Motor, visual and egocentric transformations in children with Developmental Coordination Disorder

Original Article

doi:10.1111/j.1365-2214.2006.00688.x

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd

633

Blackwell Publishing LtdOxford, UKCCHChild: Care, Health and Development0305-1862© 2006 The Authors; Journal compilation © 2006 Blackwell Pub-lishing Ltd

? 2006

32

6633647

Original Article

Imagery transformations in DCDJ. Williams

et al

.

Correspondence:Peter Wilson, Division of Psychology, School of Health Sciences, RMIT University, City Campus, PO Box 2476V, Melbourne, Vic. 3001, AustraliaE-mail: [email protected]

Original Article

Motor, visual and egocentric transformations in children with Developmental Coordination Disorder

J. Williams,* P. R. Thomas,† P. Maruff,‡ M. Butson* and P. H. Wilson*

*Division of Psychology, School of Health Sciences, RMIT University, Melbourne, Vic., †School of Curriculum, Teaching, and Learning, Griffith University, Mt. Gravatt Campus, Brisbane, Qld, and ‡The Neurophysiology and Neurovisual Research Unit, Mental Health Research Institute of Victoria, Parkville, Melbourne, Vic., Australia

Accepted for publication 12 June 2006

Abstract

Aim

This study aimed to test the internal modelling deficit (IMD) hypothesis using the mental

rotation paradigm.

Background

According to the IMD hypothesis, children with Developmental Coordination Disorder

(DCD) have an impaired ability to internally represent action. Thirty-six children (18 DCD) completed

four tasks: two versions of a single-hand rotation task (with and without explicit imagery instructions),

a whole-body imagery task and an alphanumeric rotation task.

Results

There was partial support for the hypothesis that children with DCD would display an atypical

pattern of performance on the hand rotation task, requiring implicit use of motor imagery. Overall,

there were no significant differences between the DCD and control groups when the hand task was

completed without explicit instructions, on either response time or accuracy. However, when imagery

instructions were introduced, the controls were significantly more accurate than the DCD group,

indicating that children with DCD were unable to benefit from explicit cuing. As predicted, the controls

were also significantly more accurate than the DCD group on the whole-body task, with the accuracy

of the DCD group barely rising above chance. Finally, and as expected, there was no difference between

the groups on the alphanumeric task, a measure of visual (or object-related) imagery.

Conclusions

The inability of the DCD group to utilize specific motor imagery instructions and to

perform egocentric transformations lends some support to the IMD hypothesis. Future work needs

to address the question of whether the IMD itself is subgroup-specific.

Keywords

Developmental Coordination Disorder, motor co-ordination, motor imagery

Introduction

In recent years, cognitive neuroscience models of

motor control have been used to understand the

basis of movement difficulties experienced by chil-

dren with Developmental Coordination Disorder

(DCD). One interesting hypothesis to emerge is

that these children manifest an underlying deficit

in generating and/or monitoring internal models

of action – termed the internal modelling deficit

(IMD) hypothesis (Wilson

et al

. 2004). This paper

further tests this account using a simple but elegant

cognitive paradigm, mental rotation.

Internal modelling has become an important

concept in motor control and learning as it pro-

vides a framework for understanding variation in

motor proficiency and changes that occur with

maturation (Wolpert 1997; Frith

et al

. 2000).

Internal models are of two types: the

inverse model

(or controller) generates the motor commands

634

J. Williams

et al

.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd,

Child: care, health and development

,

32

, 6, 633–647

necessary to achieve a desired goal state; e.g. neu-

rones of the intraparietal sulcus have a crucial role

in coding target position in body-centred co-

ordinates (Flash & Sejnowski 2001). By compari-

son,

forward models

use a copy of the motor com-

mand (or efference copy) to predict the future state

of the moving limb(s). Forward internal models

provide stability to the motor system by predicting

the outcome of movements before slow, sensorim-

otor feedback becomes available (Wolpert 1997);

feedback delays can be as large as 250 ms (Frith

et

al

. 2000). The forward (predictive) model is then

compared with the desired (or goal) state to

generate online error correction signals that are

transmitted to the motor-planning system (or con-

troller); this allows subtle and rapid adjustments

to the force-timing components of a movement.

The original motor intention appears to emanate

from premotor cortex; however, a functional loop

between premotor and parietal cortex appears to

subserve this forward planning process (Sirigu

et al

. 2004). The forward model is also combined

with sensory feedback once this comes online. This

feedback, especially about the actual outcome of a

movement, is then used to tune the inverse model.

The IMD hypothesis suggests that one of the

underlying causes of DCD may be an inability to

utilize internal models of motor control accurately.

Converging support for the IMD hypothesis has

come from several paradigms including covert ori-

enting of attention (Wilson

et al

. 1997), double-

step saccade tasks (Katschmarsky

et al

. 2001) and

motor imagery (Maruff

et al

. 1999; Wilson

et al

.

2001, 2004). Motor imagery involves the mental

rehearsal or simulation of a motor task in the

absence of overt movement (Sirigu

et al

. 1995;

Decety 1996) and is believed to be an internal

model of a prospective movement, coming to con-

sciousness only because the overt movement has

been inhibited (Crammond 1997). As well, by

anticipating the sensory outcome of a movement

before it is executed, forward internal models can

be engaged in imagery training to select optimal

action sequences (Wolpert 1997; Katschmarsky

et

al

. 2001). In short, motor imagery exhibits many

of the properties of represented action and is con-

sidered a valid means of describing the content of

these internal models for action (Decety & Grézes

1999; Wilson

et al

. 2001; Jeannerod 2003). In

several studies, we have shown that children with

DCD have deficits in motor imagery which we

argue reflect abnormalities in this internal model-

ling process.

Mental rotation is one such paradigm that mea-

sures implicit motor imagery. Typically, mental

rotation involves presentation of object- or limb-

related stimuli at varying degrees from upright.

When required to make a decision about the

presented stimulus (e.g. Is it a left or right hand?),

response times (RTs) typically increase with the

angle of orientation, peaking at 180

°

. A number of

studies have shown that for judgements of hand-

edness using single-hand stimuli, participants

often report imagining moving their own hands

into the orientation of the stimulus, thus utilizing

motor imagery (Parsons 1987, 1994; Parsons & Fox

1998; Parsons

et al

. 1998). This is supported by the

finding that the time taken to make a left-right

decision is closely related to the time taken to actu-

ally move one’s hand into the orientation of the

stimulus (Parsons 1994), and neuroimaging data

showing that mental (limb) rotation activates cor-

tical networks that are required to plan a physical

rotation of the hand (Kosslyn

et al

. 1998; Parsons

& Fox 1998; Thayer

et al

. 2001; Tomasino

et al

.

2005).

The mental rotation paradigm used by Wilson

and colleagues (2004) involved presenting a single-

hand stimulus at angular increments of 45

°

, with

participants required to judge handedness. The RT

pattern of controls matched that commonly seen

in normal adults, with RT increasing linearly with

increases in stimulus orientation away from 0

°

.

However, the DCD group’s RT pattern was atypi-

cal, showing a relatively shallow trade-off between

time and angle – that is, their performance speed

was significantly less constrained by increases in

angle when compared with the controls, whereas

accuracy did not differ between groups. So, at

greater angles, the DCD group were able to com-

plete the task faster, with comparable levels of accu-

racy. Based on these results, it was suggested that

children with DCD might have used an alternate

form of imagery that involved representing the

hands as concrete objects or as body parts viewed

from an external perspective – i.e. a visual (or

Imagery transformations in DCD

635



,

32

, 6, 633–647

object-based) imagery. Further, the shallow RT

function suggested that if the DCD group were

representing the hands as objects, invariant fea-

tures of the hands (such as the angle between the

thumb and index finger) could have been used as

critical features to aid the perceptual decision

about handedness, thereby reducing the processing

load on the required mental transformation and

maintaining accuracy.

There were, however, two significant limitations

of the Wilson and colleagues’ (2004) study. First,

no independent measure of visual imagery was

used to confirm whether an object-based strategy

might be a viable alternative for children with

DCD. Mental rotation of object-related stimuli

(e.g. three-dimensional cube objects or alphanu-

meric characters) is often used to elicit visual imag-

ery, with studies showing that this form of imagery

activates neural areas distinct from those activated

by the rotation of hands (e.g. Kosslyn

et al

. 1998).

Using positron emission tomography, Kosslyn and

his colleagues found that while the rotation of hand

stimuli activated dorsal stream pathways and asso-

ciated frontal motor areas, the mental rotation of

objects activated ventral stream pathways. Thus,

the current study aimed to extend the work of

Wilson and colleagues (2004) by including tests of

motor and visual imagery within a mental rotation

paradigm.

The second limitation of Wilson and colleagues

(2004) was that hand stimuli were presented in

either palm or back view, but were combined for

analysis. Although some studies show that a typical

RT pattern is observed when views are combined

(de Lange

et al

. 2005), others show distinct patterns

for palm and back view (Parsons 1987, 1994). In

the current study, we presented single hands in

back view only as responses to these stimuli typi-

cally show a linear RT trade-off, not only when

handedness judgements are required but also when

subjects are merely asked to imagine their own

hands in the position of the stimulus without indi-

cating handedness (Parsons 1994). Hence, we

could be sure that a ‘typical’ response pattern

would reflect a linear trade-off for angle in RT.

We also reasoned that if children with DCD have

a reduced ability to perform movement-related

mental transformations, then they would be resis-

tant to any brief induction that is designed to

encourage explicit use of movement imagery. Thus,

the hand mental rotation task was completed twice,

first without and then with explicit motor imagery

instructions. The instructions were designed to

encourage children to adopt a first-person perspec-

tive when performing the task. Previous studies

have shown that instructions of this type can sig-

nificantly alter the response patterns of unimpaired

groups on mental rotation (Roelofs

et al

. 2002).

Finally, it has been noted more recently that yet

another form of mental transformation, one

involving a change in egocentric viewing perspec-

tive, occurs when

whole-body stimuli

are used in

a mental rotation paradigm (Zacks

et al

. 2002b,

2003). An egocentric transformation refers to

‘imagined movements of the observer’s point of

view relative to the environment’ (Zacks

et al

. 2003,

p. 1660). This is a more complex transformation

than that required in the hand task, and allows

participants to view the environment from differ-

ent perspectives, including those of other people

(Zacks

et al

. 2002a). Hence, the ability to utilize this

type of transformation is important for interaction

with the environment. As well, being able to take

on the (physical) perspective of other people may

aid in skill development, as it allows an observer

to see the task from the performer’s perspective,

supporting imitation and mental rehearsal of a

modelled action. Such a task has not been used

previously with children with DCD. As these chil-

dren have difficulty interacting with and negotiat-

ing their environment (Larkin & Hoare 1991;

Wilson

et al

. 2001) and find imitative actions prob-

lematic (e.g. Dewey 1993), we would expect that

the ability to perform egocentric transformations

may be impaired in DCD. The inclusion of this

whole-body task allowed us to explore this possi-

bility further.

Importantly, there are some interesting differ-

ences between the whole-body imagery task and

other mental rotation tasks. The task is presented

in a similar manner to other mental rotation tasks

eliciting motor or visual imagery, but does not

elicit a typical RT trade-off for angle, as partici-

pants usually imagine themselves in the position of

the whole body, rather than imagining a rotation

of the stimulus through the frontal plane. Hence,

636

J. Williams

et al

.



,

32

, 6, 633–647

there is little effect for angle. This unique RT pat-

tern will be discussed further in the

Discussion

.

To summarize, the study presented here had a

number of aims. First, we wanted to further exam-

ine the performance of children with DCD on a

motor imagery task and determine the impact of

imagery instructions on their ability to perform

the task. We also wanted to explore whether these

children would be impaired on a whole-body task,

which required egocentric transformations. Finally,

we wanted to rule out the existence of a general

imagery deficit in children with DCD using a visual

imagery task. Based on our previous findings, we

predicted first that the DCD group would display

an atypical performance pattern in the hand rota-

tion task. Second, we expected that although con-

trols may show some benefit from explicit imagery

instructions, the DCD group’s performance would

show minimal change. This prediction was based

on two main observations: first, the view that the

ability to utilize verbal prompts about movement

representation presupposes that the performer has

learned the implicit correlation between real and

imagined action and hence can model accurately

the internal co-ordinates of an action; and second,

on findings in the sport psychology domain that

show athletes who have greater general imagery

ability are better able to use explicit mental imagery

training (Martin

et al

. 1999). Third, we expected

that DCD children would be less efficient than con-

trols in their performance of the whole-body task.

Finally, given that the IMD should have little

impact on the ability of DCD children to perform

visual imagery, we expected no group differences

on the alphanumeric task.

Methodology

Participants

The initial sample consisted of 39 children aged

between 7 and 11 years. Children were selected via

a method which has been successfully used in our

previous studies (Maruff

et al

. 1999; Katschmarsky

et al

. 2001; Wilson

et al

. 2001, 2004). First, teachers

of children in grades 2–6 from a standard primary

school were asked to refer children whose motor

co-ordination they believed to be below an age-

appropriate level. These children were included in

the DCD group if they scored below the 15th per-

centile on the Movement Assessment Battery for

Children (Movement ABC; Henderson & Sugden

1992) and there was no known neurological or

physical pathology present. Similar steps are com-

monly used when identifying children with DCD

for research purposes (Geuze

et al

. 2001). Thirteen

children were identified as having DCD using

the above method. A further three children were

included who previously were identified as having

DCD and were part of a wait-list DCD intervention

group tested in another study. Teachers from the

same primary school above also referred children

whom they believed to have age-appropriate motor

co-ordination to form the control group. These

children were included in the control group if they

scored above the 20th percentile on the Movement

ABC. No children referred by their teachers as hav-

ing motor co-ordination below an age-appropriate

level scored above the 15th percentile on the Move-

ment ABC in the current study; however, three

other non-referred children were excluded given

that they scored between the 15th and 20th percen-

tiles. As a result, the DCD group was made up of

18 participants (nine female, nine male) with a

mean age of 9.7 years (SD

=

0.7 year), a mean

Movement ABC Total Movement Impairment

Score (TMIS) of 14.6 (SD

=

3.0) and a mean Move-

ment ABC percentile rank of 5.2 (SD

=

3.5). The

control group consisted of 18 participants (nine

female, nine male) with a mean age of 9.2 years

(SD

=

1.4 years), a mean TMIS of 5.6 (SD

=

1.9)

and a mean percentile rank of 41.6 (SD

=

16.9).

Results of independent

t

-tests showed no signifi-

cant age difference between the two groups,

t

(24.85)

=

1.41,

P

=

0.172, but there was a signifi-

cant difference between the groups for Movement

ABC percentile rank,

t

(18.41)

=

−

8.93,

P

<

0.001.

Test apparatus

Rotation tasks

The E-Prime software package (Psychology Soft-

ware Tools, Pittsburgh, PA, USA) was used to

present all stimuli for the rotation tasks, recording

participant responses to the nearest 1 ms. For each

Imagery transformations in DCD

637



,

32

, 6, 633–647

task, the stimuli were centred on the computer

screen, and were presented in 45

°

increments

between 0

°

and 360

°

. All stimuli were presented

after a random delay of between 2 and 3 s, with the

stimuli remaining on the screen until the partici-

pant responded, or 10 s had expired, whichever

came first. Stimuli images are described below.

Hand rotation

The stimuli were high-resolution images of a left

or right hand, measuring 9 cm by 8 cm. All the

hands were in the back view (see Fig. 1a). Partici-

pants completed five practice trials, followed by 40

test trials. Data were combined for stimuli having

the same angle of rotation, irrespective of orienta-

tion. For example, data for the 45

°

and 315

°

trials

were combined, as both stimuli were 45

°

from

upright. This is a common technique used in men-

tal rotation paradigms (see, for example, Harris

et

al

. 2000; Roelofs

et al

. 2002), which increases the

reliability of RT estimates at each angle by increas-

ing the number of trials. Thus, there were eight

trials at each of five angles (0

°

, 45

°

, 90

°

, 135

°

and

180

°

), of which four were left hands and four were

right hands. All trials were randomly presented for

each participant.

Alphanumeric rotation

The stimuli were single figures, either the capital

letter F or the number 5, each measuring 9 cm by

6 cm. The stimuli either could face the normal,

correct direction, or were mirror-reversed (see

Fig. 1b). As with the hands, participants completed

five practice trials, followed by 40 test trials, half

containing the letter F and the other half the num-

ber 5. After combining data for stimuli with the

same angle of rotation, there were eight trials per

angle. All trials were again randomly presented.

Whole-body task

The stimuli for this task were high-resolution

images of a man with either his left or right arm

held out to his side, or crossed over his body, the

figure measuring 9 cm by 6 cm (see Fig. 1c). As

with the other tasks, participants were given five

practice trials, followed by 40 test trials. There was

an equal and random distribution of left and right

arms held out either across or to the side of the

body. Like the alphanumeric task, this gave eight

trials per angle, after data from the same angles

were combined.

Procedure

For all rotation tasks, participants were given 10 s

to record a response. If no response was recorded

after 10 s, the stimulus disappeared and the next

trial began. All trials were followed by a random

delay of between 2 and 3 s, regardless of whether

or not a response was registered.

Hand rotation

Two hand rotation tasks were used in this study: a

hand rotation task with no instruction (

HR-NI

)

and a hand rotation task with instruction (

HR-WI

).

For the

HR-NI

condition, participants were asked

to determine whether the stimulus was a left or a

right hand as quickly and as accurately as possible,

and to respond by pressing one of two designated

keys on the keyboard. They were not given any

Figure 1.

Task stimuli. (a) Hand stimulus (left hand, 0

°

of rotation); (b) Alphanumeric stimuli (number 5, mirror-reversed at 0

°

); and (c) Whole-body stimuli (left arm across, 0

°

).

(a) (b) (c)

638

J. Williams

et al

.



,

32

, 6, 633–647

guidance on technique or strategy. In the

HR-WI

condition, participants were asked to imagine their

own hand in the position of the stimulus and to use

this as a guide when deciding whether it was a left

or right hand. Again, they were asked to respond as

quickly and accurately as possible. For both tasks,

participants’ hands were covered with a towel to

prevent them looking at their own hands or hold-

ing their hands up to the screen for comparison.

Before their hands were covered, participants lo-

cated the appropriate response keys on the com-

puter keyboard –

d

for left and

k

for right. These

keys were marked with blu-tack to ensure partici-

pants could identify them underneath the towel.

Alphanumeric rotation

Participants were asked to determine whether the

F or 5 was the ‘right’ or ‘wrong’ way around as

quickly and accurately as possible. They again

responded by pressing one of two designated keys

on the computer keyboard:

d

if the figure was the

‘right’ way and

k

if the ‘wrong’ way.

Whole-body task

In this task, participants were asked to determine

if the man on the screen was holding out his left or

right arm. They were instructed to imagine them-

selves in the position of the man to help them

decide which arm was being held out. Again, par-

ticipants responded by pressing designated keys on

the computer keyboard:

d

if it was a left arm and

k

if it was a right arm.

Task order

Testing generally took between 45 and 60 min. Task

order was pseudo-randomized across participants,

with all participants first completing two rotation

tasks, followed by the Movement ABC, and then

the remaining two tasks. The

HR-NI

was always

completed before the

HR-WI

and the whole-body

task, to avoid leading participants in their strategy.

Finally, to avoid an immediate practice effect, the

HR-WI

was always completed in the second block

of rotation trials, resulting in a break between the

two hand tasks of approximately 30 min.

Design and analysis

For each child in each task, mean RTs and accuracy

(proportion correct) were calculated for each angle

over both directions of rotation. To remove antic-

ipatory and abnormally delayed responses, RTs less

than 250 ms (anticipatory) and responses greater

than 2.5 times the mean RT for each angle for each

individual participant were excluded from analysis

(cf. Kosslyn

et al

. 1998). Each child’s mean RT was

plotted against angle of rotation. A linear curve was

fitted to the RT plot and a least squares method was

used to obtain slope, intercept and

r

2

. Mean regres-

sion estimates were, thus, derived for each group.

Response time and accuracy data were tested for

violations of the assumptions of normality and

homogeneity of variance. For the RT data, the nor-

mality and variance assumptions were not tenable

for several variables, but were corrected using

square root and power transformations, where

appropriate.

For response accuracy, negative skew could not

be corrected using data transformations. However,

the

F

statistic in

ANOVA

is highly robust to the effect

of skewness (Lindman 1974) and to departures of

normality when the

df

error

are greater than 20

(Tabachnick & Fidell 1996).

For RT data, multivariate planned contrasts

comparing groups (DCD-Control) were con-

ducted for each task using the mean regression esti-

mates as dependent measures (slope, intercept and

r

2

). As well, a two-way repeated measures

ANOVA

[5(angle)

×

2(group)] was conducted for each task.

Response accuracy data were also submitted to a

5

×

2 repeated measures

ANOVA. The analysis of

repeated measures data was conducted using the

multivariate approach to ANOVA to protect against

violations to the assumption of sphericity.

Results

Hand rotation task

RT

For both hand rotation tasks, mean regression esti-

mates were calculated within each group for RT on

angle and can be seen in Table 1. For the HR-NI

Imagery transformations in DCD 639

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Child: care, health and development, 32, 6, 633–647

task, a planned contrast on the regression estimates

did not reveal a significant multivariate effect for

group (P = 0.095). Univariate tests revealed a sig-

nificant difference between the two groups for

intercept, F1,32 = 5.29, P = 0.028, η2 = 0.142, while

slope approached significance, F1,32 = 3.85,

P = 0.059, η2 = 0.107. There was no significant

difference between the groups for r2 (P = 0.11). For

the HR-WI task, there was no significant multivari-

ate effect for group (P = 0.33) and no significant

univariate effects for slope (P = 0.076), intercept

(P = 0.23) or r2 (P = 0.44).

For the HR-NI task, repeated measures ANOVA

on mean RT revealed no angle by group interaction

(P = 0.40; Fig. 2a). There was a significant effect for

angle, Wilks’ Λ = 0.239, F4,29 = 23.10, P < 0.001,

η2 = 0.761, but no effect for group (P = 0.63). The

effect for angle was explored using pairwise com-

parisons with Bonferroni adjusted α levels, which

revealed significant differences on mean RT

between all angles (all P < 0.001) with the excep-

tion of 0° and 45°, P = 1.00; and 135° and 180°,

P = 0.134.

For the HR-WI task, repeated measures ANOVA

on mean RT revealed no angle by group interaction

(P = 0.32; Fig. 2b). There was a significant effect for

angle, Wilks’ Λ = 0.156, F4,29 = 39.22, P < 0.001,

η2 = 0.844, but no effect for group (P = 0.71). The

effect for angle was explored using pairwise com-

parisons, revealing significant differences between

all angles (all P < 0.05) other than 0° and 45°(P = 1.00).

Accuracy

For the HR-NI task, repeated measures ANOVA on

mean accuracy revealed no angle by group interac-

tion (P = 0.75; Fig. 3a). There was no significant

Table 1. Mean regression estimates for the DCD and control groups for each task

Task Group n Slope Intercept r2

Hands NI DCD 17 7.05 (6.12) 1829.05 (722.76) 0.60 (0.31)Control 17 11.18 (6.04) 1359.64 (431.59) 0.79 (0.11)

Hands WI DCD 17 7.54 (4.58) 1472.01 (590.84) 0.59 (0.35)Control 17 10.98 (6.26) 1238.32 (512.68) 0.73 (0.22)

Whole body DCD 18 4.56 (4.97) 2188.55 (1146.42) 0.39 (0.36)Control 18 3.25 (6.71) 3008.04 (1225.37) 0.41 (0.27)

Alphanumeric DCD 18 5.21 (5.21) 1601.74 (886.34) 0.57 (0.27)Control 18 7.52 (7.17) 1501.47 (544.77) 0.60 (0.33)

SD in parentheses.DCD, Developmental Coordination Disorder; NI, no instruction; WI, with instruction.

Figure 2. Mean response time (ms) by angle (°) for (a) the hand task with no instruction and (b) the hand task with instruction. DCD, Developmental Coordination Disorder.

DCD

Control

Angle (degrees)

00

500

1000

1500

2000

2500

3000

3500

4000(b)

45

Res

pons

e T

ime

(ms)

90 135 180

DCD

Control

00

500

1000

1500

2000

2500

3000

3500

4000(a)

45

Res

pons

e T

ime

(ms)

90 135 180

640 J. Williams et al.


effect for angle (P = 0.092) or group (P = 0.50). For

the HR-WI task, repeated measures ANOVA again

revealed no angle by group interaction for accuracy

(P = 0.30; Fig. 3b). There was, however, a signifi-

cant effect for angle, Λ = 0.618, F4,29 = 4.49,

P = 0.006, η2 = 0.382, and a significant group

effect, F1,32 = 4.94, P = 0.034, η2 = 0.134. The effect

for angle was explored using pairwise comparisons

and revealed that all other angles differed signifi-

cantly from 180°: 0° vs. 180°, P = 0.003; 45° vs.

180°, P = 0.003; 90° vs. 180°, P = 0.002; and 135°vs. 180°, P = 0.020.

Whole-body task

RT

Mean regression estimates were calculated within

each group for RT on angle and can be seen in

Table 1. A planned contrast on the regression esti-

mates did not reveal a significant multivariate

effect for group (P = 0.20). Univariate tests

revealed a significant difference between the two

groups for intercept, F1,34 = 4.74, P = 0.037,

η2 = 0.122, but there was no significant difference

between the groups for slope (P = 0.51) or r2

(P = 0.76).

Repeated measures ANOVA on mean RT revealed

no angle by group interaction (P = 0.34; Fig. 4a).

There was a significant effect for angle, Wilks’

Λ = 0.551, F4,31 = 6.32, P = 0.001, η2 = 0.449, and

the effect for group approached significance,

F1,34 = 3.56, P = 0.068, η2 = 0.095. The effect for

angle was explored using pairwise comparisons,

revealing differences at three combinations of

angle: 0° and 135°, P = 0.007; 0° and 180°,

P = 0.003; and 90° and 180°, P = 0.034.

Accuracy

Repeated measures ANOVA on mean accuracy

revealed no angle by group interaction (P = 0.65;

Fig. 4b). There was no significant effect for angle

(P = 0.36), but there was a strong group effect,

F1,34 = 9.50, P = 0.004, η2 = 0.218.

Alphanumeric task

RT

Mean regression estimates were calculated within

each group for RT on angle and can be seen in

Table 1. A planned contrast on the regression esti-

mates did not reveal a significant multivariate

effect for group (P = 0.73). Univariate tests

revealed no significant differences between the two

groups for slope (P = 0.28), intercept (P = 0.69) or

r2 (P = 0.43).

Repeated measures ANOVA on mean RT revealed

no angle by group interaction (P = 0.63; Fig. 5a).

There was a significant effect for angle, Wilks’

Λ = 0.376, F4,31 = 12.86, P < 0.001, η2 = 0.624, but

Figure 3. Mean response accuracy (% correct) by angle (°) for (a) the hand task with no instruction and (b) the hand task with instruction. DCD, Developmental Coordination Disorder.

DCD

Control

DCD

Control

Angle (degrees)

00

10

20

30

40

70

60

50

80

90

100

(b)

45

Acc

urac

y (%

cor

rect

)

90 135 18000

10

20

30

40

70

60

50

80

90

100

(a)

45

Acc

urac

y (%

cor

rect

)

90 135 180



no significant effect for group (P = 0.98). The effect

for angle was explored using pairwise comparisons,

revealing significant differences between the fol-

lowing combinations of angle: 0° and 135°,

P = 0.001; 0° and 180°, P < 0.001; 45° and 90°,

P = 0.005; 45° and 135°, P < 0.001; 45° and 180°,

P < 0.001; 90° and 135°, P = 0.003; and 90° and

180°, P < 0.001.

Accuracy

Repeated measures ANOVA on mean accuracy

revealed no angle by group interaction (P = 0.55;

Fig. 5b). There was a significant effect for angle,

Wilks’ Λ = 0.478, F4,31 = 8.46, P < 0.001, η2 = 0.522,

but there was no effect for group (P = 0.37). The

effect for angle was resolved using pairwise

comparisons. There were significant differences

between the angles 0° and 135°, P = 0.009; 0° and

180°, P < 0.001; and 45° and 180°, P = 0.007.

Discussion

The aim of the current study was to further exam-

ine the IMD hypothesis in children with DCD

using a well-validated cognitive neuroscientific

method for eliciting both motor and visual imag-

ery. In broad terms, we expected that this deficit

would be evident in those tasks requiring the use

of motor imagery, and that children with DCD

Figure 4. Whole-body task – (a) mean response time (ms) and (b) mean response accuracy (% correct), by angle (°). DCD, Developmental Coordination Disorder.

DCD

Control

DCD

Control

Angle (degrees)

00

500

1000

1500

2000

2500

3000

3500

4000

(a)

45

Res

pons

e T

ime

(ms)

90 135 180 00

10

20

30

40

70

60

50

80

90

100

(b)

45

Acc

urac

y (%

cor

rect

)

90 135 180

Figure 5. Alphanumeric task – (a) mean response time (ms) and (b) mean response accuracy (% correct), by angle (°). DCD, Developmental Coordination Disorder.

DCD

ControlDCD

Control

00

10

20

30

40

50

60

70

80

90

100

(b)

45

Angle (degrees)

Acc

urac

y (%

cor

rect

)

90 135 18000

500

1000

1500

2000

2500

3000

3500

(a)

45

Res

pons

e T

ime

(ms)

90 135 180



would show little benefit from explicit instructions

designed to encourage use of motor imagery.

Hand task

Analysis of RT for the HR-NI task revealed a sig-

nificant difference between the control and DCD

groups for intercept, with slope also approaching

significance. There was no difference between the

groups for accuracy. In our previous work using a

similar hand rotation task (Wilson et al. 2004), we

found the DCD group showed a significantly lower

slope than controls, but there was no difference in

RT between the groups at 0° or for accuracy.

Although the findings of the two studies were sim-

ilar in some respects, there were also some differ-

ences, the most obvious being the differences in RT

patterns. In our previous study, the DCD group

tended to respond faster at larger angles, whereas

in the current study, there was no significant effect

involving group and though the DCD group did

appear to respond faster at 180° (see Fig. 2a), this

was not significant. Also, the values for slope were

greater in the current study than those in the pre-

vious study for both groups; in the Wilson and

colleagues’ (2004) study, slope for the DCD group

was 0.9 and 3.6 for controls, whereas in the current

study, slope was 7.0 for the DCD group and 11.2

for controls. These differences in the RT function

may be the result of changes to task constraints in

the current study. Participants in the current study

had longer to respond than they had in previous

studies (up to 10 s), permitting greater response

variability. Also, participants’ hands were covered

in the current study to prevent them looking at

their own hands to complete the comparison; by

removing this possible strategy, we may have

altered the task approach of some participants,

prompting them to generate a mental transforma-

tion without any possible (visual) reference to

actual hand orientation. Finally, the Wilson and

colleagues’ (2004) study presented hands in both

back and palm views, but only the back view was

presented to participants in the current study.

The presence of linear RT functions suggests that

some form of mental rotation was occurring for

both controls and DCD participants in the current

study. Given that the difference between the groups

for slope was not significant, and there was no

difference between the groups for accuracy, these

results on their own are not supportive of the IMD

hypothesis. Taken together, however, the results of

the HR-WI task (discussed below) provide some

support for the view that children with DCD may

have an underlying difficulty completing the imag-

ined transformations from an internal (or egocen-

tric) perspective, ala true motor imagery.

The purpose of the explicit imagery instructions

was to encourage the children to adopt a first-

person perspective when completing the hand task,

thereby reducing the likelihood of other (object-

based) strategies being used. When we look at the

results of the HR-WI, it is evident that the instruc-

tions had little impact on the RTs of either group.

For both groups, the regression estimates were sim-

ilar when compared with the HR-NI task, with the

exception of intercept for the DCD group. Further,

ANOVA revealed no effect for group and no inter-

action in the HR-WI. However, the lack of apparent

change in RT patterns for either group does not

necessarily mean that the imagery instructions had

no impact on performance. Again, research has

shown that for stimuli depicting the back of hands,

RT patterns closely resemble the traditional mental

rotation pattern (Parsons 1987, 1994); indeed, it is

difficult to differentiate performance strategy

based on RT alone because a visual (or object-

based) strategy is likely to result in a similar RT

profile to that seen when using motor imagery.

Instead, we must also look at performance

accuracy.

There was a striking difference in accuracy

between the two groups when specific imagery

instructions were introduced. In the HR-WI task,

the control group was now significantly more accu-

rate than the DCD group, demonstrating a marked

improvement in their accuracy without sacrificing

RT. Hence, although the DCD group showed a typ-

ical RT function on this task (similar to that in the

HR-NI task), they were unable to utilize the imag-

ery instructions to improve their task accuracy,

unlike controls. The lack of improvement in the

DCD group indicates that they continued to use

the same strategy as that used in the HR-NI task.

Previous studies using the Visually Guided

Pointing Task (VGPT) in children with DCD sug-



gest that they have a reduced ability to accurately

utilize motor imagery (Maruff et al. 1999; Wilson

et al. 2001). In the HR-NI task, however, the DCD

group were equally as accurate as the controls.

Given this, we argue that if a motor imagery deficit

does exist in children with DCD, the DCD group

in the current study must have used an alternative

strategy to complete the task as accurately as con-

trols. A form of visual imagery may have provided

such an alternative. Then, when imagery instruc-

tions were introduced, the DCD group were unable

to make use of the instructions, continuing on

instead with their alternative strategy. The pattern

of results, combined with our earlier work, suggests

that controls were able to enlist motor imagery on

both tasks and, moreover, were able to utilize

instructions to effectively improve their accuracy.

One alternative explanation for the performance

of controls is that they might have used a similar

approach to the DCD group in the HR-NI task, and

then enhanced their performance by switching to

motor imagery following the provision of instruc-

tions. Conversely, both groups might have

attempted to use motor imagery to complete the

HR-NI task, achieving moderate levels of success

in doing so. With the introduction of imagery

instructions, the controls were able to utilize those

instructions to refine their performance, whereas

their age-matched peers in the DCD group were

not able to improve.

Another possibility concerns practice effects. It

could be argued that there was a practice effect

when participants completed the HR-WI task, as

they had previously completed the same task with-

out specific instructions. The improved perfor-

mance of the controls could thus be attributed to

practice, with this group able to benefit more from

the previous task than the DCD group. This seems

unlikely, however, as steps were taken to overcome

practice effects. The tasks were completed in sepa-

rate blocks, with the Movement ABC completed in

between, providing about 30 min between tasks.

Also, the number of trials used in this study (40

trials per task) is far below the number of trials

used in other hand rotation tasks, including Wilson

and colleagues (2004), who presented 80 trials.

Short-term practice effects are not considered an

issue in hand rotation studies that employ large

numbers of trials (such as Parsons 1994); results

are rarely analysed in blocks. Further, one study on

the impact of practice in the mental rotation of

alphanumeric characters found that after large

amounts of practice, the performance of children

became significantly faster, as well as more accurate

(Kail & Park 1990). Although no such study has

been conducted using hand stimuli, the results

with alphanumeric characters suggest that if chil-

dren in the current study were benefiting from

practice, we would have seen a reduction in RT, as

well as increases in accuracy. It is clear in Fig. 2 that

there was no such effect for RT.

It would appear that the control group were able

to successfully utilize motor imagery instructions

to improve their performance, but the DCD chil-

dren were unable to do the same. That is, although

the DCD group appeared to use mental rotation,

and may indeed have attempted to use motor

imagery, they were unable to perform as accurately

as an age-matched control group. We argue that

this deficit is due to a reduction in the DCD group’s

ability to perform imagined transformations from

an internal perspective, consistent with the IMD

hypothesis. We believe that use of functional neu-

roimaging techniques like fMRI will provide the

most crucial challenge for the IMD hypothesis.

Such techniques will enable us to isolate the neural

networks supporting mental rotation in DCD, and

determine with clarity whether systems subserving

object-based transformations are favoured over

those serving internal modelling, even when limb-

related stimuli are presented.

Whole-body task

For the whole-body task, the DCD group was con-

sistently faster in responding across all angles, a

trend which was close to reaching significance. The

RT trade-off was similar for both groups, but there

was a strong group difference in accuracy. The con-

trol group was significantly more accurate than the

DCD group at four of the five angles (with 90° the

only exception). So, the DCD group was somewhat

faster but less accurate than controls, suggesting a

speed-accuracy trade-off. This may have resulted

from an increase in task complexity. The accuracy

for both groups in this task was far below that in



either the hand or alphanumeric rotation tasks:

around 70% for controls and at chance levels for

the DCD group. Children in the DCD group may

emphasized speed of responding in order to dem-

onstrate competence because they found the

required imagined transformations too difficult.

Both groups did show a linear RT function for

the whole-body task (Fig. 4). Adult data suggest

that if participants are imagining themselves in the

position of the figure on the screen there should be

little, if any, trade-off, as rotation as such, does not

occur (Zacks et al. 2002a,b, 2003). Studies by Zacks

and colleagues have shown a negligible correlation

(using Pearson’s r) between angle and RT of −0.18

(Zacks et al. 2002b) and 0.17 (Zacks et al. 2003).

This compares with r2 values of 0.39 for the DCD

group and 0.41 for controls in the current study

(equating to Pearson’s r values of 0.62 and 0.64

respectively). This indicates that participants in the

current study may have been rotating the figure in

the frontal plane, before making an imagined

transformation of their own body. Parsons (1987)

discusses the use of different strategies in his study

of whole-body transformations. His adult partici-

pants indicated that they performed transforma-

tions of their own body to compare it with an

external stimulus (that on the screen). It was

argued that this strategy would be more efficient

than imagining the rotation of the stimulus as it

requires the internal representation of only one

entity, one’s own body. Rotating the stimulus to the

upright position and comparing that with the rep-

resentation of one’s own body is more complex; it

requires the imagined stimulus to be maintained in

an upright position in the mind while an egocen-

tric transformation of one’s own body occurs for

comparison. As this strategy involves rotation of

the stimulus in the frontal plane, we would expect

some form of RT trade-off to occur and given this

strategy is believed to be more complex, we would

also expect some decrease in accuracy.

There is some anecdotal evidence for this strat-

egy: when asked how they solved this task, some

children described the process of rotating the

stimulus to the upright and then imagining them-

selves in that position. In the study presented

here, the control group were approximately 500–

1000 ms slower to respond on the whole-body

task across all angles, a trend that neared signifi-

cance, and were more accurate than DCD partici-

pants. Perhaps this additional time corresponds to

either more care on the part of the control group,

or it is representative of the time taken to perform

this imagined transformation of their own bodies

around the vertical axis. That is, both groups were

rotating the figure to the upright, but the control

group then made an additional imagined trans-

formation of themselves; the DCD group may

have found this final transformation difficult, and

instead guessed which hand was outstretched.

This is a plausible explanation given that accuracy

of the DCD group was largely at chance levels,

and that accuracy for both groups was affected

very little by changes in angle. Why children

appear to prefer this strategy to that of a direct

egocentric transformation of their own body is

unclear; the task has not been used previously

with children, but perhaps its complexity may

have had some impact upon the choice of strategy

for both groups. The apparent inability of the

DCD group to perform the final egocentric trans-

formation is reflective of the difficulty they have

interacting with their environment (Larkin &

Hoare 1991).

Alphanumeric task

The RT pattern demonstrated by both the DCD

and control groups showed a typical mental rota-

tion pattern, similar to that found in previous

alphanumeric rotation tasks (Kail et al. 1980; Kail

1985; Harris et al. 2000; Harris & Miniussi 2003).

There were no differences between the groups for

RT or accuracy, indicating that in this study, the

DCD group had no difficulty performing mental

rotation when the task did not involve a motor

component. This is an important finding in terms

of the IMD hypothesis, as a general deficit on men-

tal rotation tasks needed to be excluded before one

could argue that children with DCD were impaired

in their ability to perform hand rotation tasks.

General discussion

The current findings provide partial support for

the IMD hypothesis in children with DCD. Most



persuasive are data from the hand rotation task,

where specific imagery instructions had a signifi-

cant impact upon accuracy for the control group,

but not the DCD group. Further, this benefit for

the control groups occurred without sacrificing RT,

with no difference for this measure between the

DCD and control groups in either hand task. Chil-

dren with DCD were also impaired in their ability

to perform the whole-body task accurately when

compared with the controls, another task which

gave specific imagery instructions and requires

egocentric body transformations. However, no dif-

ferences were found in the alphanumeric task, a

task involving visual imagery without any motor

component. These results suggest that an underly-

ing inability to accurately generate and/or monitor

forward models of motor control impacts nega-

tively upon a child’s motor development, reinforc-

ing our earlier findings using the VGPT (Maruff et

al. 1999; Wilson et al. 2001).

Despite support for the IMD hypothesis over a

series of studies, the heterogeneous nature of DCD

suggests that this is not the sole underlying cause of

the disorder. Indeed, a number of our studies have

revealed the emergence of subgroups within DCD

groups, with some participants more affected than

others, and some not affected at all. For example, in

Wilson and colleagues’ (2001) study involving the

VGPT, there were participants in the DCD group

who did conform strongly to Fitts’s law, with sim-

ilar values to those of controls. Also, in another

mental rotation study yet to be published, partici-

pants were placed into subgroups based on their

explanations of how they completed the rotation

tasks. Using this method, we identified a subgroup

of children with DCD who did not refer to mental

rotation when explaining how they completed the

task; this group’s performance patterns differed

from the DCD subgroup who did refer to mental

rotation, and to all controls, indicating they were

either impaired in their ability to use motor imag-

ery or using another, less successful strategy. Fur-

ther, though not presented here because of small

sample size, we did separate the DCD group into

mild and severe DCD based upon their Movement

ABC scores – severe DCD scored below the 5th per-

centile on the Movement ABC, whereas mild DCD

scored between the 6th and 15th percentiles. From

the descriptive data, those children with severe

DCD appeared to have a greater impairment in

their motor imagery performance than those with

mild DCD. This finding is currently being followed

up with a larger sample and the preliminary results

appear to support those described above.

An intriguing avenue for further research would

be to determine whether children with particular

forms of motor impairment are more affected than

others by an IMD. For example, we need to profile

participants’ motor skills and examine whether

fine or gross motor skills are more affected by an

IMD, and whether ability profiles map onto action-

specific deficits in the use of motor imagery.

We suggest the need for larger studies incorpo-

rating a number of additional standardized motor

tasks in addition to the Movement ABC, with a

larger sample of control, DCD severe and DCD

mild participants. These additional tasks should be

selected carefully to avoid floor and ceiling effects

when testing children of different ages. This would

aid in the development of motor impairment

profiles for children displaying IMDs. The identi-

fication of ability profiles associated with particular

forms of motor imagery deficit would enable prac-

titioners to tailor different forms of intervention to

the individual child.

References

Crammond, D. J. (1997) Motor imagery: never in your

wildest dream. Trends in Neuroscience, 20, 54–57.

Decety, J. (1996) The neurophysiological basis of motor

imagery. Behavioural Brain Research, 77, 45–52.

Decety, J. & Grézes, J. (1999) Neural mechanisms sub-

serving the perception of human actions. Trends in

Cognitive Sciences, 3, 172–178.

Dewey, D. (1993) Error analysis of limb and orofacial

praxis in children with developmental motor deficits.

Brain and Cognition, 23, 203–221.

Flash, T. & Sejnowski, T. J. (2001) Computationl

approaches to motor control. Current Opinion in

Neurobiology, 11, 655–662.

Frith, C. D., Blakemore, S.-J. & Wolpert, D. M. (2000)

Abnormalities in the awareness and control of

action. Philosophical Transactions of the Royal Society

of London. Series B, Biological Sciences, 335, 1771–

1788.

Geuze, R., Jongmans, M., Schoemaker, M. M. & Smits-

Engelsman, B. C. M. (2001) Clinical and research



diagnostic criteria for Developmental Coordination

Disorder: a review and discussion. Human Movement

Science, 20, 7–47.

Harris, I. M. & Miniussi, C. (2003) Parietal lobe con-

tribution to mental rotation demonstrated with

rTMS. Journal of Cognitive Neuroscience, 15, 315–

323.

Harris, I. M., Egan, G. F., Sonkkila, C., Tochon-Danguy,

H. J., Paxinos, G. & Watson, J. D. G. (2000) Selective

right parietal lobe activation during mental rotation.

Brain, 123, 65–73.

Henderson, S. E. & Sugden, D. A. (1992) Movement

Assessment Battery for Children. The Psychological

Corporation, London, UK.

Jeannerod, M. (2003) Simulation of action as a unifying

concept for motor cognition. In: Taking Action: Cog-

nitive Neuroscience Perspectives on Intentional Acts

(ed. S. H. Johnson-Frey), pp. 139–163. MIT Press,

Cambridge, MA, USA.

Kail, R. (1985) Development of mental rotation: a speed-

accuracy study. Journal of Experimental Child Psychol-

ogy, 40, 181–192.

Kail, R. & Park, Y. (1990) Impact of practice on speed of

mental rotation. Journal of Experimental Child Psy-

chology, 49, 227–244.

Kail, R., Pellegrino, J. & Carter, P. (1980) Developmental

changes in mental rotation. Journal of Experimental

Child Psychology, 29, 102–116.

Katschmarsky, S., Cairney, S., Maruff, P., Wilson, P. H.

& Currie, J. (2001) The ability to execute saccades on

the basis of efference copy: impairments in double-

step saccade performance in children with Develop-

mental Coordination Disorder. Experimental Brain

Research, 136, 73–78.

Kosslyn, S. M., Digirolamo, G. J., Thompson, W. L. &

Alpert, N. M. (1998) Mental rotation of objects ver-

sus hands: neural mechanisms revealed by positron

emission tomography. Psychophysiology, 35, 151–

161.

de Lange, F. P., Hagoort, P. & Toni, I. (2005) Neural

topography and content of movement respresenta-

tions. Journal of Cognitive Neuroscience, 17,

97–112.

Larkin, D. & Hoare, D. (1991) Out of Step: Coordinating

Kids’ Movement. Active Life Foundation, Nedlands,

WA, Australia.

Lindman, H. R. (1974) Analysis of Variance in Complex

Experimental Designs. W.H. Freeman, San Francisco,

CA, USA.

Martin, K. A., Moritz, S. E. & Hall, C. R. (1999) Imagery

use in sport: a literature review and applied model.

Sport Psychologist, 13, 245–268.

Maruff, P., Wilson, P. H., Trebilcock, M. & Currie, J.

(1999) Abnormalities of imagined motor sequences in

children with Developmental Coordination Disorder.

Neuropsychologia, 37, 1317–1324.

Parsons, L. M. (1987) Imagined spatial transformation

of one’s body. Journal of Experimental Psychology:

General, 116, 172–191.

Parsons, L. M. (1994) Temporal and kinematic proper-

ties of motor behavior reflected in mentally simulated

action. Journal of Experimental Psychology: Human

Perception and Performance, 20, 709–730.

Parsons, L. M. & Fox, P. T. (1998) The neural basis of

implicit movements used in recognising hand shape.

Cognitive Neuropsychology, 15, 583–615.

Parsons, L. M., Gabrieli, J. D. E., Phelps, E. A. &

Gazzaniga, M. S. (1998) Cerebrally lateralized mental

representations of hand shape and movement.

Journal of Neuroscience, 18, 6539–6548.

Roelofs, K., van Galen, G. P., Keijsers, G. P. J. &

Hoogduin, C. A. L. (2002) Motor initiation and

execution in patients with conversion paralysis.

Acta Psychologia, 110, 21–34.

Sirigu, A., Cohen, L., Duhamel, J. R., Pillon, B., Dubois,

B., Agid, Y. & Pierrot-Deseilligny, C. (1995) Congru-

ent unilateral impairments for real and imagined hand

movements. Neuro Report, 6, 997–1001.

Sirigu, A., Daprati, E., Ciancia, S., Giraux, P.,

Nighoghossian, N., Posada, A. & Haggard, P. (2004)

Altered awareness of voluntary action after damage to

the parietal cortex. Nature Neuroscience, 7, 80–84.

Tabachnick, B. G. & Fidell, L. S. (1996) Using Multivari-

ate Statistics, 3rd edn. Harper Collins College Publish-

ers, New York, NY, USA.

Thayer, Z. C., Johnson, B. W., Corballis, M. C. & Hamm,

J. P. (2001) Perceptual and motor mechanisms for

mental rotation of human hands. Neuro Report, 12,

3433–3437.

Tomasino, B., Borroni, P., Isaja, A. & Rumiati, R. I.

(2005) The role of the primary motor cortex in mental

rotation: a TMS study. Cognitive Neuropsychology, 22,

348–363.

Wilson, P. H., Maruff, P. & McKenzie, B. E. (1997)

Covert orienting of visuospatial attention in

children with Developmental Coordination Disorder.

Developmental Medicine and Child Neurology, 39,

736–745.

Wilson, P. H., Maruff, P., Ives, S. & Currie, J. (2001)

Abnormalities of motor and praxis imagery in chil-

dren with DCD. Human Movement Science, 20, 135–

159.

Wilson, P. H., Maruff, P., Butson, M., Williams, J., Lum,

J. & Thomas, P. R. (2004) Impairments in the internal

representation of movement in children with Devel-

opmental Coordination Disorder (DCD): a mental

rotation task. Developmental Medicine and Child Neu-

rology, 46, 754–759.



Wolpert, D. M. (1997) Computational approaches to

motor control. Trends in Cognitive Sciences, 1, 209–216.

Zacks, J. M., Mires, J., Tversky, B. & Hazeltine, E. (2002a)

Mental spatial transformations of objects and per-

spective. Spatial Cognition and Computation, 2, 315–

322.

Zacks, J. M., Ollinger, J. M., Sheridan, M. A. & Tversky,

B. (2002b) A parametric study of mental spatial trans-

formations of bodies. Neuroimage, 16, 857–872.

Zacks, J. M., Gilliam, F. & Ojemann, J. G. (2003) Selec-

tive disturbance of mental rotation by cortical stimu-

lation. Neuropsychologia, 41, 1659–1667.

Documents

Motor, visual and egocentric transformations in children with Developmental Coordination Disorder