L5 - Motor Learning and Robot-Assisted Therapy

Question 1

What is motor learning?

Answer 1

A set of processes associated with practice, leading to a relatively permanent change in the capacity for skilled behaviour.

Question 2

What are the 5 elements that motor learning involves?

Answer 2

• instruction (explicit strategies)
• reinforcement
• error-based adaptation
• motor acuity
• use-dependent learning

Question 3

What was the issue with patient HM?

Answer 3

Lesion to the hippocampus, it was completely gone. Complete lack of short term memory.

Question 4

What did seminal studies find and show about the role of instruction, using patient HM?

Answer 4

Tested on a range of motor tasks; mirror learning, sequence learning (difficult, so typically tested over a number of days)

Even though HM had no recollection of the task, learning accumulated over the days of learning.

Therefore, motor learning is implicit. But on each day, task instructions were supplied. Later studies show that learning of motor tasks does not occur without instructions. Suggests learning is not completely implicit, and requires explicit elements (instructions).

Question 5

If people are not given any explicit instructions on a task, how good are they at performing it?

Answer 5

Bad - we don’t do well at making exploratory behaviours.

Question 6

What did Manley and Dayan (2014) find about awareness, success and variability in motor learning (exploration behaviours)?

Answer 6

PPS made reaches on a scale. Aware pps were told that there were more rewarding regions than others on the scale, unaware were not told.

Aware - exploration, finding rewarding areas/reaches.

Unaware - very little change in aiming direction throughout

Question 7

Define reinforcement based learning.

Answer 7

Actions are selected with increased or decreased frequency based on reward or punishment.

Question 8

Describe the difference between the effects of short term and long term reward on behaviour

Answer 8

Short term reward - just good enough, rewards for small changes in behaviour. Can limit the amount of exploration behaviour and therefore the amplitude of overall improvement.

Long term reward - waiting for larger changes/improvements in behaviour that are closer to optimal levels, before reward is supplied.

Question 9

What is error-based adaptation?

Answer 9

Cerebellar-dependent reduction in errors in response to a novel perturbation, through the updating of a forward model.

Question 10

What is the error augmentation approach?

Answer 10

Initial error is amplified. If PPS aim with slight clockwise error, the system changes the feedback to a huge clockwise error. Induces fast and large changes in behaviour/performance, but they are often short-lived.

Question 11

How can motor adaptation be used to improve gait in stroke patients?

Answer 11

Split belt treadmills are used to alter the speed of step required for each leg. A stroke patient learns a symmetric pattern of walking, in the same way that a neurotypical participant can learn an asymmetric pattern of walking, in the hope that their disturbed gait can be improved.

Question 12

What is the issue with using motor adaptation to improve gait in stroke patients?

Answer 12

Improvements on split belt treadmills do not generalise to normal, non-treadmill walking - both for the learning of asymmetric walking by neurotypicals, and for the learning of symmetrical walking by patients.

Question 13

What is motor acuity?

Answer 13

Reducing motor variability and improving smoothness with practice.

Pinnacle of stroke rehabilitation.

Question 14

What is use-dependent learning?

Answer 14

Movement repetition leads to future behaviour being biased by that movement’s dynamics.

Shown for movement direction, speed and reaction time.

Question 15

What is the speed of washout following use-dependent learning?

Answer 15

Very quick

Question 16

What is an example of use-dependent learning with TMS?

Answer 16

• TMS applied to the motor cortex, thumb moves in a particular direction (e.g. up and right) very consistently.
• Sham stimulation is applied, and PPS are trained to move the thumb in the opposite direction (e.g. down and left).
• TMS is applied again, he thumb now moves in the trained direction without thought.

(However, it very quickly returns to the original direction in the absence of further training-use-dependency.

Question 17

What do stroke rehabilitation methods fail to do, in the context of separable brain areas?

Answer 17

Understand that all patients will be different in the rehab methods that are best to use, depending on the area of damage that the stroke caused (i.e. no use training using instruction if the stroke was in the DLPFC).

Question 18

Which part of the brain are instructions processed in?

Answer 18

DLPFC, ACC (frontal), and Basal Ganglia

Question 19

Which part of the brain is error-based adaptation processed in?

Answer 19

Cerebellum

Question 20

Which part of the brain is reinforcement processed in?

Answer 20

Basal ganglia, SMA, maybe M1.

Question 21

Which part of the brain is responsible for motor acuity?

Answer 21

SMA and M1

Question 22

Which part of the brain is responsible for use-dependent learning?

Answer 22

Mainly M1

Question 23

Which form of learning is most important for stroke rehab?

Answer 23

According to Krakauer - Instruction and reward & failure based reinforcement (Galea agrees with this conclusion).

Question 24

What is the caveat of the most effective forms of motor learning?

Answer 24

Neural substrates for the expression of learning must be intact, i.e. the motor system must be able to execute motor commands. This also means an intact corticospinal tract.

Therefore, rehab-based learning can only function within the ‘residual performance envelope’ that the remaining nervous system is capable of.

Question 25

Can task specific training enhance spontaneous recovery in humans?

Answer 25

No

Although Krakauer believes task specific training and focusing on motor principles during spontaneous recovery can increase its effect.

(contrasting evidence says that this may have a negative effect, however. So not a straightforward answer, but on balance up to this point in the lecture series, the answer is no)

Question 26

What is the evidence that suggests early training can enhance spontaneous recovery?

Answer 26

In rats, starting re-training 5 days after stroke is much more effective than waiting 2 weeks.

Suggests that early training may speed up or increase the efficiency of spontaneous recovery processes - if there was no effect, the outcomes of training would be the same regardless of the time that re-raining began.

Question 27

What do some believe about conventional rehab during the sensitive period?

Answer 27

It is too low in dose/intensity that it fails to exploit motor learning. Leads patients to learn compensatory strategies instead of focusing on reducing impairment, just to decrease their time spent in hospital.

Question 28

What does a large amount of CIMT early after stroke lead to, in rats?

Answer 28

Increased lesion volume and reduced use of the affected limb in rats.

Therefore, perhaps we should not rehab in high doses/intensities early post-stroke.

Question 29

What does a large amount of CIMT early after stroke lead to, in humans?

Answer 29

Those reiciecing higher doses of CIMT are worse on task specific improvements (function) assessed by the ARAT and on impairments, tested by Fugl Meyer score, than controls AND those receiving low doses of CIMT.

Suggests negative impact of large doses of rehab early on post-stroke.

Question 30

What does the ARAT test for?

Answer 30

Tests task-specific improvements post-stroke. E.g. picking up cups, etc

Question 31

What does Krakaeur believe about the application of high doses of rehab, namely CIMT and robotics post stroke?

Answer 31

When high doses are applied, it is typically 6 months post stroke. The brains of stroke patients at this chronic stage are similar to that of a healthy adult, with little neuroplasticity occurring. Therefore it seems to late to be applying high doses, and could be more effective at an earlier stage.

Question 32

What motor learning principle is robot assisted therapy based on?

Answer 32

Use dependent learning

Question 33

What are the two main approaches within robotic therapy?

Answer 33

• robot guides/constrains the arm to more normal trajectories (reinforcement based shaping)
• robot applies forces that causes patients’ trajectories to have even larger errors, with the after-effect causing normal movements (adaptation based error augmentation).

Question 34

What is reinforcement based shaping?

Answer 34

One of the main approaches within robotic therapy:

robot guides/constrains the arm to more normal trajectories

Question 35

What is adaptation based error augmentation?

Answer 35

One of the main approaches within robotic therapy:

Robot applies forces that causes patients’ trajectories to have even larger errors, with the after-effect causing normal movement

Question 36

Which is the more beneficial method of robotic therapy?

Answer 36

Reinforcement based shaping

Question 37

What is the clinical efficacy of using robot-assisted therapy compared to usual care, or ICT?

Answer 37

When compared with usual care, the effectiveness of usual care actually decreases, with lower Fugl Meyer scores, while there is no significant difference in FM score caused by robotic therapy.

When compared with ICT, intensive comparison therapy (physio), you do get an improvement, but only 2-4 points on FM, which is not enough to feel benefits, or classify as clinical improvement (6-7 points).

Question 38

What is the hypothesis of methods used in Rosenthal’s (2019) study using robots to guide or assist task movements?

Answer 38

Better recovery of motor function will occur when training is with movements that have an intermediate difficulty for the patient - at 10% steepest gradient (SG) - the movements of steep transition in performance capability.

Question 39

What did Rosenthal (2019) find while using steepest gradient methods to guide or assist task movements with robots?

Answer 39

At baseline, there were no differences between the control and SG (steepest gradient) groups, both having mild to moderate impairments on the FM.

After 5 days you get improvements in the control and SG group, but the improvement in the SG group is 50% greater, suggesting that individualisation of robot assisted therapy and make it relative to each individual’s impairment, you get a greater amount of improvement on the FM.

Question 40

What are the caveats of Rosenthal’s (2019) study on training with movements that correspond to regimes of high performance variation?

Answer 40

Improvements are small - 50% improvements were not a clinically significant amount in terms of movement on FM scores.

No difference after 3 months, so improvements are not long-lasting.

Question 41

What is the general consensus on robot-assisted therapy, and what is this based on?

Answer 41

It works, but only as well as dose-similar conventional therapies.

Based on a meta analysis of the 67 robotic studies between 1997 and 2011, finding very small FM score changes overall (2).

Most studies were on chronic stroke patients. Only 5 were based on the first 3 months post-stroke, and only 1 showed a FM change greater than 5.

Question 42

What are the summary points?

Answer 42

• rehab can only operate within the residual performance envelope that the nervous system is capable of.
• reinforcement-based tasks (shaping) appear more long-lasting and more widely used.
• most motor learning based rehab has been applied in chronic stage, with relatively low-dose/intensities with little clinical impact.
• rehab during chronic phase is thought not to be able to improve impairment, but only improve function via compensation.
• high dose/intensity motor learning rehab could be more effective during the first 3 months when spontaneous recovery is still present.
• but some work suggests that starting too early can have negative effects on motor impairments, so perhaps the best way is to start early, but slowly ramp up the dose throughout the acute stage.

Question 43

Key reading - essential: Chapter 7, Oxford Textbook of Neurorehabilitation (Dietz & Ward)

Answer 43

Ok boss