practice amount, specificity, variable, constant... 1
TRANSCRIPT
1
Practice
Amount, specificity, variable, constant...
2
Performance & learning
• What is learning, really?– Performance is observed, learning inferred– Performance can improve without improved
learning– Learning can improve without improved
performance
3
Amount of practice
• Do we become less dependent on the environment, or more?– Important implications – should you practice
powerlifting in front of a mirror to aid form? (Proteau & Temblay, 1998)
– Motor program theories could suggest you should• Repeat, repeat, repeat, and the process becomes
increasingly independent of sensation (“you could do it with your eyes closed” - http://www.youtube.com/watch?v=Yshe4BcN_Mg).
– Proteau and colleagues thought otherwise...
4
Amount of practice
• As we learn, do we rely less on feedback?• Proteau’s task (1987, 1992)...
5
Amount of practice
• As we learn, do we rely less on feedback?• Proteau’s typical paradigm...
– Task: 90cm movement in 550msec– Condition 1: 200 trials with vision– Condition 2: 2000 trials with vision– Test condition: No vision
Has also used walking (see next slide), force control, and others
Sometimes as little as 20 vs. 200 trials too.
6
Amount of practice
• As we learn, do we rely less on feedback?• Typical Results:
The full vision practice condition typically transfers to a no vision condition badly, and this gets worse as full vision practice increases
7
Amount of practice
• As we learn, we seem to rely more on the information that is present and used when we learn– For the powerlifting form example – mirrors not a
good idea (Proteau & Tremblay, 1998)– Also think of learning to type, drive (shifting gear),
play piano (watching fingers) and so on– “learning is specific to the source or sources of
afferent information that are more likely to ensure optimal performance”
8
Amount of practice
• More recent findings:– Weak vs. strong visual cues
(still a reaching task) – weak vision transfers as well as no vision to a no vision condition
– Weak vision encourages processing of other sources of information like proprioception
9
Variability of practice
• Imagine you’re trying to teach catching– Should you make it as simple as possible, by
choosing only one type of ball, one type of throw, one catching technique…etc…
– Or not?
10
Variability of practice
• Schema Theory (Schmidt, 1975)– More variability means more generalized schema
for learning– Like a regression rule– Your performance of the right movement depends
on the proximity of previous behavior to the desired behavior
11
Variability of practice
• Supported?– Generally, I’d say so, provided key assumptions are
met• Are the participants genuinely novices?• Is sufficient practice given to form a strong enough
prediction rule?• Is prediction of a novel version of the task ultimately
required?• See Schmidt and Shapiro (1982) for a summary, and
Schmidt & Lee’s texts for more recent summaries.– Does not imply that the governing theory is accepted
• Now as for organization of variability...
12
Contextual interference
Practice order (3 tasks – A, B, and C)
Blocked
All A’s
…then all B’s
…then all C’s
All A’s
…then all B’s
…then all C’s
Serial
A
CB
Random
Who knows – it’s
random!
Amount of contextual interference
Low High
13
Contextual interference
• Practice order (3 tasks – A, B, and C)
Stimulus light goes off Color signifies which
movement pattern to perform
Pick up tennis ball Knock down barriers Replace tennis ball RT and MT measured
14
Contextual Interference effect
• From the classic study (Shea & Morgan, 1979)
• Practice – Low CI is better (time is being measured, so smaller scores are better)
• Retention – High CI is better
15
Contextual Interference
• Theory– 2 primary hypotheses
• Elaboration– Compare the sequence of tasks practiced within blocked and
random practice – what kinds of comparisons between or among the tasks are promoted by each type of practice?
– “inter-task” versus “intra-task” processing.
16
Contextual Interference
• Theory– 2 primary hypotheses
• Action plan reconstruction– Compare the sequence of tasks practiced within blocked and
random practice – how long, on average, do you have to wait before the task is repeated in each practice order?
– Brown-Peterson (1958), Peterson-Peterson (1960)
Recall worsens as interval “A” increasesA
A
Recall improves (!) as interval “A” increases
17
Contextual Interference
• Which hypothesis is best supported?
– Please note I’m not saying this is proof of one theory’s predominance – it’s too complicated for that (think external validity!) – but it is interesting evidence in this instance.
18
Contextual Interference
• Predictions, and task
19
Contextual Interference
• A typical trial pattern (showing when TMS is applied)
20
Contextual Interference
• Which hypothesis is best supported?Blocked practice
groups unaffected by TMS
Random practice groups affected
by TMS
Blocked Groups Random Groups
21
Contextual Interference
• Now for something completely different
We’ll see that these findings may severely limit the generalization of the CI effect
22
Contextual Interference
• Task:
– Notice: overall duration varies across tasks; relative timing does not
23
Contextual Interference
• Task:– With this task, you can
vary overall duration without varying rhythm
• see previous slide– Or both– Or vice versa
• E.g. – 300-200-400– 400-300-200– and 200-300-400
24
Contextual Interference
• Findings– Experiment 1:
• The more consistent the practice type, the better people perform in retention and transfer
• Here, ratio feedback was provided for all groups (i.e. how well did the person do in performing the required rhythm?)
25
Contextual Interference
• Experiment 2:– Feedback type has a radical effect
on this outcome– Hard to grasp, but depending on
feedback, effect is almost reversed– Generally, whatever results in
stability of RT during practice works (random practice with segment feedback did this, & so did blocked practice w/ratio feedback)
– Implication is that random practice is not good for learning tasks that require new relative timing patterns...
– ...but it is good if only absolute timing is required to change.
26
Contextual Interference
• Applied work (e.g. – there’s lots more)...http://www.youtube.com/watch?v=CIG3El76ltg&feature=related A task with
“switched” relative timing
27
Contextual Interference
• Smith & Davies (1995)– Used a Pawlata roll
• Compared progressive part learning of a full roll with either alternate (high CI) or blocked (low CI) practice
• All transferred to both a full and a half roll one week later (score is 5 - average # attempts prior to success)
28
Contextual Interference
• Since then...– Still celebrated as a general effect (in some places)
• Does not seem to be the case• Shea’s (& colleagues) work clearly important• Findings largely limited to overall timing (simple
adaptations of already known movements)• Exceptions?
– Smith & Davies (1995, see also Smith, 2002, Smith et al, 2003) may be a result of negative transfer rather than CI (though this certainly matters too).
– See Barreiros, Figueiredo, & Godinho (2007) for a review of applied work. They say successful applications are somewhat rare (c. 40%)
• Subsequent work emphasizes the disconnect between simple and complex tasks (see next slide)
29
Contextual Interference
• Complexity as a moderator (for CI & others)
A good review paper for the final
30
Contextual Interference
• Different neural substrates responding to different practice structures?
TMS again, used this time to disrupt
particular brain sites (published in 2010)
31
Contextual Interference – concluding comments
• So what does all that mean?– There do seem to be fundamental differences in
the brain’s reaction to the different practice types– These differences seem to be associated with
different memory activities– Could be that random practice enhances recall-
retrieval practice, while constant (or blocked) practice does not enhance recall-retrieval, but does a better job of allowing people to learn new movement patterns.
32
Part vs. Whole practice
• Segmentation, fractionation, simplification, component interdependence...– Do the parts fit together naturally, or can they be
easily separated?• Think of a free throw – should you practice the knee
movement and the arm movement separately?• Juggling...from the annals of 257 (Spring 2000) – Knapp
& Dixon (1952) revisited.
33
Part vs. Whole practice
Sections broken down by 3 mastery levels
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
1 2 3 4 5 6 7 8 9 10
Blocks of 40 attempts
Av
era
ge
# c
atc
he
s 11am Beginners
11am Intermediates
11am Masters
12.35pm Beginners
12.35pm Intermediates
12.35pm Masters
11am class: move through practice stages quickly (get to the full juggling phase as soon as possible)
12:35pm class: practice each stage thoroughly (master each stage before moving on)
Similar findings have been published by Knapp & Dixon, 1952.
34
Part vs. Whole practice
• In this case, part practice of juggling didn’t work well
• Seems that the skill is highly organized, and as such should not be practiced in parts– Task complexity and organization (Naylor and
Briggs, 1963)• More recent research...
35
Part vs. Whole practice
• Part/whole practice for Polyrhythms
Unimanual works well regardless of
training type
Polyrhythm does not benefit from
part practice
36
Our readings this week...
37
Blandin, Toussaint & Shea, 2008
• Guidance effect & specificity of practice• Expt. 1:
No vision of arm
“simple” one-joint movement
Goal pattern
P = proprioception onlyPV = prop + vision33% = 33% feedback frequency100% = 100% fdbk frequency
38
Blandin, Toussaint & Shea, 2008
• Guidance effect & specificity of practice• Expt. 1:
Slight tendency for 100% frequency to
have less error during learning
Both PV conditions fail to perform in non-vision conditions – effect
greater with 33% than 100% - reasons?
P = proprioception onlyPV = prop + vision33% = 33% feedback frequency100% = 100% fdbk frequency
N.Sig.
Sig.
198 trials
39
Blandin, Toussaint & Shea, 2008
• Guidance effect & specificity of practice• Expt. 2: PV only, # trials varied
Difference only exists for PV groups after 396 trials, not after 54
– and was in same direction as after 198 trials.
PV33% = 33% frequencyPV100% = 100% frequencyBoth either 54 or 396 trials
Sig.
54 trials 396 trials
40
Porter & Magill (2010)
• Gradually increased levels of CI.• Expt. 1: three putting tasks (vary by force only)
Note “increasing” group seems better than blocked early on – yet supposed to be more difficult?? (Note only sig effect
was both gps better than random)
Note no “standard” CI
effect
41
Porter & Magill (2010)
• Expt. 2: 3 basketball passes (vary by coordination)
Only vertical error
consideredDistance
constant – 5m
Do these tasks require the acquisition of new forms of coordination?
42
Porter & Magill (2010)
• Expt. 2: 3 basketball passes (vary by coordination)
Practice: No significant
effects
Retention & transfer: Increasing < Random <
Blocked
43
Porter & Magill (2010)
• Overall summary: Small variations (force only): increasing better than both
random and blocked Large variations (change in movement pattern): increasing
better than both random and blocked; random better than blocked
Is this to do with motor programs? Is CI leading to better learning of motor programs, or new forms of coordination?