Transcript lecture9

Describe your results in the ball-catching lab.
Describe the relation between the first and second labs.
The second lab demonstrated (a) that the feedback delay can be 200 msec
or more and (b) when intercepting a rapidly moving object, this delay is too
great for accurate targeting. Therefore the subject must learn to predict
where the target will be. That is, the subject must switch from a feedback
mode to a feedforward mode.
In the first lab, we see evidence of prediction (feedforward) when the eye
saccades to the bounce point ahead of the ball. If the eye had followed the
ball (using feedback) it would take too long to relocate it after the bounce.
How does the speed of a movement (e.g. reaching) affect
its accuracy? Explain why.
Role of Visual Feedback
Question: why does error increase with speed?
Note: 50 cm/sec = 5cm/100msec
Describe the effect on reaching movements of large-fiber
sensory neuropathy (degeneration of the afferent fibers
from the muscles and skin) when visual feedback is
eliminated? Sketch the movements to illustrate your answer.
Consequences of loss of feedback on reaching
Large fibre sensory
neuropathy leads
to loss of proprioceptive
feedback
from muscles
Loss of vision leads to
moderate errors.
Errors in direction,distance
Normal:
proprioception
only
No vision or proprioception
Vision compensates
for lack of proprioception
Loss of proprioception and vision causes large errors. Vision of hand at start can
reduce the size of the errors. Vision alone can almost completely compensate for
loss of proprioception.
How can examples from robotics help us understand the
human visuo-motor system? Give examples from
videos.
Robotics helps understand what theoretical problems
need to be solved, and the consequences of particular
solutions to those problems.
Eg flipping an egg:
Grasping: need location information. How is it found?
How is grasp formed? Passive compliance helps solves
the problem. Seems likely humans solve it that way too.
Move to pan. Need to know where pan is. How is this
found? Note that telling the program the location is not a
good general solution because someone might move the
pan.
Etc
Could solve this problem using visual feedback, but
feedback is slow, as shown when Pook guided the robot
arm with her own arm. Feedback delays made it very
inefficient.
What is measured by the standard deviation and
the standard error of the mean. How are they related?
The standard deviation is a measure of the spread or variability in
a population or set of measurements.
The standard error of the mean is a measure of the variability of
the mean of a set of measurements.
The standard error of the mean is equal to the standard deviation
divided by the square root of N, where N is the number of
measurements in the sample.
Describe the visual capabilities of Mike May after his sight
was restored. What are the implications of this?
Describe the visual capabilities of Mike May after his sight
was restored. What are the implications of this?
Lost vision at age 3 - scarred corneas. Restored at age 40;
Poor acuity.
Answer to Molyneux’s question: Mike May couldn’t tell difference between sphere and
cube. Improved, but does it logically rather than perceptually. (cf other cases)
Color and motion sensitivity good.
Cannot recognize faces. (eyes, movement of mouth distracting)
Can’t perceive distance very well.
Can’t recognize perspective.
No size constancy or lightness constancy/ segmentation of scene into objects, shadows
difficult.
Vision most useful for catching balls (inconsistent with Held & Hein??) and finding things
if he drops them.
Note: fMRI shows no activity in Infero-temporal cortex (corresponding to pattern
recognition) but there is activity in MT, MST (motion areas) and V4 (color). Other parts of
brain take over when a cortical area is inactive.
MT/MST (motion)
V4 (color)
Infero-temporal cortex
Implications?
Basic object perception (recognition and segmentation) requires
experience. (Experience prior to 3 yrs not enough.)
Geometric cues about scene structure (perspective, distance) also
require experience.
Color and motion more robust - either present at birth, or acquired
before 3yrs, and preserved without continued experience.
Describe sequence of eye movements in an everyday task.
Give reasons.
Making breakfast:
Upon entering the kitchen: saccade to the cupboard on the
basis of memory, as I know cereal is located there.
Approach cupboard and saccade to door handle to guide
hand to open door. Search for cereal with several saccades,
maybe landing on boxes of similar size and appearance.
When saccade lands on the correct box, stay fixating to
guide the grasp of the box. Rotate body and head to exit the
cupboard and make a big saccade to the cupboard
containing the bowls. Fixate the cupboard while I walk there
and make a fixation to the handle to guide opening….
Goal of Lab 3:
Can we demonstrate adaptation to new sensory-motor
relationships?
How fast is the adaptation?
Are some relationships easier to learn than others?
Method
Virtual environment: head mounted display with virtual racquetball.
PhaseSpace monitors hand and head position. Auditory cue when S hits ball.
Environment obeys normal dynamics. (ie laws of physics/ bounces etc)
Visual scene translated by 0.5 m or compressed (90 deg field compressed into 50
deg)
Task:
Procedure: 20 baseline, 40 trials in altered environment, 20 recovery trials.
Data recorded: XYZ position of hand and ball every 17 msec.
Also whether S hit the ball or not.
Results
How do we measure adaptation?
Successful versus unsuccessful hit
Trajectories: X vs Y relative to ball.
Plot performance as proportion correct every 5 trials.
Plot baseline, adaptation, and recovery.
How complete is the adaptation? (compare last adaptation trial with baseline)
Is there any after effect? (compare recovery trials with baseline)
How does trajectory change with practice? Do trajectories get closer to baseline?
Is there a difference between translation and compression? Why?
Discussion
Review findings.
Evaluate extent of adaptation. Was plasticity demonstrated?
Are subjects really learning a new set of relationships of just learning to
ignore the visual feedback?
How could we distinguish these possibilities?
Neural basis of adaptation?
Possible sites: Posterior parietal cortex (AIP, MIP),
supplementary motor area, pre-motor, motor cortex, cerebellum,
basal ganglia …
Ability to adapt to new relationships requires cerebellum
Why do we need to retain plasticity for new
visuo-motor relationships?
1. Need to adjust to changes in body size during development.
2. Need to adjust to damage/aging.
3. Need to adjust to environmental changes eg ice, loads etc.
4. Need to learn arbitrary mappings for tool use etc.
5. Need to acquire new motor skills.
6. Visuo-motor coordination is a computationally difficult problem for
the brain. Need flexibility to correct errors.
Role of Experience in
Development of Visuo-motor
coordination
Held & Hein
1
2
Both kittens get visual experience and motor experience
1. Visual experience correlated with motor
commands/proprioceptive feedback/vision of limbs
2. Gets both, but uncorrelated. Kitten 2 -abnormal visuomotor coordination.
Adaptation to different relation between
vision and movement.
George Stratton
–Wore inverting lens for 8 days
If he saw an object on the right he would reach with his right hand and discover
he should have reached with his left. He could not feed himself very well, could
not tie his shoelaces, and found himself severely disoriented. His image of his
own body became severely distorted. At times he felt his head had sunk down
between his shoulders,and when he moved his eyes and head the world slid
dizzyingly around.
As time went by Stratton achieved more effective control of his body. If he saw an
object on the right he would reach with his left hand. He could accomplish normal
tasks like eating and dressing himself. His body image became almost normal
and when he moved his eyes and head the world did not move around so much.
He began to feel as though his left hand was on the right, and his right hand on
the left. If this new location of his body was vivid, the world appeared right side
up, but sometimes he felt his body was upside down in a visually right-side-up
world.
After removing the prisms, he initially made incorrect reaching movements.
However, he soon regained normal control of his body.
Adaptation to different relation between
vision and movement.
George Stratton
– Wore inverting lens for 8 days
– Believed that we learn visual directions by
associating visual experiences with other forms of
sensory feedback (e.g. proprioceptive).
– Alternatively…
Adaptation results from learning correlation betweeen
vision and actively generated motor commands
(Held, 1965).
Function of Different Areas
monitor/plan
movements
target selection
saccade decision
inhibits SC
saccade command
signals to muscles
Schematic Representation of Feedback and Feed-forward Systems
Eg: pursuit, reaching, grasping
Eye velocity=image velocity
sensory
Motor command
retinal velocity
delay
Load/fatigue/current position
Eg: saccade, throwing
wind
ballistic
Learnt motor command
guided
Flipping an Egg
Autonomous control: robot is pre-programmed - no human input
Problems to be solved:
1. Grasp spatula
locate handle (vision)
some mechanism to translate location into arm movement
some mechanism for controlling fingers - “passive compliance”
2. Move to pan
locate pan (vision)
translate location into arm movement
3. Lower spatula to pan
vision or
proprioception: lower until force > 0
4. Flatten
proprioception: rotate until forces on fingers are equal
5. Locate egg
vision or
proprioception: move forwards until horizontal force > 0
Describe 2 functions of eye movements and give an example of each.
Types of Eye Movement
Information Gathering
Voluntary (attention)
Stabilizing
Reflexive
Saccades
vestibular ocular reflex (vor)
new location, high velocity, ballistic
body movements
Smooth pursuit
optokinetic nystagmus (okn)
object moves, velocity, slow
whole field image motion
Vergence
change point of fixation in depth
slow, disjunctive (eyes rotate in opposite directions)
(all others are conjunctive)
Fixation: period when eye is relatively stationary between saccades.
Draw a sketch of the brain showing the structures involved
in the generation of a saccadic eye movement. Specify the
function of these structures (to the extent that this is possible)
Flipping an Egg (ctd)
Problems to be solved:
6. Lift
need to keep spatula level: vision or proprioception (keep tension constant)
7. Flip
need to learn how much to rotate hand: rotate until forces = 0
Limitations of autonomous control:
inflexible - can’t adapt to changed circumstances
requires high precision
Mike May - world speed record for downhill skiing by a blind person.
Lost vision at age 3 - scarred corneas.
Optically 20/20 - functionally 20/500 (cf amblyopia)
Answer to Molyneux’s question: Mike May couldn’t tell difference between sphere and
cube. Improved, but does it logically rather than perceptually. (cf other cases)
Color: an orange thing on a basket ball court must be a ball.
Motion: can detect moving objects, distinguish different speeds.
Note: fMRI shows no activity in Infero-temporal cortex (corresponding to pattern
recognition) but there is activity in MT, MST (motion areas) and V4 (color). Other parts of
brain take over when a cortical area is inactive.
Cannot recognize faces. (eyes, movement of mouth distracting)
Can’t perceive distance very well.
Can’t recognize perspective.
No size constancy or lightness constancy/ segmentation of scene into objects, shadows
difficult.
Vision most useful for catching balls (inconsistent with Held & Hein??) and finding things
if he drops them.