L4 - Motion Flashcards
What 3 types of perceptual information do we gain from motion?
Figure-Ground Segregation
Extraction of three-dimensional structure
Visual guidance of action
How does motion help figure-ground segregation?
When there is camouflaged it is hard to detect, however when it moves it becomes easier to identify.
How does motion aid in the extraction of three-dimensional structure?
If something is static you don’t get as good of an idea of it’s volume until it moves
How does motion aid in visual guidance of action?
Motion information is crucial when interacting with the environment in order to avoid harming oneself and moving fluently in the world
What did Barlow and Levick (1965) discover?
Neurons that are responsible for motion.
What happens to the motion neuron if a stimulus in the visual field of a neuron is moving in the preferred direction?
You get an increase in the response of the neuron.
What happens to the motion neuron if a stimulus in the visual field of a neuron is moving in the null direction?
You get no change in response.
Why was Barlow and Levick’s discovery of motion neurons a big deal in psychology?
It showed that there is a direct relationship to what is happening in the brain in response to the environment.
Neurons were coding movement in the world.
Are cells in the retina directionally selective?
Yes
Suggested building block for motion
What is directional tuning?
Humans can detect all orientations of movement through many specific neurons.
How might you model motion detection?
Have a stimulus that changes its location over time and record responses at different locations.
T1->T5 (bottom left) is stacking on top of one another. If we do this we get orientation in space and time.
T1 = time 1 etc.
What is a simple motion detection unit in the brain?
How does it work?
A comparator neuron
It has two input signals, one of which is delayed. If a stimulus is moving in one direction the delayed receptor and the normal receptor will reach each other at the same time, indicating movement in a certain direction.
Who proposed the simple motion detection unit model?
What does it allow you to do?
Reichardt (1961) (From his work analysing flies)
Allows you to detect the difference in motion between A and B
We can think of motion as orientation in _____ and _____.
Space and Time
Who proposed the Energy Model?
Adelson and Bergen (1985)
Burr, Ross and Morrone (1986)
Which model is this?
How does it work?
Reichardt model
Which model is this?
How does it work?
Energy Model
What is the extra stage in the energy model compared to the Reichardt model?
Oriented Linear Response
What does the Oriented Linear Response allow us to do in the energy model?
Allows us to aggregate information for motion in space and time.
What are the 3 benefits of the energy model?
No need to discrete correspondence (allows us to aggregate over more instances, if our filters don’t quite match the aggregation makes movement easier to catch)
Specificity: Motion response is to a particular space, time and spatial frequency.
More flexible definition of a feature: Ability to detection motion in random noise patterns
What is the Aperture problem (barber pole illusion)?
The grading is moving from left to right, however, the bars look like they are moving up and down.
It looks like it changes direction depending on the orientation of the aperature
Why does the Aperture problem exist?
The motion of a repetitive pattern is essentially featureless.
What is meant by local vs global features in regards to the Aperature problem?
Local receptive fields (motion detectors in the retina) do not have enough information to disambiguate direction of the object’s motion.
In order to disambiguate the information, we use the orientation information of the edges of the aperture to determine the direction (global).
This is why changing the shape of the aperture will make the lines seem like they are moving in a different direction.
Do you get the barber pole illusion if the motion information if you use a non-repetitive pattern?
No, because it has features you will be able to determine the correct direction of motion.
What does ‘pooling of information’ mean in regards to disambiguating motion direction?
We can’t use just local information if the moving formations are featureless.
Therefore we pool ‘movement’ information and the ‘form’ information to generate a sense of direction.
What are 3 tools we use to disambiguate direction?
Line endings (where the line ends)
Form information from the edges of the aperture
Global pooling of local information
What is the average integration time for cells in V1?
100ms
They collect information over 100ms before they make some kind of response
How does the delay in V1 cells help us disambiguate motion direction?
Through detecting a blur, then by filtering it using a motion deblurring mechanism.
How did Geisler (1999) test the V1 motion deblurring theory?
Showing stimulus in two different directions and adding ‘noise’ to see if it interferes with the performance
How do we use motion streaks to detect motion?
If you are moving along you have orientation information and you are also getting orientation information along the axis of that (with a slight delay from V1) and you can use that to disambiguate motion.
Can we use form information alone to generate a perception of coherent global motion?
How was this demonstrated?
Yes
Showed using Glass patterns: series of points on a screen randomly and you apply a rule (e.g. be radial)
- No one dot is the same but because there is a global rule they see motion.
What part of the brain does ‘glass’ patterns activate?
What does this imply?
MT/V5 (for motion)
Implies that our brain sees the individual dots as motion.
What did Albright, 1984 discover about cells in MT?
They are sensitive to the direction of motion, but also orientation and pattern information.
Does orientation information contribute to perceived motion direction?
Yes
What did Johansson (1963) demonstrate?
What is this called?
That people are able to recognise people walking, even when there are only points of light on the joints.
Biological Motion
What does Biological Motion suggest about how our brain processes movement information?
That there is a high-level constraint on low-level information.
We pool our low-level information and combine it with high-level information we have about form to see biological motion
