Peter Wilson

School of Psychology,

Australian Catholic University

Melbourne Campus

A developmental cognitive neuroscience perspective on motor rehabilitation: The case for VR-augmented

therapy

Peter H. Wilson1, Dido Green2, Jonathan Duckworth2, Jan Piek4

1School of Psychology, ACU, Melbourne, Australia2School of Occupational Therapy, Oxford-Brookes University, Oxford UK

3Creative Media & Communication, RMIT University, Melbourne4Curtin University, Perth, Australia

Collaborating Team

Aims1. Sketch the core processes that help explain motor

development (and deviations from it)

2. Present a conceptual model for VR-based, paediatric movement rehabilitation

3. Highlight the importance of multimodal augmented feedback in training predictive control

What processes drive typical skill development in children?

Body schema – linked to multimodal integration

Internal modeling –

Mapping output signals and their effects on the body

Enables prediction and online control

Prediction develops rapidly over childhood:

Force control

Postural adjustments

Online control

Progressive integration of feedback & feedforward mechanisms; well

developed by 8-9 years.

Developmental Coordination DisorderDCD = Motor clumsiness in children, not explained by a medical condition (like MS, etc.)

DSM-V category

Issues

What are core deficits?

Is there a neural locus?

Is severe DCD a mild form of CP??

Modeling atypical development: A multi-level perspective

Sx of DCD Sx of DCD

Atypical movement patternsAtypical movement patterns

Neurocognitive deficitNeurocognitive deficit

Atypical neural maturationAtypical neural maturation

Genetic markersGenetic markers

E

N

V

I

R

O

N

M

E

N

T

Key levels of analysis

• Maturational integrity of the CNSMaturational integrity of the CNS

• Motor control mechanismsMotor control mechanisms

• Biomechanics: kinetics & kinematicsBiomechanics: kinetics & kinematics

• Cognitive control mechanisms Cognitive control mechanisms

• Energetics – arousal, motivation, etc. Energetics – arousal, motivation, etc.

• OthersOthers

Top-down is still important

Meta-Analysis

Aims

1.Quantitative review of the published performance data since 1997

2.Identify the motor control, learning, and cognitive control deficits that best discriminate between children with and without DCD

3.Identify patterns that suggest causal mechanisms

Wilson et al. (in press). Developmental Medicine & Child Neurology.

Contributing Studies

Final “sample” of studies = 129.

1785 effect sizes

Main Findings - Motor Control

The main motor control deficits relate to:

1. Predictive control / forward internal modeling

> 5 categories with v. high effect size (d > 1.0)

2. Timing and coordination of interceptive actions

3. Postural stabilisation, adjustment & anticipation

4. Rhythmic coordination

5. Gait dynamics

Neuro-computational model 1

Magescas et al. (2009)

Limb Movement

Error signal

Error signal (used to train the model = learning)

efference copy

Effector Movement

Sensory Feedback

Error signal

Motor Command

Intended State

Controller [maps perception to

movement]

Forward Model [maps movement to perception]

Estimated Actual State (or sensed state)

Predicted State (Parietal &

Cerebellum)

Goal (Frontal &

ACC) Affordances coordinate

transformations (Parietal: IPS)

Neuro-computational model 2

Forward modeling – the simple one!

From Nowak (2007)

Issues for Treatment

Practice does not guarantee skill in CP and DCD!

Excess neural noise greater performance variability in DCD

Feedback dependency in DCD, with costs for speed & online control

Need techniques that will build predictive control

Concurrent augmented feedback + Attentional cuing.

Augmented FeedbackForms of AF

Knowledge of Results (KR)

Knowledge of Performance (KP), and

Concurrent AF (C-AF).

The benefits of AF are well documented mainstream motor learning (Carson and Kelso 2004; Gordon & Magill 2012).

rehabilitation of brain injury (Winstein, Wing et al. 2003; van Vliet & Wulf, 2006).

BUT, the quality of evidence varies.

Little is known a/b the relative effect of visual, verbal, video and kinematic feedback, and optimal scheduling.

How does C-AF work?

C-AF (which promotes an external focus of attention) can assist recovery of upper-limb function in ABI (e.g., Quaney et al. 2010).

C-AF may do so by training motor prediction (aka forward modeling) – empirical question.

Sobering is the fact that most verbal instructions (esp. in clinical settings) are likely to induce an internal focus of attention (Durham, Van Vliet et al. 2009).

Training (motor) prediction using C-AF

C-AF serves two main purposes: 1.Provides children with additional input on the outcomes of their actions: reinforces the relationship b/n motor output & consequence. builds body schema

2.Encourages the child to focus their attention on the effects of their movement (Wulf & Prinz 2001).

Hypothesis: VR-augmented therapy works by training predictive control (and associated body schema “knowledge”)

Virtual workspace design – Embodied interaction

VR-based systems are the perfect vehicle for C-AF:

Multimodal C-AF

Aesthetic design

Intuitive (tangible) interfaces

Client-centred

Augmented Feedback

Exploratory interaction / “training”

Exploratory Tasks – Squiggles

Exploratory Tasks – Mixer

Exploratory Tasks – Swarm

Multimodal AF works in paediatric rehab (but is part of a treatment package)!

Green & Wilson (2011) - Mixed hemiplegic group

Green, Wilson, & Lin (2012) – As above

BUT, we still don’t know the relative impact of different components of the system

FIN

System measuresAccuracy

% overlap b/n target & object

Movement SpeedRate of movement (m/s)

EfficiencyDeviation of object from straight line path % score

Within-group Evaluation

Within-group design

Performance assessed at 3 time points:

Pre-test 1 4 weeks before VR therapy

Pre-test 2 Immediately before VR therapy

Post-test Immediately after 4 weeks of VR therapy

Participants9 patients (5 male) with severe TBI, recruited from

Epworth Hospital, Melbourne.

Age range: 23 – 49 years.

Note: Both L-R side affected, but L somewhat more in 6/10

PTA range: 28 – 630 days.

Inclusion criteria < 50 y-o Score of 2+ for muscle activity, Oxford Scale Cognitively able to understand the program, and provide consent

System measures

Accuracy

Movement Speed

Efficiency

Standardised measures

(1) Box and Block testblocks moved in 60 s

(2) McCarron Assessment of Neuromuscular Dysfunction (MAND)• Timed nuts-&-bolts task = Bimanual

Neurobehavioral Functioning Inventory (NFI)

A measure of cognitive & functional impairments in TBI. Sub-scales are:

Depression

Somatic

Memory/attention

Communication

Aggression

Motor

Adapted Presence Questionnaire Sub-scales:

• Involvement/Control - engagement & ability to exert control

• Interface Quality – how intuitive and easy to use

• Distraction – ability to “isolate” user from external environ’t.

• Sensory factors - richness of VE & multimodal info.

5-point Likert scale (1=not at all; 5=a great deal).

Results

Data analysis

Planned comparisons for 6 DVs

Pre-Test Contrast - Pre1 vs. Pre2

Pre-Post Contrast – (Pre1, Pre2) vs. Post

Subjective evaluation of the exploratory VEsAveraged over the three exploratory environments, mean

ratings were:

Involvement/Control - 3.90 (SD=0.54)

Interface Quality - 4.13 (SD=0.06)

Distraction - 4.84 (SD=0.15)

Sensory factors - 4.00 (SD=0.18)

Conclusions

VR-based therapy “value adds” to upper limb rehab for TBI

Training effects tend to be task-specific, with some functional gains, BUT

Training effects more mixed on standardised measures. Why? Bimanual performance was not targeted by our system. (See also other VR work).

Subjective evaluation of the exploratory VEs

High presence engendered by the exploratory VEs, esp. task involvement / control. Sensory stimulation was strong Engagement was high, & distraction low Sense of control

Motivational incentive for patients in creating their own feedback effects (visual and auditory).

The combination of goal-based and exploratory environments holds great promise.

Where to from here?

Paediatric VR

Re-Action Project (London) – Green, Lin, & Wilson

Soft Graspable User Interfaces

The art and science of interface designAffordance, tactility, curiosity, play

Co-located environments

The power of social facilitation and cooperation

“WOW! Where’d you get

that move

from?”

Never you mind. I’ve

been practising!

Blending motor & cognitive virtual rehabilitation

FIN