Providing Feedback During the Learning Experience
- Feedback is essential for learning.
- Individualized feedback is also important.
- In a clinical setting typically one on one vs. classroom setting has a high
teacher to learner ratio.
o Must make sure that you provide enough feedback to ensure
TRANSFER (i.e learning).
o Help them just enough
- To know definitions associated with inherent and augmented feedback
o Ultimate goal is for people to rely on inheret
- To understand the effect of augmented feedback on performance vs. learning
- To understand the difference between knowledge or performance (KP) and
knowledge of result (KR)
- To understand the effect of KP and KR manipulations on performance and
learning, and appropriately apply them in practical settings.
- Augmented Feedback: Any information that “supplements the information
that is naturally available.” – Swinnen, 1996
- Inherent Feedback: Information naturally available
Ex. lateral displacement while walking on a beam might be a relevant cue that is not
Which of the following sources of feedback cannot be augmented?
- Muscle Spindles
- NONE OF THE ABOVE
ALL of the sources of feedback can be augmented!
Ex. GTO- can get a better understanding with force plates or force transducers.
Ex. Vestibular- Can feed information from accelerometer on to the tongue. Tactile
information can be provided about a lost vestibular input.; also can provide some
guidance by manually manipulating a body over a body in gymnastics, for example.
Ex. Muscle Spindle- Can get a better understand bout length of muscle with
Ex. Visual: Glasses; Auditory: Hearing aids!
Augment Within or Between Senses
- Video camera is important quick replays provide a better representation
of where you are in space.
o Use different viewing angles - Binoculars
- Hearing aids (can pick up auditory information more accurately b/c hearing
aids have a high temporal velocity)
- Vibration device for posture
o But how do you provide enough feedback to make sure that when you
take tools away, they are still good! Easy to improve performance,
hard to learn.
Do not exploit them/rely on them…
- Concurrent Feedback: During performance
- Terminal Feedback: After performance
o Terminal feedback can either be immediate or delayed
o Immediate: Right after the movement
o Delayed: with a time delay after the end of the movement
- Distinct: based on a specific trial
- Accumulated: based on a number of trials not always possible.
- Knowledge of result (KR): Information pertaining to the outcome of the
o Ex. Someone falls on head
We are still inclined to give very obvious information to the
learner “Hey, you fell on your head!”
- Knowledge of Performance (KP): Information pertaining to the execution of
o People often disagree with what is the right technique.
o Needs to be taken with a grain of salt unless there is a very solid
understanding of biomechanics (proper technique)
Which of the following augmented FB types is impossible?
- Inherent Concurrent
- Concurrent knowledge of result (don’t know the result during the
movement. Sometimes you do though)
- Immediate terminal
o It’s not a good idea though!
- Accumulated knowledge of performance (ex. focus on a few items that are
- All of the above are possible
The effects of augmented feedback- learning
- Often gives the illusion that learning is taking place! **
- Bilodeau, Bilodeau, and Schumsky (1959)
o Linear position task Can’t see limbs 9 and 14 mm away from
o KR person improves. Isn’t the person learning? o Quantify mean absolute error
o What results looks like: NO KR Don’t learn; KR Learn
o But they are missing a retention test.
o They focused on the No-KR and the KR group for 19 trials
o Compare first 5 trials WITH KR for the No-KR and KR-19 trial groups
Knowledge of Performance
- In practice
o Kinematic FB (Position, velocity, acceleration)
o Biofeedback (EMG)
o Kinetic (Force)
Sometimes you need to explain the information that you gather via these
methods to the learner.
- Video Replay
o Can learning be optimized with video reply?
o Correcting cues.
o Graph Subjective form rating
By giving correcting cues, people are performing better.
Could they become dependent on corrective cues? (Or
Knowledge of Result
- Line tracing task: Effect of KR withdrawal in transfer
- Nonsense LKR: “good job, keep going” error was large in acquisition and
- Right/Wrong: Improve, and is quite consistent in transfer.
- Precise KR: good performance, more error in transfer, but still better than
- Bad feedback is simply bad. Good feedback is good, but you must expect a
decrement in performance in the transfer test.
- Bandwidth KR: Tolerance limits on errors that define when to provide
qualitative or Quantitative KR
o Provide KR only when outside an error bandwidth
Only talk to the learner when they are making a mistake
Is no KR a form of KR? Yes. They know they did it right.
Which of the following sports uses bandwidth KR as a form of augmented FB?
- Not a source of augmented feedback because you know if you got it in on your own.
- As a batter, you have an idea of whether it is in the strike zone, but
augmented feedback helps.
- Strike zone is the bandwidth Football
Knowledge of Result
- Can vary bandwidth (0% margin of error, etc.)
- First acquisition block, 0% and 5% BW groups do more errors than 10% BW
- People with 10% bandwidth start off better. They are performing better on
acquisition and retention.
- Amount of feedback given to people with smaller bandwidth is bigger!
Possible explanation of this?
- Neuromotor noise or variable error
- Example: Red line
- Constant error: difference between target distance and distance achieved
- Variable error as well! IT is normal. It is very hard to minimize variable error.
If use a bandwidth of 10% of amplitude: Don’t say anything to learner if they
are in the red box (representing 10% bandwidth)
o Let them know if they are undershooting or going to far.
- If a learner always undershoots, and then overshoots on ONE trial. Do not say
anything on the overshoot
o Must identify a good error! Don’t say anything.
o Reason why giving precise KR all the time is bad. It is likely to induce
neuromotor noise (there is already a level based on variable error).
- Must identify a good error!! Don’t say anything
Problem with Bandwidth Error
- Can’t provide Bandwidth KR in a classroom setting
- Some people need different bandwidths based on skill.
- Learner-determined KR: “KR provided after a trial, only if the learner is
asking for it”
o Why provide this?
They can start to figure out what works for them.
o Increase learning compared to expert determined KR: even with the
same KR frequency.
There is lots of literature on this, but how can we implement
this? It is not easily deliverable.
Will kids be able to know when they need KR?
Erroneous KR: Knowledge of Result containing a bias/error
- If it is within your variable error range, wrong KR should INDUCE error.
- Ex. There is a delay/bias when a light comes on and it when rally comes on.
o Buekers and Magill (1995): +100ms bias on the anticipation task if
the person is 100ms early, then KR was considered a 0ms error.
If you are biased in your KR will transfer bias to learner. o Effect is stronger if erroneous Kr is presented on every trial!