally,
we com puted angular velocities and displacem ents of
arm segm ents to the vertical axis from the start signal
to the instant of dar t release. T he positions of these two
segm ents to the vertical represent the m ost coherent
shape in term s of the topological properties of a throwing arm (Scully, 1987).
To determ ine precisely the approxim ation of participants relative m otion patterns to those of the
m odel, a cross-correlation of recognition coe cient (R )
(Sparrow et al., 1987) was calculated between the
m edian relative m otion values of the m odel and that of
every trial perform ed by a participant and included in

511
the analysis. T he cross-correlation of recognition coe cient is a m easure of the extent to which the angles
between adjacent data points from one angle angle
or velocity velocity plot are sim ilar to angles from
another plot. It is a m easure of sim ilarity between two
coordination patterns. The value of cross-correlation
(R) ranges from - 1.0 to + 1.0 according to the sim ilarity.
As R approaches zero, the velocity velocity or angle
angle relative m otions of two plots becom e increasingly
dissim ilar in shape. In the present study, therefore,
the higher the positive R-value, the greater the approxim ation to the m odel s relative m otion.
To calculate cross-correlations, we followed three
steps. First, one m ovem ent pattern was selected as
representative of the six trials dem onstrated by the
m odel. To determ ine the m odel s m edian pattern, we
plotted all relative m otions of the m odel s six trials
in a single angle angle diagram and velocity velocity
diagram . From these m otions, we selected the m edian
relative m otions of the m odel (see Fig. 2). Secondly,
after selecting the m odel s m edian pattern, all participants trials were tim e-norm alized to the sam e
num ber of data points as the m odel s pattern using a
cubic spline interpolation technique. Finally, a crosscorrelation with a zero-tim e lag was calculated between
each participant s trial and the m odel s m edian trials.
These procedures were rst followed to calculate crosscorrelations for spatial coordination (i.e. angle angle
relations) and were then repeated to com pute tem poral
coordination (i.e. velocity velocity relations).
After calculating cross-correlations, m eans and
standard deviations of cross-correlations were com puted
for each participant for the rst six acquisition trials,
the last six acquisition trials and the rst six retention
trials. Standard deviations were com puted as an index
of within-participant variability in coordination
patterns around the m odel s m edian pattern. T he
cross-correlations were averaged by using a Fischer Ztransform ation procedure. The calculated m eans and
standard deviations of cross-correlations were subm itted to separate (2 groups 3 trial blocks) two-factor
m ultivariate analyses of variance (M AN OVA) with
repeated m easures on the trial block factor.
Although there were various
M ovement control.
kinem atic variables available to exam ine the e ects of
dem onstration type on m ovem ent control, we decided
to focus on those variables re ecting the control of
elbow angle. C hanges in this angle represent the scaling
or param eterization of the coordination function of
upper- and lower-arm segm ents (Scully, 1987). T he
selected variables were: elbow angle at release, elbow
velocity at release and m ovem ent tim e (i.e. the tim e
taken from the start signal to release of the dart). For
each variable, m eans and standard deviations for each

512

Al-Abood et al.

F ig. 2. Angle angle (a) and velocity velocity (b) relative motion patterns for the six trials selected for the m odel. The abbreviations Int, Rev and Rel indicate the instants of m ovem ent initiation, arm reversal and dart release respectively.

block of six trials were calculated for each participant.

Separate m ultivariate analyses of variance (2 groups
3 trial blocks: Acquisition 1, Acquisition 2 and Reten-

tion), w ith repeated m easures on the last factor, were

com puted on the control variables.
Signi cance for all statistical analyses conducted in

513

Visual dem onstrations and skill acquisition

this study was set at a = 0.05. For repeated-m easure
analyses, the Huynh-Feldt correction factor was used
to adjust the degrees of freedom of the univariate F
because of the potential violation of the sphericity
assum ption owing to the m any levels of the repeated
m easures (see Schutz and G essaroli, 1987). F urtherm ore, signi cant m ultivariate analyses of variance were
followed by separate univariate analyses of variance and
discrim inant analysis (i.e. standardized coe cients) to
determ ine w hich of the dependent variables contributed
m ost to signi cant di erences w ithin each factor (Bray
and M axwell, 1985; Schutz and G essaroli, 1987). T he
values reported for the m ultivariate analyses of variance
are W ilks lam bda ratios. Signi cant analyses of variance
were followed by Tukey H SD tests when required.
Results
T he AN OVA on m ean outcom e
M ovement outcome.
scores showed no signi cant e ect for groups (F 1,8 =
1.28, P = 0.29) or groups trial blocks interactions
(F 11,88 = 1.21, P = 0.29). H owever, there was a signi cant m ain e ect for trial blocks (F 11,88 = 2.37, P = 0.01).
Univariate analysis, including a H uynh-Feldt procedure
to adjust the degrees of freedom because of a violation
of the sphericity assum ption, m aintained the trial block
e ect (F 11,88 = 2.37, P = 0.013). Follow -up Tukey H SD
tests on the block factor revealed an im provem ent in
m ovem ent outcom e w ith practice. T he perform ance of
both groups on the second retention block was signi cantly better than that on the rst acquisition block
(P = 0.00). N o other signi cant di erences were found,
although the di erences between the rst acquisition
block and the seventh and eighth acquisition blocks
approached conventionally accepted levels of statistical
signi cance (P = 0.06; see Table 1).
For standard deviations of outcom e scores, there
was no m ain e ect for groups (F 1,8 = 0.291, P = 0.60) or

trial blocks (F 11,88 = 1.14, P = 0.34), or a groups trial

blocks interaction (F 11,88 = 1.04, P = 0.42).
The m ultiM ovement coordin ation (relative motion).
variate analyses of variance com puted on the m eans
of velocity velocity cross-correlations and angle angle
cross-correlations showed no m ain e ect for groups
(W ilks l 2,7 = 0.865, P = 0.60) or trial blocks (W ilks
l 4,5 = 0.296, P = 0.13), or a groups trial blocks interaction (W ilks l 4,5 = 0.655, P = 0.65) (see Table 2).
For standard deviations of cross-correlations, as an
index of intra-individual variability of m ovem ent
coordination, no signi cant di erence was observed
between groups (W ilks l 2,7 = 0.861, P = 0.59) or trial
blocks (W ilks l 4,5 = 0.538, P = 0.46), and there was no
groups trial blocks interaction (W ilks l 4,5 = 0.839,
P = 0.90) (see Table 2).
Table 1. M ovement outcom es as a function of groups and
trial blocks (m ean s )

Trial
block

Norm al-videotape
group

Acq uisition
1
2
3
4
5
6
7
8
9
10

3.20
3.52
4.84
4.84
4.50
5.32
5.24
4.86
5.24
4.76

R etention
1
2

4.66 2.80
5.26 3.00

Table 2. Cross-correlations of recognition coe

and trial blocks (m ean s )

2.96
3.13
3.12
2.97
2.92
3.07
2.92
2.92
2.41
2.86

cient (R ) as a function of groups

Trial block
Acquisition 1

Acquisition 2

Retention

An gle a ngle (R)

Norm al-videotape group
Point-light group
Slow-motion group

0.46 0.10
0.42 0.08
0.39 0.09

0.56 0.06
0.52 0.06
0.43 0.05

0.52 0.09
0.51 0.06
0.42 0.05

Velocity velocity (R )
Norm al-videotape group
Point-light group
Slow-motion group

0.43 0.05
0.42 0.04
0.37 0.04

0.52 0.04
0.48 0.04
0.41 0.03

0.49 0.05
0.45 0.04
0.42 0.03

Point-light
group

3.20
4.06
3.84
3.62
3.92
3.46
4.12
4.52
3.56
4.22

3.21
2.69
2.66
2.45
2.98
2.64
3.33
2.80
2.58
2.59

4.16 3.15
5.18 3.09

514

Al-Abood et al.
Table 3. M ovement control variables as a function of groups and trial blocks (mean s )

Trial block
Acquisition 1

Acquisition 2

Retention

142 1.80
138 4.07
142 8.44

138 5.78
142 4.03
143 4.40

138 1.57
137 3.70
142 4.63

Release velocity (degrees per secon d)

Normal-videotape group
61.8 20.6
Point-light group
104 40.6
Slow-motion group
74.2 42.9

87.6 31.8
84.5 42.4
108 31.1

94.6 41.1
75.2 31.3
72.9 35.5

M ovem ent tim e (m s)

Normal-videotape group
Point-light group
Slow-motion group

1727 86
1922 135
1847 168

1636 74
1718 115
1757 123

Release an gle ( 8 )
Normal-videotape group
Point-light group
Slow-motion group

1469 156
1550 169
1597 367

T he m ultivariate analyses of variM ovement control.

ance on the m eans of elbow release angle, elbow release
velocity and m ovem ent tim e showed no signi cant
di erences between groups (W ilks l 3,6 = 0.921, P =
0.91) or trial blocks (W ilks l 6,3 = 0.152, P = 0.21), and
there was no groups trial blocks interaction (W ilks
l 6,3 = 0.266, P = 0.43). Sim ilarly, there were no signi cant e ects for groups (W ilks l 3,6 = 0.700, P = 0.51)
or trial blocks (W ilks l 6,3 = 0.281, P = 0.45), or a
groups trial blocks interaction (W ilks l 6,3 = 0.368,
P = 0.55), when a M AN OVA was com puted on the
standard deviations of the sam e dependent variables
(see Table 3).
D iscussion
T he aim of Experim ent 1 was to determ ine w hether
point-light kinematic displays, com pared to norm al
videotaped displays, support further the pick-up of relative m otion inform ation, thus facilitating the acquisition
of a m ovem ent coordination pattern. We hypothesized
that if point-light displays do facilitate the pick-up
of relative m otion inform ation, com pared to norm al
videotapes, participants obser ving point-light displays
should show a closer, or at least an equal, approxim ation
to the tem poral and spatial relative m otion patterns
dem onstrated by the m odel. This prediction was
supported in part, since no di erences in approxim ating
the m odel s relative m otion patterns were obser ved
between participants in the two groups. T his suggests
that the transform ational inform ation available in pointlight kinematic displays was su cient to convey relative
m otion inform ation to observers learning an underarm throw. T his is consistent with ndings of visual

perception research for the potency of such displays

in transm itting accurate inform ation about physical
activity identi cation and recognition (Johansson, 1973,
1975; W illiam s, 1985; Scully, 1987).
The analysis of m ovem ent control variables also
showed no group di erences caused by type of visual
dem onstrations. Again, the results suggest that pointlight displays were as e ective as norm al videotape
dem onstrations in conveying inform ation about m ovem ent control. This is also consistent w ith ndings in
visual perception research (Scully, 1987; W illiam s,
1989), indicating that the rem oval of structural inform ation has no detrim ental e ect on the perception of
control-relevant inform ation. The results are also consistent with the work of W illiam s (1985), who found no
di erences in production of m ovem ent tim ing between
observers of point-light displays and norm al dem onstrations. M oreover, the results of the present study extend
W illiam s ndings to learning contexts.
Finally, for achieving the goal of the m ovem ent (i.e.
outcom e scores), the observation of point-light displays
was also found to be as e ective as the obser vation of
norm al videotaped dem onstrations.

Experim ent 2
T he m ain aim of E xp erim ent 2 was to assess the e ect
of slow -m otion demonstrations, com pared to real-tim e
dem onstrations, on the acquisition of m ovem ent coordination and control of an aim ing task. Visual perception research has revealed m ixed results on how
slow m otion demonstrations in uence perception of
m ovem ent characteristics. Furtherm ore, the e ect of

515

Visual dem onstrations and skill acquisition

slow -m otion demonstrations on m ovem ent production
has typically received little attention in the m odelling
literature, although this m ethod of instruction is com m only used by instructors of m otor skills. T his lack
of consideration m ay have been due in part to a lack of
understanding of the nature of m ovem ent inform ation
conveyed by visual demonstrations in traditional perspectives on observational learning. However, Scully
(1988) suggested that slow -m otion displays provide
the sam e relative m otion demonstrated by norm al
videotapes, but distort the tem poral param eters of a
m odelled action. Slow-m otion displays o er intact relative m otion inform ation of an action but convey `unreal
tim e features of that action, perturbing the perception
of absolute m otion. Accordingly, a key prediction on
the use of slow -m otion dem onstrations from a visual
perception perspective is that slow -m otion will be as
e ective as norm al real-tim e dem onstrations for the
reproduction of an obser ved coordination pattern,
although it m ay im pair the approxim ation of control
aspects of the dem onstrated action.
M ethods
Five additional m ale participants (m ean
Participants.
age = 23.2 years), none of whom participated in Experim ent 1, volunteered for this experim ent. All participants
were right-handed and had norm al or corrected-tonorm al vision. They were naive to the experim ental task,
provided inform ed consent before participation and
they were told that they were free to withdraw at any
time.
The task and apparatus were the
Apparatus and task.
sam e as those used in Experim ent 1.
T he
Data collection and demonstration preparation.
procedures used to collect kinem atic data in Experim ent 1 were followed in Experim ent 2. F urtherm ore,
the videotaped demonstrations by the skilled m odel
prepared in Experim ent 1 were presented in slow
m otion for participants in the slow -m otion group in
Experim ent 2. The viewing tim e of the videotape
recording presented in slow m otion was approxim ately
4.5 m in.
The ve additional par ticipants w ho
Procedure.
volunteered for this experim ent were assigned to the
slow -m otion group. To exam ine the predictions
advanced for this exp erim ent, this group was com pared
to the norm al-videotape group w ho viewed norm al
videotape dem onstrations presented in real time in
Experim ent 1. All participants in the slow-m otion
group followed the sam e experim ental procedures as
the norm al-videotape group in Experim ent 1 for the

num ber of acquisition and retention trials, know ledge

of results and inter-block rest inter vals. H owever, the
groups di ered with respect to the nature of the visual
dem onstrations they observed; the participants in the
norm al-videotape group viewed norm al real-tim e videotaped dem onstrations (a norm al lm rate of 25 H z),
whereas those in the slow -m otion group obser ved slowm otion dem onstrations (15 H z) of the sam e m ovem ent.
Participants in both groups viewed their respective
dem onstrations before practice and interm ittently ever y
10 consecutive practice trials throughout the acquisition
session. All participants were instructed to use the inform ation available in the dem onstrations to help them
to learn the task and to im prove their perform ance.
In this experim ent, the dependent
Data analysis.
variables representing m ovem ent coordination, control
and outcom e scores, as well as the statistical analyses,
were sim ilar to those used in Experim ent 1.
Results
M ovement outcomes. T he AN OVA com puted on m ean
outcom e scores showed a signi cant e ect for groups
(F 1,8 = 9.32, P = 0.02) and trial blocks (F 11,88 = 2.31,
P = 0.01); however, there was no signi cant e ect for
the interaction between groups and trial blocks
(F 11,88 = 1.42, P = 0.18). Inspection of the m eans of
both groups indicated that the norm al-videotape group
(4.69) perform ed signi cantly better than the slowm otion group (3.10) (see Fig. 3). A univariate analysis,
including a H uynh-Feldt procedure to adjust the
degrees of freedom because of a violation of the
sphericity assum ption, m aintained the trial block
e ect (F 11,88 = 2.37, P = 0.015). Follow -up Tukey H SD
tests on the block factor revealed im provem ent in
m ovem ent outcom e with practice. T he perform ance
of both groups on the eighth acquisition block
(m ean = 4.44) and second retention block (m ean =
4.35) was signi cantly better than that for the rst
acquisition block (m ean = 2.83) (P = 0.01 and P = 0.02,
respectively).
For standard deviations of outcom e scores, there
was no signi cant e ect for groups (F 1,8 = 0.507, P =
0.50) or trial blocks (F 11,88 = 1.22, P = 0.29), and there
was no groups trial blocks interaction (F 11,88 = 0.808,
P = 0.63).
The m ultiM ovement coordin ation (relative motion).
variate analyses of variance com puted on the m eans
of velocity velocity cross-correlations (tem poral coordination) and angle angle cross-correlations (spatial
coordination) showed signi cant m ain e ects for groups
(W ilks l 2,7 = 0.227, P = 0.01); however, no signi cant
e ect was found for trial blocks (W ilks l 4,5 = 0.393,

516

Al-Abood et al.

F ig. 3. M ean movement outcom es as a function of the normal videotape group (d ) and slow-motion group (s ) across blocks of
10 practice trials.

F ig. 4. Angle angle relative motion patterns for a representative participant from the norm al videotape group (NVp) and slowm otion group (SM p) on the last acquisition trial, compared to the m edian pattern of the m odel. The abbreviations Int, Rev and
Rel indicate the instants of m ovem ent initiation, arm reversal and dart release respectively.

P = 0.24) and there was no groups trial blocks interaction (W ilks l 4,5 = 0.319, P = 0.16).
Follow-up univariate analyses of variance and discrim inant analysis on the group e ect revealed signi cant di erences between groups for both dependent
variables: spatial coordination (F 1,8 = 22.1, P = 0.00)
and tem poral coordination (F 1,8 = 16.3, P = 0.00). T he
discrim inant functions (i.e. standardized coe cients)
were - 0.70 and - 0.48 for spatial and tem poral coordination, respectively. Inspection of group m eans

for both dependent variables during acquisition and

retention revealed that participants in the norm alvideotape group had higher cross-correlations than
those in the slow -m otion group, indicating closer
approxim ation to the skilled m odel s relative m otion
(see Table 2 and F igs 4 and 5).
For standard deviations of cross-correlations, as an
estim ate of intra-individual variability of m ovem ent
coordination, no signi cant di erences were obser ved
between groups (W ilks l 2,7 = 0.719, P = 0.32) or trial

Visual dem onstrations and skill acquisition

517

Fig. 5. Velocity velocity relative motion patterns for a representative participant from the norm al videotape group (NVp) and
slow-motion group (SM p) on the last acquisition trial, com pared to the m edian pattern of the m odel. The abbreviations Int, Rev
and Rel indicate the instants of movement initiation, arm reversal and dart release respectively.

blocks (W ilks l 4,5 = 0.571, P = 0.51), and there was no

groups trial blocks interaction (W ilks l 4,5 = 0.638,
P = 0.62) (see Table 2).
T he m ultivariate analyses of
M ovement control.
variance on the m eans of elbow release angle, elbow
release velocity and m ovem ent tim e showed no signi cant di erences between groups (W ilks l 3,6 = 0.861,
P = 0.81) or trial blocks (W ilks l 6,3 = 0.423, P = 0.68),
and there was no groups trial blocks interaction
(W ilks l 6,3 = 0.327, P = 0.53).
Sim ilarly, there was no signi cant e ect for groups
(W ilks l 3,6 = 0.487, P = 0.20), or an interaction between
groups and trial blocks (W ilks l 6,3 = 0.211, P = 0.32),
when a M AN OVA was com puted on the standard
deviations of the sam e dependent variables. H owever,
a signi cant e ect was found for trial blocks (W ilks
l 6,3 = 0.021, P = 0.01).
Follow -up univariate analyses of variance and discrim inant analysis on the trial block e ect revealed
signi cant di erences between trial blocks for m ovem ent tim e only (F 2,16 = 7.73, P = 0.00). T he discrim inant function (i.e. standardized coe cient) was - 0.61.
Follow-up Tukey H SD tests revealed a signi cant
decrease in variability of m ovem ent tim e with practice.
The perform ance of both groups on the second acquisition block and retention block was m ore consistent
than that in the rst acquisition block (see Table 3).
Discussion
The aim of Experim ent 2 was to determ ine the e ect
of obser ving slow-m otion demonstrations, com pared

to real-tim e visual demonstrations, on m otor skill

acquisition from a visual perception perspective
(Scully and N ewell, 1985; Scully, 1988). According
to this perspective, slow-m otion displays m ay convey
intact perception of relative m otion inform ation of a
dem onstrated activity but m ay im pede the recognition
of absolute m otion param eters such as speed and
duration of an activity. Following this hypothesis,
and given that perception constrains action, we predicted that there would be no di erences in relative
m otion patterns between the slow-m otion and norm alvideotape groups. However, di erences due to speed
of dem onstration were expected in m ovem ent control
variables such as velocity at release and m ovem ent tim e.
T he results of this experim ent showed that, throughout the acquisition and retention sessions, participants
in the norm al-videotape group approxim ated the
m odel s tem poral and spatial relative m otions m ore
closely than those in the slow -m otion group. It was
apparent that observers of slow-m otion demonstrations
were unable to perceive intact relative m otion inform ation from these dem onstrations, resulting in a less
accurate approxim ation of the m odel s relative m otion
patterns. T his nding does not concur w ith the predictions of the visual perception perspective. Rather,
the results are m ore consistent w ith the suggestion of
Cutting and Pro tt (1982) that dem onstrating an
action in real-tim e is an indispensable procedure for
establishing accurate perception of a coordinate fram e
of reference unique to each activity. T his is also in line
with the work of Barclay et al. (1978), who reported
poor recognition of actions after the observation of
slow -m otion presentations.

518
O ne plausible explanation for the discrepancies
between the ndings of the present study and the
predictions of the visual perception perspective is
the novelty of relative m otion patterns available in the
underarm -throwing task of this study. M ore generally,
it is possible that whether observers can perceive intact
inform ation about relative m otion from slow -m otion
displays depends, to a large extent, on the observers
familiarity w ith the exp erim ental task. It is plausible that
altering the speed of dem onstrations (i.e. a scaling up
or down of the optim al relative m otion) w ill not im pede
m ovem ent perception if observers have previously
viewed the task. Accordingly, it is possible that the participants in the studies of Scully (1987) and W illiam s
(1985, 1989) were able to pick up and use appropriate
relative m otion patterns because the overarm throwing
and bow ling actions demonstrated in those experim ents
were fam iliar to them . In contrast, if a slow -m otion
m anipulation of relative m otion occurs w ith a novel
exp erim ental task, then the perceptual pick-up of the
optim al relative m otion underlying the successful perform ance m ay be perturbed, thus hindering learning
as in this study. Further research is required to verify
this explanation by exam ining the e ects of learners
perceptual experience on action perception and reproduction of novel and fam iliar tasks to be learned
through visual demonstrations presented in slow
m otion.
The analysis of m ovem ent control variables (i.e.
elbow angle and velocity at release plus m ovem ent
tim e) indicated no di erences between participants
in the norm al-videotape and slow-m otion groups in the
reproduction of these variables during acquisition
and retention sessions. T his is also inconsistent w ith
the predictions of the visual perception perspective. We
predicted that slow -m otion presentations m ay hinder
the perception of m ovem ent control variables relevant to
the speed of m ovem ent and thus m ay lead to inaccurate
scaling of m ovem ent param eters. T he present results
are also not in line with those of W illiam s (1985,
1989) and Scully and C arnegie (1998), who showed
incorrect reproduction of tim ing variables (e.g. absolute
tim ing, force) after the obser vation of slow -m otion
dem onstrations.
O ne possible exp lanation for these contradictory
ndings about m ovem ent control is that participants in
the slow-m otion group were able to discover appropriate tim ings of the aim ing action through physical
practice. T hat is, because participants in the present
study had m ore physical practice (i.e. 100 trials) than
those in the studies of W illiam s (1989) and Scully
and C arnegie (1998) 6 and 10 trials, respectively
they were able to nd the appropriate control
param eters. Slow -m otion dem onstrations m ay have
hindered observers perception of m ovem ent tim ing

Al-Abood et al.
characteristics, but had little e ect on m ovem ent production because learning was m ediated by discovery
of key param eters through physical practice. However,
because perception was not assessed in the present
study, it is not possible to con rm this.
N evertheless, subsequent additional analyses of
m ovem ent production of these control variables on
the rst practice trial suggested that this exp lanation
m ay be plausible. We ran separate t-tests on the data
of the rst practice trial for each control variable in
an attem pt to dissociate the e ect of observation of
dem onstrations from that of physical practice. T he
results showed no signi cant di erences between the
slow-m otion and norm al-videotape groups in elbow
angle at release (t 8 = 0.627, P = 0.55) and m ovem ent
time (t 8 = - 0.957, P = 0.37). However, a signi cant
di erence between groups was found for elbow velocity
at release (t 8 = - 2.66, P = 0.03). T he norm al-videotape
group (m ean = 49.5 per second) approxim ated the
m odel s velocity at release (m ean = 48.1 per second)
m uch m ore closely than the slow-m otion group
(m ean = 131 per second).
Furtherm ore, although there were no group differences in m ovem ent tim e, closer inspection of
individual data for each participant in both groups, and
of within-g roup standard deviations, revealed an interesting nding. T he m ovem ent tim es on the rst practice
trial of the participants in the norm al-videotape group
ranged from 1200 to 1790 m s (s = 230), while those
of the participants in the slow-m otion group ranged
from 1040 to 2680 m s (s = 667). T his im plies that
the search by participants for the appropriate m ovem ent tim e of the task was m ore constrained or directed
by the m odel s (m ean = 1820 m s) m ovem ent tim e in
real-tim e dem onstrations com pared to slow -m otion
presentations. Taken collectively and prom pted by
initial ndings, further analyses of m ovem ent control
variables for the rst practice trial suggested that the
observation of slow-m otion dem onstrations could have
hindered the pick-up of absolute m otion inform ation.
H owever, through physical practice, the participants
in the slow-m otion group were able to discover and
approxim ate the appropriate m ovem ent speed and tim e
essential for solving the task problem . Future research
should assess both perception and action to better
understand how m anipulating absolute m otion inform ation in visual dem onstrations m ediates action perception and reproduction.
Finally, analysis of m ovem ent outcom e scores indicated that the norm al-videotape group signi cantly
outperform ed the slow-m otion group during acquisition and retention sessions. T his suggests that the
closer approxim ation of the m odel s relative m otion
by participants in the form er group resulted in better
outcom e scores.

519

Visual dem onstrations and skill acquisition

General discussion
In the exp erim ents reported here, we exam ined the
e ect of m anipulating relative and absolute m otion
inform ation available in visual dem onstrations on the
acquisition of an aim ing task from a visual perception
perspective of observational learning (Scully and
N ewell, 1985; Scully, 1988).
As predicted, E xp erim ent 1 showed that the
inclusion of structural inform ation, found in norm al
dem onstrations, had little additional e ect on the
acquisition of m ovem ent coordination and control
or on m axim izing outcom e scores. T he use of
point-light dem onstrations, com pared to norm al
dem onstrations, was of equal value in facilitating
the acquisition of m ovem ent coordination patterns
by em phasizing the relative m otion inform ation.
The im plication is that learners can pick up relative
m otion inform ation from both norm al videotaped
dem onstrations of a skill and in point-light form ats
to direct learners attention towards transform ational
inform ation. Because of evolutionary in uences on
the hum an visual system , learners can pick up as
m uch transform ational inform ation from norm al
dem onstrations as from point-light displays to e ectively use this source of inform ation in subsequent
m ovem ent reproductions.
In Experim ent 2, som e evidence indicated that the
observation of slow-m otion dem onstrations, com pared
to presentations at regular speed, was detrim ental to
the approxim ation of the m odel s spatial and tem poral
relative m otions as well as to perform ance scores. This
nding contradicted the predictions of the visual perception perspective that slow -m otion displays convey
intact relative m otion inform ation. H owever, further
work is required to determ ine whether the e ect of the
speed of visual dem onstrations on action perception and
reproduction is a function of task constraints, such as
the novelty of relative m otion corresponding to the
dem onstrated activity. The m ore unfam iliar the relative
m otion patterns used in task perform ance, the m ore
likely that the perception of optim al relative m otion
will be hindered by slow -m otion presentation. H owever,
this im plication awaits further research before precise
suggestions for the use of slow -m otion videotapes
in m otor skill acquisition can be advanced. It could
be the case that slow -m otion dem onstrations are m ore
e ective at later stages of learning after the m ovem ent
percept and coordination functions underlying the task
have been established. In other words, slow-m otion
presentations m ight be better used at m ore advanced
stages of learning to direct learners attention to som e
speci c aspects of perform ance, particularly in m otor
skills with m any sub-com ponents varying in tem poral
param eters.

References
Al-Abood, S.A., Davids, K. and Bennett, S.J. (in press).
Speci city of task constraints and e ects of visual demonstrations and verbal instructions in directing learners search
during skill acquisition. Jour na l of M otor B ehavior.
Barclay, C., C utting, J. and Kozlowski, L. (1978). Temporal
and spatial factors in gait perception that in uence
gender recognition. Perception and Psychophysics , 23 ,
145 152.
Bray, J.H. and M axwell, S.E. (1985). M ultivariate Analysis of
Variance . Beverly Hills, CA: Sage.
Cutting, J.E. and Pro tt, D.R. (1982). The minim um
principle and the perception of absolute, com mon, and
relative m otions. C ognitive P sychology , 14 , 211 246.
Johansson, G. (1973). Visual perception of biological m otion
and a model for its analysis. Perception and Psychophysics, 14 ,
201 211.
Johansson, G. (1975). Visual m otion perception. Scienti c
Am erican , 232 , 76 88.
Koslowski, L.T. and Cutting, J.E. (1977). Recognizing the sex
of a walker from a dynam ic point-light display. Perception
and Psychophysics , 21 , 575 580.
McD onald, P.V., Van Em merik, R.E.A. and Newell, K.M .
(1989). The e ects of practice on limb kinematics in a
throwing task. Jour nal of M otor B ehavior , 21 , 245 264.
McGuire, W.J. (1961). Some factors in uencing the e ectiveness of demonstrational lms: Repetition of instructions,
slow m otion, distribution of showings, and explanatory
narration. In Student s Response in Program m ed Instr uctio n
(edited by A.A. Lumsdaine), pp. 187 207. Washington,
DC: National Academy of Science, National Research
Council.
Nelson, D.O. (1958). The e ects of slow motion loop lm on
the learning of golf . R esearch Qua rterly , 29 , 37 45.
Newell, K.M. (1985). C oordination, control and skill. In
Di ering Perspectives in M otor L ear ning , M em or y , and Control
(edited by D. Goodman, R.B. W ilberg and I.M . Franks),
pp. 295 317. Amsterdam: North-Holland.
Pedotti, A. and Ferrigno, G. (1995). Optoelectronic-based
systems. In Three-Dim ensiona l Analysis of H um an M ovem ent (edited by P. Allard, I.A.F. Stokes and J.-P. Blanchi),
pp. 57 77. Champaign, IL: Human Kinetics.
Runeson, S. and Frykholm, G. (1981). Visual perception of
a lifted weight. Jour nal of E xper im ental Psychology: H um an
Perception and Perfor m ance , 8 , 733 740.
Schutz, R.A. and G essaroli, M .E. (1987). The analysis of
repeated m easures designs involving m ultiple dependent
variables. Research Q uarterly for E xercise and Sport , 58 ,
132 149.
Scully, D.M . (1986). Visual perception of technical execution
and aesthetic quality in biological m otion. H um an M ovem ent Scienc e , 5 , 185 206.
Scully, D.M . (1987). Visual perception of biological motion.
Unpublished doctoral dissertation, University of Illinois
at Urbana-Cham paign.
Scully, D.M . (1988). Visual perception of human m ovem ent:
The use of demonstrations in teaching motor skills. B ritish
Jour na l of Physical Education Research , 4 (suppl.), 12 14.

520
Scully, D.M . and Carnegie, E. (1998). Observational learning
in motor skill acquisition: A look at demonstrations.
Irish Jour nal of Psychology , 19 , 472 485.
Scully, D.M . and Newell, K.M. (1985). Observational
learning and the acquisition of motor skills: Towards a
visual perception perspective. Jour na l of H um an M ovem ent
Studies , 11 , 169 186.
Sparrow, W.A., Donovan, E., Van Emmerik, R. and Barry,
E.B. (1987). Using relative m otion plots to measure changes
in intra-limb and inter-limb coordination. Jour na l of M otor
B ehavior , 19 , 115 129.

Al-Abood et al.
W illiam s, A.M ., Davids, K. and W illiam s, J.G. (1999). Visual
Perception and Action in Sport . London: Taylor & Francis.
W illiam s, J.G. (1985). M ovem ent imitation: Some fundamental processes. Unpublished doctoral dissertation,
University of London.
W illiam s, J.G. (1989). Visual dem onstration and m ovem ent
sequencing: E ects of timing variations of a model s actions.
Perceptua l and M otor Skills , 68 , 891 896.
Wood, G.A. (1982). Data smoothing and di erentiation
procedures in biomechanics. Exercise and Sport Science s
Review s , 10 , 308 362.