Abstract— Planning of reach actions can be inferred through

examination of the scaling of kinematic features and through a

deterministic statistical model that delineates a premovement

plan and early compensatory adjustments. The influence of

stroke-induced hemiparesis on these motor planning features is

currently unknown. The purpose of this study is to determine

the feasibility of using a 3-dimensional (3-D) virtual

environment (VE) to investigate the planning of unconstrained

reach actions in individuals post-stroke. Two non-disabled

adults and one individual post-stroke reached to 9 targets

displayed in 3 directions (straight, right 45°, left 45°) and 3

extents (8 cm, 16 cm, 24 cm) for a total of 108 trials with each

arm. Reach trajectories were analyzed, kinematics variables

extracted, and a statistical model applied to infer motor

planning. All participants were able to complete the VR task

although changes were required to ease accuracy requirements

for the participant with stroke. In the non-disabled

participants, movement time averaged < 500 msec and mean

endpoint error increased based on target extent. The

participant with stroke had similar movement times (485 ± 98

msec) when reaching with the right/ipsilesional UE but longer

movement times (761 ± 202 msec) when reaching with the

left/contralateral UE. Endpoint errors increased based on

target extent in both arms in the participant with stroke but

errors were overall larger when reaching with the

left/contralateral UE. All participants scaled peak velocity to

target distance although absolute values differed based on the

arm used and the presence of stroke-induced hemiparesis.

Scaling of peak acceleration was less consistent across

participants. Overall, over 80% of movement distance could be

explained by peak velocity and target distance in the non-

disabled participants when trials were collapsed across

direction. When reaching with the right/ipsilesional UE, the

participant with stroke demonstrated an increased reliance on

the premovement plan and decreased use of early compensatory

adjustments. A similar trend was seen with the

left/contralesional UE. Our results suggest that it is feasible to

use an immersive VE to investigate the planning of 3-D reach

actions in non-disabled adults and an individual with stroke-

induced hemiparesis. Future work will develop a calibration

system to determine system errors, compare performance in the

VE with an analogous real-word task, and extend data

collection to a larger group of individuals post-stroke.

Manuscript received April 11, 2008.

J. C. Stewart and J. Gordon are with the Division of Biokinesiology and Physical Therapy at the School of Dentistry, University of Southern California, Los Angeles, CA 90089 USA (phone: 323-442-1196; fax: 323-

442-1515; e-mail: jcstewar@usc.edu, jamesgor@usc.edu). C. J. Winstein is with the Division of Biokinesiology and Physical

Therapy at the School of Dentistry and Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles, CA 90089 USA (e-mail: winstein@usc.edu).

I. INTRODUCTION

HE ability to plan motor actions in advance of

movement onset is an essential component of skilled

functional task performance. Without anticipatory planning,

movement would be slow and laborious as peripheral

feedback is converted to movement commands. Goal-

directed reach actions incorporate feedforward mechanisms

that require anticipatory planning prior to movement onset

and feedback components [1-3]. As movement evolves,

sensory feedback information can be used to implement

corrections to the motor commands that are adaptive to task

demands, environmental conditions, and initial error [1, 4].

Sensitive kinematic measures of a reach action such as initial

peak velocity [2, 5, 6], initial peak acceleration [7, 8], and

initial trajectory direction [9, 10] can be used to infer

anticipatory capability. A consistent finding is the scaling of

initial peak acceleration and peak velocity to movement

extent [6, 7, 11].

Error corrections can be observed early after movement

onset suggesting the central nervous system anticipates the

consequences of movement commands (e.g. velocity, joint

position) and compares these with actual performance

variables immediately upon movement onset [3, 12]. Such a

motor system provides flexibility and efficiency in that a

precise motor command is not necessary for every possible

movement as the initial command can be tuned as the

movement unfolds to optimize accuracy [4, 13]. Evidence

that humans use a feedforward, premovement plan as well as

early compensatory adjustments in generating goal-directed

upper extremity (UE) actions has been demonstrated in the

performance of single-joint tasks [14-16] and in reach

actions to 2-dimensional (2-D) targets [11] but has not been

studied with unconstrained reach actions to 3-dimensional

(3-D) targets. Control strategies for unconstrained reach

movements may differ from those for more constrained reach

movements [17] suggesting a need to investigate the role of

motor planning under task conditions that more closely

resemble the requirements of functional tasks.

After stroke, individuals can be left with residual motor

deficits that impact performance of functional motor

activities and overall quality of life [18, 19]. A frequently

reported contribution to these functional limitations is an

inability to incorporate the paretic hand into daily activities

[18, 20, 21]. Deficits in the ability to execute targeted arm

movements are well documented and include impairments in

Use of a Virtual Environment to Investigate Planning of

Unconstrained Reach Actions after Stroke: A Feasibility Study

Jill Campbell Stewart, James Gordon and Carolee J. Winstein

T

978-1-4244-2701-7/08/$25.00 ©2008 IEEE 13

endpoint accuracy [22], movement speed [22, 23], joint

coordination [24, 25], and patterns of muscle activation [26]

but less is known about anticipatory planning ability. While

motor planning deficits after stroke have been demonstrated

[10, 14, 15, 27-32], research to date has not always

controlled for the limb used (contralesional versus

ipsilesional), lesion location, or lesion side. Additionally,

limited research has been conducted using tasks that require

a 3-D, unconstrained reach [10, 28, 31] and the preservation

of scaling of kinematic variables after stroke has not been

systematically investigated.

Virtual reality (VR) is an emerging technology with

promise as a rehabilitation tool. VR provides a unique

opportunity to investigate the planning of unconstrained

reach actions to 3-D targets in a manner not possible with a

real-world protocol by allowing the ability to: 1)

systematically control target position in 3-D space; 2)

sequentially present a single target while other targets are not

visible; 3) block vision of the moving arm to limit the

influence of feedback online during movement while

simultaneously allowing vision of the target and

environment; and 4) provide constraints on movement time

to facilitate fast action (versus slower movements where the

influence of feedback is greater). Several recent studies have

demonstrated the feasibility of using a virtual environment

(VE) for motor activities in individuals recovering from

stroke [33-41]. Yet, there is limited scientific research about

the effects of VR on reach kinematics. To date, evidence

suggests that the execution of reach movements in a VE have

kinematic features that have both similarities and differences

to analogous movements in the real-world [42-45]. A

systematic approach to these aspects of motor control is

needed before the full potential of VR as a tool to investigate

the planning of reach actions can be realized.

The purpose of this study is to determine the feasibility

of using a 3-D VE to investigate the planning of

unconstrained reach actions in individuals post-stroke. The

system was tested with two non-disabled adults and one adult

with stroke-induced hemiparesis. The usability of the VR

system is described based on the ability of the participants to

complete the protocol and the system to capture the

necessary data. Reach trajectories were analyzed to

determine if kinematic variables used to infer planning could

be extracted. Finally, we evaluated the feasibility of

applying a deterministic statistical model to infer

anticipatory planning of reach actions.

II. METHODS

A. Participants

Two non-disabled adults (ages 31 and 35 years) and one

adult who was 51 months post-unilateral stroke (age 75

years) resulting in left hemiparesis participated in this pilot

experiment. All participants were right-hand dominant.

Assessments in the participant with stroke indicated

moderate motor impairment with the left UE (UE Fugl-

Meyer score = 48/66), no cognitive impairment (Mini-

Mental State Exam score = 29/30), and no evidence of

hemispatial neglect (star cancellation score = 54/54).

B. Virtual Environment

An immersive VR display unit that interfaces with The

MotionMonitor (Innovative Sports Training) was used in this

experiment. This system integrates a stereoscopic workspace

(800 × 600 mm) from SenseGraphics with motion capture

capability via an electromagnetic system (Ascension).

Images are presented through a projector to a mirror that

reflects the objects into the workspace at a resolution of 800

× 600 pixels and a refresh rate of 120 Hz. Shutter glasses

sampling at 60 Hz per eye are worn by the user to provide 3-

D view of the VE. Integrated head tracking through a sensor

attached to the glasses provides visual compensation for

head movements. The virtual environment consists of a

simple background in which targets are displayed in 3-D

from a first-person perspective.

C. Experimental Task and Procedure

The home position and target locations were represented

by 25 mm spheres. An electromagnetic marker placed on the

index finger acted as the interface with the VE (displayed as

a white sphere) and measured position change during each

trial at a sampling rate of 120 Hz. Nine targets, 3 directions

(straight, right 45°, left 45°) by 3 distances (8 cm, 16 cm, 24

cm), on a single plane were presented in relation to a

‘virtual’ start position (Fig. 1). A single trial consisted of

moving the hand off of the start position after an auditory

‘Go’ signal and reaching to the indicated target as fast and as

accurate as possible. A maximum of 2 seconds was allowed

to initiate and complete the response. Vision of the arm and

hand were blocked during all trials by darkening the work

space and placing a black glove on the hand in order to

increase reliance on planning processes and limit the ability

to use visual feedback online during movement. Performance

Start

45° 45° 8 cm

16 cm

24 cm

25 mm

Fig. 1. Nine targets presented in a 3-D virtual environment. All targets

were in the same z-plane. Black circle = target.

14

feedback as to final endpoint position and target position was

provided after every trial. A total of 12 trials to each target

were performed in 4 blocks of 27 trials (108 total trials) with

the left/ipsilesional arm followed by the right/contralesional

arm.

-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.20.25

0.3

0.35

0.4

0.45

0.5

0.55

X (m)

Y (m)

0.25 0.3 0.35 0.4 0.45 0.5

-0.25

-0.2

-0.15

-0.1

Y (m)

Z (m)

A) B)

-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.20.25

0.3

0.35

0.4

0.45

0.5

0.55

X (m)

Y (m)

0.25 0.3 0.35 0.4 0.45 0.5

-0.25

-0.2

-0.15

-0.1

Y (m)

Z (m)

F) E)

-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.20.25

0.3

0.35

0.4

0.45

0.5

0.55

X (m)

Y (m)

0.25 0.3 0.35 0.4 0.45 0.5

-0.25

-0.2

-0.15

-0.1

Y (m)

Z (m)

C) D)

Fig. 3. Endpoint position for Control participant 1 reaching with the right UE in: A) X, Y (view from above); B) Y, Z (view from side). Endpoint

position for Stroke participant when reaching with the right/ipsilesional UE in X, Y (C) and Y, Z (D) and with the left/contralesional UE in X, Y (E)

and Y, Z (F). Black circles = target location, black square = start location, red circles = 8 cm target endpoints, green circles = 16 cm target endpoints,

blue circles = 24 cm target endpoints.

15

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

time

velocity (m/sec)

0 0.2 0.4 0.6 0.8 1-8

-6

-4

-2

0

2

4

6

8

time

acceleration (m*m/sec)

-0.15-0.1

-0.050

0.05

0.25

0.3

0.35

0.4

0.45

-0.24

-0.22

-0.2

xy

z

-0.1 -0.15

-0.050

0.05

0.25

0.3

0.35

0.4

0.45

-0.24

-0.22

-0.2

x

y

z

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

time

velocity (m/sec)

0 0.2 0.4 0.6 0.8 1-8

-6

-4

-2

0

2

4

6

8

time

acceleration (m*m/sec)

-0.1 -.15

-0.050

0.05

0.25

0.3

0.35

0.4

0.45

-0.24

-0.22

-0.2

x

y

z

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

time

velocity (m/sec)

0 0.2 0.4 0.6 0.8 1-8

-6

-4

-2

0

2

4

6

8

time

acceleration (m*m/sec)

A) C) B)

X Y Y

Z Z

X

Z

Y X

Time (sec) Time (sec) Time (sec)

Velocity (m/sec)

Acceleration (m

2/sec)

Fig. 2. Reach trajectory, velocity, and acceleration for a single trial to the 16 cm target in the left direction. Trajectories are plotted in 3-D. The participant sits facing the X axis with forward movement being positive Y and upward movement being positive Z. Red sphere = target, blue sphere =

start position, black sphere = movement endpoint. A) Control 2 reaching with the right UE; B) participant with stroke reaching with the

right/ipsilesional UE; C) participant with stroke reaching with the left/contralesional UE.

D. Data Analysis

Position in x, y, and z were recorded for each trial. All

data were filtered with a 2nd

order, low-pass Butterworth

filter using a 10 Hz cutoff. Tangential velocity and

acceleration were derived from the resultant change in

position. Movement for each trial was defined from the

onset to the offset of the velocity peak. Onset/offset was

determined using a custom MatLab program that searched

backward/forward through the time series from the peak until

velocity dropped to 0 m/sec or changed directions. Due to

overt adjustments present in the velocity profiles of the

participant with stroke, an additional criteria of a minimum

velocity of 0.03 m/sec was add to the cutoff determination.

For each trial, the following variables were extracted: peak

velocity, peak acceleration, movement distance, movement

time, and resultant endpoint error. Descriptive statistics

were used to summarize findings for each participant.

A previously described statistical model [11, 15, 16] was

used to determine the percent variance of movement distance

explained by the initial premovement plan and the percent

explained by early compensatory adjustments. This model

uses regression and correlation techniques to determine

whether movement distance is achieved through advanced

planning based on peak acceleration or peak velocity, early

compensatory adjustments made to the trajectory based on

target distance, or a combination of both when performing a

fast, discrete reach action. For this feasibility study, the

model was tested using both peak acceleration and peak

velocity as predictor variables.

III. RESULTS

A. System Usability and Task Feasibility

All participants were able to perform the VR task with each

arm. Control 1 successfully completed 107 of 108 trials

(99%) with the right UE and 106 of 108 (98%) trials with the

left UE. For Control 2, a system error occurred during

performance of the second block with the left UE causing the

display software to shut down. Ten trials were not

completed due to this error. Therefore, this participant

performed only 98 trials with the left UE. Of the trials

performed, 105 of 108 trials (97%) with the right UE and 96

of 98 (98%) trials with the left UE were completed

successfully. All unsuccessful trials resulted from errors in

task performance by the participant such as not obtaining the

‘Home’ position accurately or not making a movement

toward the target prior to the trial timing out. These trials

were discarded and not included in data analysis.

16

The participant with stroke was unable to meet the

criteria set for being on the ‘Home’ position with the

ipsilesional UE. Therefore, the error tolerance between the

home position and the cursor was increased from 0.01 to

0.03 meters to allow task completion. Subsequently, this

participant completed 81 trials (3 blocks) each with the

ipsilesional and contralesional UE; the final block of 27

trials was not performed due to time constraints. Of the 81

trials performed, 72 trials (89%) were successfully

completed with the right/ipsilesional UE and 78 (96%) with

the left/contralesional UE. Unsuccessful trials resulted from

errors in task performance (same as control participants) and

were not included in further analyses.Task ExecutionFigure

2 presents data for a representative trial of one non-disabled

participant reaching with the right UE and the participant

with stroke reaching with both arms. In control participants,

the path taken to the target was relatively straight with a

single peaked velocity. Mean peak velocity to individual

targets ranged from 0.321 to 1.151 m/sec with the right UE

and 0.363 to 1.305 m/sec with the left UE. Mean peak

acceleration ranged from 2.94 to 12.38 m/sec2

with the right

UE and 3.23 to 11.30 m/sec2

with the left UE. While

absolute values of peak velocity (0.268 to 0.859 m/sec) and

peak acceleration (2.63 to 8.92 m/sec2) were slightly lower

when reaching with the right/ipsilesional UE, patterns of

performance as to straightness of hand path and single

peaked velocity profile were similar to controls. With the

left/contralesional UE, the participant with stroke tended to

take a longer path to the target and velocities frequently

contained multiple peaks with more than one 0 crossing in

acceleration. Mean peak velocity to individual targets

ranged from 0.333 to 0.787 m/sec, slightly lower than that

for control participants and reaches with the right/ipsilesional

UE. Mean peak acceleration was also lower in the

left/contralesional UE and ranged from 2.80 to 6.85 m/sec2.

Mean movement time across all trials were 442 ± 117 msec

for Control participant 1 and 465 ± 93 msec for Control

participant 2. The participant with stroke had similar

movement times (485 ± 98 msec) when reaching with the

right/ipsilesional UE but longer movement times (762 ± 202

msec) when reaching with the left/contralateral UE.

Endpoint positions for individual trials for one non-

disabled participant and the participant with stroke are

shown in Fig. 3. Overall, mean absolute error increased as

target distance increased (Fig. 4) for all participants although

errors were larger in the participant with stroke when

reaching with the left/contralesional UE. Additionally, errors

varied based on target direction (Fig. 4). Errors appeared to

by systematic in direction towards negative X, negative Y,

and positive Z (Fig. 3).

B. Anticipatory Planning

All participants scaled peak velocity to target distance

(Fig. 5) although absolute values differed based on the arm

used and the presence of stroke-induced hemiparesis.

Scaling of peak acceleration was less consistent across

participants. In reaches to the right targets with the left UE,

both the control participants and the individual with stroke

(Fig. 5C) had an initial, lower acceleration peak followed

quickly by a larger peak.

A deterministic statistical model as previously described

[11, 14-16] was tested with both peak velocity and peak

acceleration. Peak acceleration did not correlate well with

target distance and therefore was not a strong predictor of

movement distance. The results of the model using peak

velocity as the predictor variable are shown in Figure 6. All

three participants used both a premovement plan (% variance

of movement distance explained by peak velocity) and early

compensatory adjustments (%variance of movement distance

explained when target distance was added to the model)

when reaching to 3-D targets. Overall, over 80% of

movement distance could be explained by these two

components when trials were collapsed across direction.

When reaching with the right/ipsilesional UE, the participant

with stroke demonstrated an increased reliance on the

premovement plan and decreased use of early compensatory

.08 .16 .24

Target Distance (m)

0.00

0.05

0.10

0.14

0.19

0.24

Mean Error (m)

Left UE

Right Targets Straight Targets Left Targets

.08 .16 .24

Target Distance (m)

0.00

0.03

0.06

0.10

0.13

0.16

Mean Error (m)

Right Targets Straight Targets Left TargetsB)A)

Fig. 4. Endpoint error as a function of target direction and target extent for: A) Control 2 reaching with the left UE; and B) participant with stroke reaching with the left/contralesional UE (note change in Y axis scale).

17

adjustments. The finding was consistent with previous

findings with single-joint movements [15]. A similar trend

was seen with the left/contralesional UE. The model

explained a higher proportion of the variance to movement

distance when each movement direction was analyzed

separately (>90% for most conditions). Differences between

the control participants and the individual with stroke were

not as consistent when data was analyzed by target direction.

IV. DISCUSSION

VR provides a unique opportunity to extend previous

work on the planning of reach actions with a single-joint or

to 2-D targets [6-8, 11] to unconstrained reach actions to 3-D

targets, movements that more closely resemble those

required for functional tasks. The purpose of this study was

to determine the feasibility of using a 3-D, immersive VE to

investigate the planning of unconstrained reach actions in

individuals with stroke-induced hemiparesis. We were able

to create task conditions that allowed the systematic

presentation of sequential 3-D targets, vision of the target

and workspace while blocking vision of the arm during

movement, and constraints on movement time. Both non-

disabled participants and the participant with stroke were

able to complete the VR based task. Caution should be taken

when interpreting the results of the kinematics measures and

statistical model presented here due to the small number of

subjects. Future work will determine the feasibility of using

this paradigm in a larger group of individuals with stroke and

provide sufficient power for interpretation of behavioral

findings.

The participant with stroke had a larger percentage of

error trials when reaching with the ipsilesional versus the

contralesional UE. This difference between arms is likely

due to the order in which the task was performed. The

participant reached with the ipsilesional UE first and may

have still been learning and adjusting to the task constraints.

When reaching with the contralesional UE, the general

features of the task may have been better understood

resulting in fewer error trials.

Accuracy requirements for successful achievement of the

‘Home’ position had to be relaxed in order for the participant

with stroke to perform the task. The purpose of using a

‘Home’ position at the start of each trial was to provide a

consistent start location across trials and subjects.

Loosening the accuracy requirements for the participant with

stroke made the start position variable, impacting the actual

distance between the start and target locations. In order to

better control the start position for different subjects and

arms, a physical start switch will be added and integrated

into the VR system.

This study demonstrates the feasibility of using the

described VR system to collect the large number of trials and

extract the kinematic variables needed to perform a detailed

investigation of the planning of reach actions. The

A)

0 0.1 0.2 0.3 0.4 0.5 0.60

0.2

0.4

0.6

0.8

1

Time (sec)

Velocity (m/sec)

8 cm

16 cm

24 cm

0 0.1 0.2 0.3 0.4 0.5 0.6-8

-6

-4

-2

0

2

4

6

8

10

Time (sec)

Acceleration (m*m/sec)

8 cm

16 cm

24 cm

B)

0 0.1 0.2 0.3 0.4 0.5 0.60

0.2

0.4

0.6

0.8

1

Time (sec)

Velocity (m/sec)

8 cm

16 cm

24 cm

0 0.1 0.2 0.3 0.4 0.5 0.6-8

-6

-4

-2

0

2

4

6

8

10

Time (sec)

Acceleration (m*m/sec)

8 cm

16 cm

24 cm

C)

0 0.2 0.4 0.6 0.8 1 1.2 1.40

0.2

0.4

0.6

0.8

1

Time (sec)

Velocity (m

/sec)

8 cm

16 cm

24 cm

0 0.2 0.4 0.6 0.8 1 1.2 1.4-8

-6

-4

-2

0

2

4

6

8

10

Time (sec)

Acceleration (m*m/sec)

8 cm

16 cm

24 cm

Time (sec) Time (sec) Time (sec)

Velocity (m/sec)

Acceleration (m

2

/sec)

Fig. 5. Mean velocity and acceleration profiles to the right targets: A) with the right UE for Control 1; B) with the right/ipsilesional UE for the stroke

participant; C) with the left/contralesional UE for the stroke participant (note the change in X axis scale). Each line represents a different target distance.

18

individual with stroke presented with moderate motor

impairment (UE Fugl-Meyer motor score of 48/66). Yet, he

was able to complete 81 trials with the contralesional UE.

The last block of 27 trials was not completed due to time

constraints related to the software changes that had to be

made to alter the error tolerance as described above. The

participant may have been able to complete all 108 trials had

time allowed. Applying a deterministic statistical model to

understand the planning of unconstrained reach actions to 3-

D targets yielded similar results as previous studies in non-

disabled adults performing a goal-directed, single-joint reach

[15] and a reach to a 2-D target [11] and in individuals with

sensorimotor stroke performing a single-joint reach with the

ipsilesional UE [15]. We also applied this model to the

reach actions performed with the contralesional UE which

has not been previously reported. This method will provide

useful insight into whether impairments in the planning of

reach actions are present after stroke and inform the

development of novel rehabilitation techniques.

Mean endpoint errors were larger than has been

reported in previous studies [6, 11] with errors greater than

10 cm to some targets. Additionally, as shown in Figure 3,

endpoint position tended to be consistent but off the target in

the control participants. This consistency suggests that

participants perceived ‘good’ performance (i.e. being close

to the target) from the feedback provided after each trial.

They did not appear to be making corrections from trial to

trial to improve accuracy. The errors also appear in a

consistent direction (negative X, negative Y, positive Z)

making the possibility of systematic errors in the system

reasonable. The direction of errors was similar in the

participant with stroke although the overall variability was

higher. The pattern of endpoint errors found in this study

may be due to errors in the presentation of targets by the

display software, measurement error by the system, error in

workspace calibration, lack of sufficient instructions and

practice for participants to adequately determine being ‘on’

or ‘off’ target, or true errors in planning related to task

demands. Research is currently underway to determine

which of these factors may be impacting the data.

V. CONCLUSION

Our results suggest that it is feasible to use an immersive

VE to investigate the planning of 3-D reach actions in non-

disabled adults and an individual with moderate motor

impairment due to stroke. Relevant kinematic variables

could be extracted and a deterministic statistical model

applied. Future work will focus on task and system changes

to improve experimental controls, development of a

calibration device to allow systematic determination of the

source of errors, comparison of findings within the VE with

those from an analogous real-world task, and extension of

data collection to a larger group of individuals post-stroke.

Fig. 6. Premovement plan (black bar) and early compensatory adjustments (gray bar) based on peak velocity for the control participant shown in

Figure 4 when reaching with the right UE and the stroke participant when reaching with the right/ipsilesional UE and left/contralesional UE. The

number at the top of each bar represents the total variance explained by both components. The top row contains the results for all targets collapsed

across direction. The bottom row contains the results for two directions (right targets and left targets).

19

ACKNOWLEDGMENT

Innovative Sports Training provided the immersive

display unit for this experiment.

REFERENCES

[1] P. N. Sabes, "The planning and control of reaching movements," Curr

Opin Neurobiol, vol. 10, pp. 740-6, Dec 2000. [2] V. Brooks, "Some examples of programmed limb movements," Brain

Res, vol. 71, pp. 299-308, May 17 1974. [3] M. Desmurget and S. Grafton, "Forward modeling allows feedback

control for fast reaching movements," Trends Cogn Sci, vol. 4, pp. 423-431, Nov 1 2000.

[4] E. Todorov, "Optimality principles in sensorimotor control," Nat

Neurosci, vol. 7, pp. 907-15, Sep 2004. [5] S. Bouisset and F. Lestienne, "The organisation of a simple voluntary

movement as analysed from its kinematic properties," Brain Res, vol. 71, pp. 451-7, May 17 1974.

[6] J. Gordon, M. F. Ghilardi, and C. Ghez, "Accuracy of planar reaching movements. I. Independence of direction and extent variability," Exp

Brain Res, vol. 99, pp. 97-111, 1994. [7] J. Gordon, M. F. Ghilardi, S. E. Cooper, and C. Ghez, "Accuracy of

planar reaching movements. II. Systematic extent errors resulting from

inertial anisotropy," Exp Brain Res, vol. 99, pp. 112-30, 1994.

[8] R. L. Sainburg and S. Y. Schaefer, "Interlimb differences in control of

movement extent," J Neurophysiol, vol. 92, pp. 1374-83, Sep 2004.

[9] J. Wang and R. L. Sainburg, "The symmetry of interlimb transfer

depends on workspace locations," Exp Brain Res, vol. 170, pp. 464-

71, Apr 2006.

[10] D. J. Reinkensmeyer, A. McKenna Cole, L. E. Kahn, and D. G.

Kamper, "Directional control of reaching is preserved following

mild/moderate stroke and stochastically constrained following severe

stroke," Exp Brain Res, vol. 143, pp. 525-30, Apr 2002.

[11] J. Messier and J. F. Kalaska, "Comparison of variability of initial

kinematics and endpoints of reaching movements," Exp Brain Res,

vol. 125, pp. 139-52, Mar 1999.

[12] M. Desmurget, C. M. Epstein, R. S. Turner, C. Prablanc, G. E.

Alexander, and S. T. Grafton, "Role of the posterior parietal cortex in

updating reaching movements to a visual target," Nat Neurosci, vol.

2, pp. 563-7, Jun 1999.

[13] R. S. Woodworth, "The accuracy of voluntary movements," Psychol

Mongr, vol. 3, pp. 1-114, 1899.

[14] B. E. Fisher, L. Boyd, and C. J. Winstein, "Contralateral cerebellar

damage impairs imperative planning but not updating of aimed arm

movements in humans," Exp Brain Res, May 5 2006.

[15] B. E. Fisher, C. J. Winstein, and M. R. Velicki, "Deficits in

compensatory trajectory adjustments after unilateral sensorimotor

stroke," Exp Brain Res, vol. 132, pp. 328-44, Jun 2000.

[16] J. Gordon and C. Ghez, "Trajectory control in targeted force impulses. III. Compensatory adjustments for initial errors," Exp Brain Res, vol. 67, pp. 253-69, 1987.

[17] M. Desmurget, M. Jordan, C. Prablanc, and M. Jeannerod,

"Constrained and unconstrained movements involve different control strategies," J Neurophysiol, vol. 77, pp. 1644-50, Mar 1997.

[18] N. E. Mayo, S. Wood-Dauphinee, R. Cote, L. Durcan, and J. Carlton,

"Activity, participation, and quality of life 6 months poststroke," Arch

Phys Med Rehabil, vol. 83, pp. 1035-42, Aug 2002. [19] S. M. Lai, S. Studenski, P. W. Duncan, and S. Perera, "Persisting

consequences of stroke measured by the Stroke Impact Scale," Stroke,

vol. 33, pp. 1840-4, Jul 2002.

[20] S. M. Lai, S. Studenski, L. Richards, S. Perera, D. Reker, S. Rigler, and P. W. Duncan, "Therapeutic exercise and depressive symptoms after stroke," Journal of the American Geriatrics Society, vol. 54, pp.

240-7, Feb 2006. [21] D. S. Nichols-Larsen, P. C. Clark, A. Zeringue, A. Greenspan, and S.

Blanton, "Factors influencing stroke survivors' quality of life during subacute recovery," Stroke, vol. 36, pp. 1480-4, Jul 2005.

[22] C. E. Lang, J. M. Wagner, A. J. Bastian, Q. Hu, D. F. Edwards, S. A.

Sahrmann, and A. W. Dromerick, "Deficits in grasp versus reach

during acute hemiparesis," Exp Brain Res, vol. 166, pp. 126-36, Sep 2005.

[23] J. M. Wagner, C. E. Lang, S. A. Sahrmann, D. F. Edwards, and A. W. Dromerick, "Sensorimotor impairments and reaching performance in subjects with poststroke hemiparesis during the first few months of recovery," Phys Ther, vol. 87, pp. 751-65, Jun 2007.

[24] M. F. Levin, "Interjoint coordination during pointing movements is disrupted in spastic hemiparesis," Brain, vol. 119 ( Pt 1), pp. 281-93, Feb 1996.

[25] M. F. Levin, S. M. Michaelsen, C. M. Cirstea, and A. Roby-Brami, "Use of the trunk for reaching targets placed within and beyond the reach in adult hemiparesis," Exp Brain Res, vol. 143, pp. 171-80, Mar 2002.

[26] J. M. Wagner, A. W. Dromerick, S. A. Sahrmann, and C. E. Lang, "Upper extremity muscle activation during recovery of reaching in subjects with post-stroke hemiparesis," Clin Neurophysiol, vol. 118, pp. 164-76, Jan 2007.

[27] J. J. Daly, Y. Fang, E. M. Perepezko, V. Siemionow, and G. H. Yue, "Prolonged cognitive planning time, elevated cognitive effort, and relationship to coordination and motor control following stroke," IEEE Trans Neural Syst Rehabil Eng, vol. 14, pp. 168-71, Jun 2006.

[28] P. S. Pohl, D. L. Filion, and S. H. Kim, "Effects of practice and unpredictable distractors on planning and executing aiming after stroke," Neurorehabil Neural Repair, vol. 17, pp. 93-100, Jun 2003.

[29] M. R. Velicki, C. J. Winstein, and P. S. Pohl, "Impaired direction and

extent specification of aimed arm movements in humans with stroke-

related brain damage," Exp Brain Res, vol. 130, pp. 362-74, Feb

2000.

[30] K. Y. Haaland, J. L. Prestopnik, R. T. Knight, and R. R. Lee,

"Hemispheric asymmetries for kinematic and positional aspects of

reaching," Brain, vol. 127, pp. 1145-58, May 2004.

[31] D. G. Kamper, A. N. McKenna-Cole, L. E. Kahn, and D. J.

Reinkensmeyer, "Alterations in reaching after stroke and their relation

to movement direction and impairment severity," Arch Phys Med

Rehabil, vol. 83, pp. 702-7, May 2002.

[32] S. Y. Schaefer, K. Y. Haaland, and R. L. Sainburg, "Ipsilesional

motor deficits following stroke reflect hemispheric specializations for

movement control," Brain, Jul 11 2007.

[33] M. Holden, E. Todorov, J. Callahan, and E. Bizzi, "Virtual

environment training improves motor performance in two patients

with stroke: case report," Neurol Rep, vol. 23, pp. 57-67, 1999.

[34] A. S. Merians, D. Jack, R. Boian, M. Tremaine, G. C. Burdea, S. V.

Adamovich, M. Recce, and H. Poizner, "Virtual reality-augmented

rehabilitation for patients following stroke," Phys Ther, vol. 82, pp.

898-915, Sep 2002.

[35] A. S. Merians, H. Poizner, R. Boian, G. Burdea, and S. Adamovich,

"Sensorimotor training in a virtual reality environment: does it

improve functional recovery poststroke?," Neurorehabil Neural

Repair, vol. 20, pp. 252-67, Jun 2006.

[36] J. C. Stewart, S. C. Yeh, Y. Jung, H. Yoon, M. Whitford, S. Y. Chen,

L. Li, M. McLaughlin, A. Rizzo, and C. J. Winstein, "Pilot trial results from a virtual reality system designed to enhance recovery of skilled arm and hand movements after stroke.," in International

Workshop on Virtual Rehabilitation, New York, NY, 2006, pp. 11-

17. [37] J. C. Stewart, S. C. Yeh, Y. Jung, H. Yoon, M. Whitford, S. Y. Chen,

L. Li, M. McLaughlin, A. Rizzo, and C. J. Winstein, "Intervention to

enhance skilled arm and hand movements after stroke: A feasibility

study using a new virtual reality system," J Neuroengineering

Rehabil, vol. 4, p. 21, 2007. [38] J. E. Deutsch, J. Latanio, G. Burdea, and R. Boian, "Post-stroke

rehabilitation with the Rutgers Ankle System: a case study,"

Presence, vol. 10, pp. 416-430, 2001. [39] S. Flynn, P. Palma, and A. Bender, "Feasibility of using the Sony

PlayStation 2 gaming platform for an individual poststroke: a case

report," J Neurol Phys Ther, vol. 31, pp. 180-9, Dec 2007. [40] S. H. You, S. H. Jang, Y.-H. Kim, M. Hallett, S. H. Ahn, Y.-H.

Kwon, J. H. Kim, and M. Y. Lee, "Virtual Reality-Induced Cortical Reorganization and Associated Locomotor Recovery in Chronic Stroke: An Experimenter-Blind Randomized Study," Stroke, vol. 36,

pp. 1166-1171, June 1, 2005 2005.

20

[41] Y. R. Yang, M. P. Tsai, T. Y. Chuang, W. H. Sung, and R. Y. Wang, "Virtual reality-based training improves community ambulation in individuals with stroke: A randomized controlled trial," Gait Posture,

Mar 19 2008. [42] A. Y. Dvorkin, M. Shahar, and P. L. Weiss, "Reaching within video-

capture virtual reality: using virtual reality as a motor control paradigm," Cyberpsychol Behav, vol. 9, pp. 133-6, Apr 2006.

[43] A. Viau, A. G. Feldman, B. J. McFadyen, and M. F. Levin, "Reaching in reality and virtual reality: a comparison of movement kinematics in

healthy subjects and in adults with hemiparesis," J Neuroengineering

Rehabil, vol. 1, p. 11, Dec 14 2004. [44] J. Broeren, K. S. Sunnerhagen, and M. Rydmark, "A kinematic

analysis of a haptic handheld stylus in a virtual environment: a study in healthy subjects," J Neuroeng Rehabil, vol. 4, p. 13, 2007.

[45] T. Kuhlen, K. F. Kraiss, and R. Steffan, "How VR-based reach-to-grasp experiments can help to understand movement organization within the human brain," Presence, vol. 9, pp. 350-359, 2000.

21