1 Augmented Feedback Chapter 4
Dec 23, 2015
1
Augmented Feedback
Chapter 4
2
Note
For those of you who have recently taken KNR 257 (motor learning and performance) you’ll note similarity between the slides. No surprise – it’s the same material.
The difference comes at the end, where more recent findings are reviewed in preparation for next week’s research readings
3
Feedback types After performance…
•Sensory feedback (Task-intrinsic)•Visual
•Proprioceptive
•Auditory
•Tactile
•Augmented feedback (Task-extrinsic)•Knowledge of results (KR): information about
the outcome
•Knowledge of performance (KP): information about the movement
4
Relative importance of feedback
Sometimes it’s essential for learning•Critical feedback needed for learning is
not “available” or not interpretable for whatever reason•Unseen target
•Disease/disability - loss of sensation
•Task-intrinsic feedback is there, but can’t be understood (timing)
5
Relative importance of feedback
Sometimes it may not be needed•Sensory feedback available, understood,
and usable
•Duplicating information that is already available•E.g. Saying “you hit it” when the person can
clearly see they did (not only redundant, but annoying)
6
Relative importance of feedback
Sometimes it may enhance learning•They can learn without it, but it speeds
up learning•Complex skills requiring new patterns of
multi-limb coordination
•Aids the search through the “perceptual-motor workspace” (directs attention, aids in cue usage and so on)
•E.g. golf shots, most sports skills
•We’ll discuss this more towards the end of this slide set and next time
7
Relative importance of feedback
It may even make things worse•Feedback after every trial (guidance
hypothesis, see later)
•Concurrent feedback (but again see later)•In both cases, the idea is that there’s an
inappropriate amount of attention paid to the augmented feedback
8
KR & KP – the lab & the “real world”
Teachers & coaches use KP almost exclusively
Motor learning research has been founded mostly on KR•Problem (external validity)?
•Maybe – needs to be borne in mind for the next few slides
•Don’t just assume it’s all junk though – it has to be argued, and argued well
9
The “small & simple” paradigm Principle 1: Feedback must be
prescriptive for folk to learn from it, so we need to study it in such situations
Principle 2: the task must be simple enough that folk can learn it in the time available, so that we can say something about learning•The “small and simple” paradigm met both these
objectives by: •Using simple tasks that only required a small
amount of practice to learn•Using tasks where feedback was essential to
learning (so feedback was prescriptive), and examining how different doses of feedback affected learning
10
The “small & simple” paradigm Feedback is prescriptive: provides
guidance towards correct performance
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7 8
0% KR50% KR100% KR
11
The “small & simple” paradigm Does this mean 100% feedback
improves learning?
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7 8
0% KR50% KR100% KR
12
The “small & simple” paradigm The guidance hypothesis…why
does this happen?
0
10
20
30
40
50
60
70
80
90
1 2 3 4 5 6 7 8 Ret
0% KR50% KR100% KR
13
Feedback “scheduling” Reducing guidance…improving
learning•Relative frequency…less is more
•Summary…100% feedback but only 10% of the time
•Self-selected frequency (tend to choose less frequent)•Choose to get feedback when accurate, and
works best when the choice is after performance (Chiviakowsky & Wulf, 2005)
•Bandwidth feedback…
14
Feedback “scheduling”0% (no) bandwidth
(100%KR)10% bandwidth (?% KR)
425ms425ms
475ms475ms
525ms525ms
575ms575ms
Target (500ms)
425ms425ms
“…“…..”..”
“…“…..”..”
575ms575ms
‘Correct’
450
550
15
Feedback “scheduling” Reducing guidance…improving
learning•Bandwidth feedback…
•Provides guidance (correction) only when necessary (big errors)
•Also tells people NOT to correct when they are reasonably accurate
16
Feedback “scheduling” Bandwidth feedback…more than
just a relative frequency effect
20
30
40
50
60
70
1 2 3 4 5 6 Ret
5% BW"Yoked"
From:
Lee and Carnahan (1990)
17
Feedback “scheduling” Bandwidth feedback…effects on
performance?•“Blocking”…similar to guidance ideas
•When you increase the bandwidth, you decrease relative frequency of feedback
•In doing so you reduce “blocking” of sensory feedback
•Maladaptive short-term corrections
18
Feedback “scheduling” Bandwidth feedback…effects on
performance?•Maladaptive short-term corrections
•Increased bandwidths reduce attempts to correct very small errors in performance
•Note influence of “no-feedback” trials
19
Feedback “scheduling” Bandwidth feedback…learning?
•Larger bandwidths (up to a point) may improve learning
Cause:•Blocking
•Reduction in MSTC.•You want the bandwidth to be sized so that it
reduces RF to an appropriate level and reduce the occurrence of MSTC
20
More recent findings
Recent research has suggested guidance ideas are not always accurate•Sometimes frequent feedback seems
desirable even for learning
•The effect of the feedback on attentional focus seems to be important
21
High frequency is good Wulf & Shea
(1999)•Concurrent
feedback
•The more often it was presented, the better people performed
•One of the papers reviewed for next week seems to offer an explanation...
-160
-140
-120
-100
-80
-60
Pret
est 1 10 11 20 21 30
TrialsR
ela
tive forc
e o
nse
t (%
)
100% FB
50% FB
Control
Perf
orm
ance
im
pro
vem
ent
22
High frequency can be good Retention data from a
soccer kicking study Pay attention to:
• Feedback frequency
• Attentional focus
• Interaction between the two
1
2
3
4
5
6
1 2
RetentionAcc
ura
cy S
core
Ext-100 Ext-33 Int-33 Int-100
23
Next week
Please read the first paper in the list•Read first for comprehension (quickly)
•Read again for critique (slower)•Q1: What do you think of the external validity of
the method of altering focus of attention to improve learning? Think of a new situation, set of ages etc, and whether the findings will change
•Q2: What do you think of the construct validity of the cues used to manipulate focus of attention? (specifically, is it just focus of attention that is manipulated across the two sets of cues?)