Lecture 8 and chapter 8 - motion
Motion sensitivity in humans
- Every visually guided organisms detects motion
- Color, depth perception and form perception are not detected by all organisms
- 8% of male population → color blindness
- 7-10% of population → depth (stereo vision) does not work
- Most people become aware of their deficit only when explicitly tested for
stereo vision
- Many areas are motion sensitive
Motion sensitivity in different species
- Retina
- Rabbits, frogs, turtles, birds, squirrels
- Not all same mechanisms
- Not in humans
What is motion?
- A change in position of an object over time?
- No
- Not only of an object, because moving patterns are also motion but are not
an object that change in position
- Objects can be based on motion
- Motion perception is not necessarily based on objects
- To be able to perform an action one needs visual feedback, make eye movements
and more or less stabilize the retinal image
- Every visually guided organism detects motion, while not all organisms can perceive
color, depth or form
- Also, think about problems in vision, color blindness does not affect your day to day
life that much, just like strabismus, but however problems in detecting and
discriminating visual motion severely impairs our life
, - Motion is the moving of something going from A to B; something we can only
perceive when it is moving
Detection of global motion - process
- Is done by the magnocellular layers in the LGN (lateral geniculate nucleus),
which project to the V1 (primary visual cortex) and then go on to the MT (middle
temporal cortex)
- The vast majority of cells in the MT are direction specific and show little sensitivity
for form and color
- Experiment
- Information that the MT is indeed critically involved in processing global
motion comes from (lesion) studies in monkeys where they first trained
monkeys to detect correlated dot motion
- Were able to do this when 2-3% of the dots were moving in that direction,
after lesions only when 20-30% of the dots were moving
Space-time diagram
- Used to display the motion that is used in a particular experiment
- The flatter the slope, the faster the stimulus goes
- The steeper the slope, the slower the stimulus goes
- We do not have to recognize the object to see motion
- Motion can define the object
- Gradient detector: gradient in space and gradient in time derivative
- Energy detector: the smaller the lines are, the higher the frequency
- Thus, it does a Fourier analysis in space and time
Aperture problem
- Looking with V1, it will be hard to attribute what belongs to what
- Objects are defined by different characteristics
- Object recognition: our brain must put the parts (V1) together to a coherent whole
- Objects also need to be manipulated: we have to navigate in our environment
- Thus, we have perception for recognition (what/ ventral part) and for action
(dorsal/ where path)
- The problem
- The fact that when a moving object is viewed through an aperture (or
receptive field), the direction of motion of a local feature or part of the object
may be ambiguous
- It is an example of a correspondence problem: problems faced by the
motion detection system of knowing which feature in Frame 2 corresponds to
which feature in Frame 1. The aperture is in this case the opening that allows
, only a particular view of that object. Remember that every V1 cell sees the
world through a small aperture. Thus, none of the V1 cells can with certainty
tell which visual elements correspond to another when an object is moving.
The correct observation comes when these responses are combined.
Bilocal correlator/ Reichardt detector
- Based on a physiological measurement of the difference between point A and point B
and you look at span and delay, which is basically the same as space and time
- In physiological systems, all come down to variations on the bilocal correlator theme
- One possible objection to the Reichardt model is apparent motion
- It does not require continuous motion in order to fire
- An image that appears, then disappears and then reappears in a latter
receptive field in a short window will drive cells to respond just as strongly as
if the image has moved
- This is what is called apparent motion: the illusory impression of smooth
motion resulting from the rapid alternation of objects
- These cells are speed sensitive
- If it does not arrive at the same time, to the correlator it will not give
information further
- These cells are selective for motion direction
- You do not have to get the form before you define the object since motion can
define the object
- Alter speed sensitivity by two things
- Change the delay (the longer the delay, the slower)
- Change the span
Luminance Signal to Noise Ratio (LSNR method)
- How we can measure speed sensitivity
- Used in science and engineering that compares the level of a desired signal to the
level of background noise
- SNR is defined as the ratio of signal power to the noise power, often expressed in
decibels
- A ratio higher than 1:1 (greater than 0 dB) indicates more signal than noise
, First and second order motion
- There have been double dissociations where people still have first order motion but
not second order motion and vice versa
- The motion aftereffect is also more abundant in second order motion then in first
order motion
First order motion
- Sensitive to the displacement of luminance contours and the displacement of color
- The change in position of luminance-defined objects (objects that are delineated by
differences in reflected light) over time
Second-order motion / 2nd order motion - process
- Anything not being first order motion
- The motion of an object that is defined by changes in contrast or texture (not by
luminance)
- Its detector responds to displacement of second-order attributes even in the absence
of displacement of first order features
- The fact that almost all humans (and other animals) can perceive second-order
motion is a strong indicator that the visual system does not simply track objects’
positions over time to calculate movement
- Why we can see 2nd order motion while motion detectors can only see luminance
gradients:
1. One preprocessing pathway
- Cells that define process specific textures, those feed in bilocal correlator +
cell … = motion detector that detects motion of the texture
- Form processing can precede motion processing
2. Separate cortical pathways
- One that goes from V1 - MT for first order
- One that goes from V1 - V2 - V3 - MT for first and second order
3. Two motion layers
- There are multiple stages of motion process