KNPE 261 weeks 5-8 Flashcards
A system is open loop when it
doe snot take feedback into account
efference copy
copy of motor command that allows for the prediction of the action outcome and the sensory consequences
Why cant we tickle ourselves
we predict sensory consequence via efference copy
invariant features GMPs contain
-relative timing
-relative force
-relative sequence of events
parapeters GMPs accept
-overall duration
-overall force
-effector used
How does concept of GMP address storage and novelty problems?
storage: reduce amount of programs by half… general motor program but adjust/specify for certain parameters
Novelty: movement is not truly new, you adapt other programs to complete a “new” movement
relationship between speed and accuracy
- fast movements are less accurate
-accurate movements are slower
Fitts’ Law Experiment details
-tap b/w 2 targets as rapidly as possible maintaining 95% accuracy
Main outpus in fitts law
movement time
variables in fitt’s law
movement amplitude and target width
Fitts Paradigm
Index of difficulty= log (2(amplitude)/width)
when you increase ID, you increase movement time
Fitts Equation
movement time = a + b x ID
a= y int
b= slope
Units of ID
bits of information
if movement amplitude is doubled, what happend to ID
increases by 1 (harder)
if width is doubled what happens to ID
decreases by 1 (easier)
How would you get an ID of 0
no amplitude; the 2 targets are touching
Graphing FItt’s Data
+ve linear data
can you have the same ID with different amplitudes and widths
yes
FItts Law holds for
-children
-lower limb movements
-under magnification
-imagined movements
-percieved movements
fitts law is based on
visual feedback
open loop movements examined by Schmidt et al
-participants performed movements to thin target line
-width did not change but time required to make movement did
goal of open loop movements experiment by schmidt et al
determine “spread” about target as a function of D and T
schmidt formula for open loop movements
effective width = a+b (D/T)
Schmidts Law
as time increases, effective width increases
as speed increases, effective width increases
Breaking Fitts Law violation
when contextual target cues were present
-last target faster when all possibilities were shown; preparing for worst case scenario
where do violations of fitts law occur
all last targets when all posibilities shown
what did glazebrook et al find
fiits movement violations were not planning for worst case scenario but controlling via feedback, efficent movement corrections as seen tat most change happens in the middle not the beginning, indicating corrections not planning
glazebrook et al experiment
-performed movements to target location (first, middle or last in an array)
-measured movement variability across trajectory as an indicator of planning (differences earlier would indicate planning, differences later indicate control)
-differences emerged later meaning violation based on movement correction NOT planning
Do we plan for worst case scenario
no, we adapt efficently
what causes speed-accuracy tradeoffs?
Impulse variability theory
Impulse Vriability theory
-the variability in the duration of a group of contractions is related to the mean duration
-the variability in force produced increases as a function of the force produced
Implulse variability and speed accuracy tradeoffs
variability increases to about 65% but then decreases at higher levels of force output
How do we test impulse variability and speed accuracy tradeoffs
aim to target but add resistance OR tell them to do it in less time to increase force
Schmidt and Sherwood
-applied load
-when load was greater then 60% of participant max, variability decreased
-least variable at fastest
movement time in a reciprocal movement in linearly related to
movement amplitude and target width
Recap violaition of fitts law
movements to last target is faster than movements to second last target
definition of motor learning
set of processes associated with practice or experience that leads to a relatively permanent change in capability for movement
what is a process
set of events that lead to a product or state of change
what process are we interested in in motor learning
retrieving a motor program from memory
motor learning processes are
assumed; we think some events have occured for there to b learning as a result of practice
Practice
purposeful repetition of a skill or behaviour
Experience
the fact or state of having been affected by or gained knowledge through direct observation or participation
Motor Learning is
relatively permanent
define relatively permanent
change of state is not readily reversible
**any change that is readily reversible is not related to learning
How do you know if you have learned something
some underlying change that is stable, learning should have some lasting effect
Product of learning
ability to move skillfully in a particular situation
goal of motor learning
strengthen the quality of internal state such that the capability of the skill will be altered
goal as researcher of motor learning
understand nature of these internal states
Capability for movement
-stresses the role of the internal states that leads to the skilled behaviour
-factors can affect movement outcome but have little impact on internal states
Learning
-not directly observable
-highly complex phenomena
-we often have to infer changes based off behaviour
complexity of assessing neural basis of learning
difficult if changes in neural and physiological activity is correlated with learning or changes in behaviour
Donald and Hebbs, Hebbian processes
neurons that fire together, wire together
Neural networks (geoffrey Hinton)
-most neurons recieve input from other neurons
-these inputs are weighted
-neurons can adapt weights (how much they contribute)
Activation in networks can be observed by looking at
-Outputs: electrical activity
-Energy consumption - bloodflow
Functional connectivity analysis
-Examine the time-series of FMRI data (change sin BF) in different brain regions
-correlate the time-series between different regions of interest
-examine strength of relationships
McGregor and Gribble Functional connectivity and motor learning
have shown functional connectivity can predict motor learning
McGregor and Gribble 2017
-measured participants connectivity prior to observational learning protocol
-found participants who had higher levels of connectivity in sensorimotor regions perform better in learning
Adaptation
the iterative process of adjusting ones movement to new demands
Motor adaptation
the trial to trial modification based on error feedback
characteristics of motor adaptation
-movement retains identity
-change occurs with repetition or practice and is gradual over minutes
-person must de-adapt over behaviour
how do you know if someone has de-adapted
they show an aftereffect
Forcefield adaptation paradigm
-initial increase in error when exposed to forcefield
-adapt to forcefield and plateau
-in post adaptation phase, error in opposite direction
Prism Adaptation
adapt to visual distortion then deadapt and show a little error but not as much error as you did before adapted
measuring motor learning
Typical experiment: participant exposed to task (acquisition), performance plotted as a function of trials, examine consistency
4 types of performance curves
- Linear curve
- Negatively accelerated curve
- positively accelerated curve
- Sigmoid curve
common curves for motor skill learning
positively accelerated and sigmoid
Factors affecting performance
-between participants variability
-within-person variability
-ceiling effects
-floor effects
Between participants variability
-performance curves usually represent grouped data so individual differences get washed out
within-person variability
-performance of individual varies trial to trial
-average curve may not do a good job of catching individual variations
ceiling effects
limits at top of scale
floor effects
limits at bottom of scale
ceiling and floor effects
changes in performance levels become intensive (harder to quantify) to changes in learning
ex.
-it is easier to improve score at mid-level
-reducing scores in golf is easier when strokes are high
performance curve considerations
how acquisition is measured can affect the interpretation of performance curve (ex. must be within 30% vs 5% of target diameter to be “ on target”)
How would we want to compare virtual reality (new method) training to physical practice (old method) in learning a surgical skill?
have 2 groups, one exposed to new method and one exposed to old
how do we separate the transient effects seen in acquisiton from that of “real learning”
use 24 hour retention interval
retnention tests
testing particiant on the same task after a time interval
*longer retention interval, the more transient effects are reduced
Transfer tests
involve new variations of the practiced task
-new speed/conditions
-similar but not same task
how to examine results of retention/transfer tests
look at if retention and transfer has improved or declined compared to acquisition, as retention and transfer indicates learning the most
Learning with robotic guidance
robot increased performance in acquisition but then when taken away, it showed decreased learning
how can we modify exams to make them truly assess learning?
1) longer time between being introduced to content and taking exam
2) Application questions, not reciting
Motor adaptation leads to
transient effects on movement capability (observed during acquisition)
Best way to evaluate learning
retention and transfer
what type of change is learning
a relatively permanent change
what type of change is adaptation
transient
What variable does not consider target
variable error
Dorsal streams mainly process
action
Ventral Streams mainly process
perception
Golgi tendon organs sense muscle
tension
Muscle spindles sense muscle
stretch
3 issues with motor programming theory
-storage problem
-degrees of freedom
-novelty problem
what mechanism can sensory information be used for
feedback
levels of open loop control
executive and effector
executive sends motor program to the effectors and effector carries out the instructions without modification
Example open loop feedback
Oven, stays at the same temperature whether the food is burnt or not
learning is what type of change
relatively permanent
adaptation is what type of change
transient
what two types of tests can learning be evaluated using
retention and transfer
what is a possible issue with examing a new vs old method and how do we address it?
testing environment is same for one group but different for another - we can do double transfer test
double transfer design
in acquisition, both groups do 1 form of skill, in retention and transfer both groups do both forms of skill and assess
performance variables
influence performance in transient ways
(effect of variable disappears when conditions altered)
-tools that affect performance
Learning Variables
influences performance in relatively permanent ways
(effect of variable stays when conditions are altered)
-tools that lead to permanent changes
ceiling effect
for well practiced skills, it is hard to assess if additional practice will result in improvements
learning curves indicate
skills can be learned using different patterns
Overlearning
the process of having a person or continue to practice after they have reached performance ceiling
How do we measure if any more performance effects are happening during the overlearning trials?
compute savings score
Savings score
have many trials the overlearning group saved in reaching their ceiling
ex. participant A took 5 less trials to reach ceiling than participant B
other ways to assess learning
-performance on secondary task
-measuring indices of effort
-measuring response latency
-generalizability of learning
performance on secondary task
gives idea of how much attention is needed to perform a task - well learned tasks require less attention
measuring indices of effort
physio or psychological markers - higher heartrate for activity indicates more difficult for participant
measuring response latency
peed of correct response or movement performance
generalizability of learning
varying parameters of task
what about learning curves quantifies rate of learning
steepness
problems with learning curves
change in acquisition does not show learning therefore only assess learning in retention and transfer
ways practice can be distributes
massed vs distributed
Massed practice
practicing with very little rest between trials
Distributed practice
practicing with longer rest periods between trials
research on practice distributions are usually conducted using
continuous tasks (pursuit-tracking) task
conclusion on optimizing practice conditions
shorter practice durations are less effective than longer ones
what is a potential confounding factor of for having long practice periods
fatigue
Fatigue
-could play a role in tasks such as pursuit-rotor task
Bourne and Archer 1956
found shorter resting times between trials resulted in less learning
Ammons 1950
-used pursuit rotor task and found groups with more rest performed better
Distributed practice and learning 2 key findings (Ammons)
- Only small differences remained after the last transfer trials
- However, differences in performance re-emerged on the 24 hour transfer trial
Baddeley and Longman
looked at typing skills in postal workers and found that more distributed practice led to better retention when practice hours were constant
Distributed vs Massed Practice Anderson 2015
-explored learning in VR
-participants learned more with distributed practice
Distributed and Massed Practice tradeoff
massed: reduced benefits per time training, but takes less total time
distributed: most learning per time training but takes more time
how can we make practice more effective and efficent?
if we treat time as a limited resource
constant practice
one variation of a task per practice session
variable practice
many variations of a task within a practice session
what skills does variable practice produce more learning
open and closed
Shea and Kohl 1990
-examined whether participants trained under constant or variable practice performed better in retention
-used force production tasks where participants either produced constant or variable force
-variable practice outperformed constant practice group
McKraken and Stelmach, 1977
-investigated learning of obstacle hit task
-participants had to knock barrier in goal time
-used either constant or variable practice
Variable practice in transfer
performed better than constant
variable practice increases the
generalizability of learning
Schema Theory
-with practice, people develop rules (schemas) about their own motor behaviour
-contains relationship between environmental inputs and outputs
-rules maintained in memory and can be used in new parameters
variable practice and schema theory
variable practice produces a schema for selecting the parameters of the generalized motor program
when is variable practice more powerful
-childrens motor learning
-learning absolute timing of a task (relative is better with constant practice)
WHy is learning hard to quantify
variables like criteria used to assess people and we have no measure of learning potential
Blocked practice
practicing the same tasks repeatedly in a predictable manner: AAA BBB CCC
Random Practice
practicing multiple tasks in an unpredictable manner: BCA ACB CBA
when is constant practice better than variable practice
acquisition & when the cost of varying practice is too high
can learning occur during plateaus
yes - and overlerarning results in savings
when is practice most effective
when distributed over time
functional connectivity in the brain
-timeseries correlation between BOLD signals (oxygenated to deoxygenated blood) in different regions of the brain
high coorelation is interpreted as
greater contectivity (signals vary in similar ways over time)
low correlation interpreted as
low connectivity
McGreggor and Gribble
-found that participatns with greater connectivity were better at observational learning
results of blocked and random practice experiment
random performed worse in acquisition but better in retention and transfer
Contextual inference
inference that is generated due to context associated with skilled practice
*reduces performance in acquisition but provides advantage in learning
how do we generate high contextual inference
randomizing order in which tasks are performaed
when comparing constant and variable practice:
performance during acquisition is better for constant and performance during learning is better for variable
when comparing blocked and random practice
performance during acquisition if better for blocked practice and performance during learning is better for random practice
performance learning paradox
variables that enhance performance were thought to be good for learning (this is not the case)
example when blocked practice leads to more learning
volleyball serves
learner influences of contextual inference
-magnitude may differ with age or expertise
downfall of random practice
can be discouraging due to psychological impact
difference between random and variable practice
random practice is changing the task (dont know whats coming next) variable practice is changing the parameters (doing the same task different ways)
Hall et al
examined effect of blocked and random practice in NCAA athletes
-pitchers performed blocked and random practice and then were tested in both a blocked and random test
found: random facilited transfer in blocked and random testing, and even skilled athletes can benefit from random practice
Elaborative Processing Hypothesis
rnadom practice promotes more comparitive and contrastive analyses of actions required to complete the tasks to previous tasks
Forgetting and Reconstruction hypothesis
teh processes associated with the reconstruction of teh motor program are trained to a greater extent by random practice than blocked practice
when you perform a task you draw motor program from long term memory and action is constructed to stay in working memory. If the same program is sued for subsequent trial it stays in working memory. reconstructing motor program involved drawing it from long term memory
Cuddy and Jacoby
-examined recall words in a list
found: words repeated with lots of words between are recalled more effectively than words repeated with no/short lags *this is spacing effect
spacing effect
as word spacing increases, so does the ability to recall
contextual inference: forgetting and reconstruction
-any activity between trials that causes short-term forgetting should promote learning
-any activity that limits reconstruction should be detrimental to performance
Lee et al Spacing effect
-prticipants performed blocked, random and random practice with a model
-model was a detrement to learning, memory aids hinder learning
-shows the struggle is important in reconstruction stage
what does giving a learner hints do?
short term benefit, but long term is detremental
Neural basis of contextual inference
random practice decrease in error in 2 brain regions
Pauwels et al 2018
-random practice led to better performance than blocked
-random practice larger decreases is default mode network activation
Serial practice
combines predictableness of blocked and non-repetitiveness of random
results of serial practice
no different than random - indicating non-repetitive part of practice is IMPORTANT