3- Visual Motion Processing Flashcards by Anya Laila

What 4 main things is motion good for?

1-Motion-based image segmentation
2-Navigation
3-Depth from motion
4-Structure from motion

How well did you know this?

Not at all

Perfectly

What happens when there is bilateral damage of V5?

Loss of vision
-see the world as static snapshots

How well did you know this?

Not at all

Perfectly

When is feature tracking used?

When there are:
-few objects/features,
-long intervals or large displacements

How well did you know this?

Not at all

Perfectly

Limitations of local motion analysis:
Explain Aperture problem

Due to V1’s small receptor fields, local motion detectors in V1 only ‘see’ a small part of the image and respond to motion that is only orthogonal to luminance edges.

As a result the output of any one motion detector may not be a valid indicator of the overall direction an object is moving.

What is a solution to this?

To combine the responses of detectors with receptive fields located in different regions of space

(take the outputs of lots of neurons in V1 and put them together to figure out what the actual motion of object is)

How well did you know this?

Not at all

Perfectly

As this moves, the dots are perceived as an object (cylinder). Yet the direction of rotation is actually ambiguous (can move back and forth) This is this representing?

Structure from motion
-biological motion (dots in the form of a walking man)

How well did you know this?

Not at all

Perfectly

What is this centre of heading representing?

Navigation

How well did you know this?

Not at all

Perfectly

What is this representing?

Motion-based image segmentation

How well did you know this?

Not at all

Perfectly

Optic flow patterns reduces an expanding flow field (eg. driving down a country road), locating the centre of expansion (centre of the road) tells us our direction of heading is known as…

Navigation

How well did you know this?

Not at all

Perfectly

Which stream is responsible for motion?

Dorsal

How well did you know this?

Not at all

Perfectly

What is this representing?

Depth from motion

How well did you know this?

Not at all

Perfectly

What area in the primary visual cortex is responsible for motion processing?

V5
Dorsal stream

How well did you know this?

Not at all

Perfectly

Dividing the image into regions based on the motion of objects within the scene is known as…

Motion-based image segmentation

How well did you know this?

Not at all

Perfectly

What tells us how far away objects are depending on the speed at which they move (eg. things closer will move faster, things further will be perceived as slower)?

Depth from motion

How well did you know this?

Not at all

Perfectly

Feature tracking / cognitive strategy / long-range motion are all direct or indirect motion detection methods?

Indirect

How well did you know this?

Not at all

Perfectly

How is motion computed in the visual brain?

Indirect method:

Feature tracking

Independent analysis of spatial displacements (where) and temporal intervals (timings)

-by viewing where an object is in 2 different time points and acknowledging their location has changed (car moving after traffic light)

How well did you know this?

Not at all

Perfectly

What method to detect motion infers speed and direction of motion?

feature tracking

How well did you know this?

Not at all

Perfectly

How is motion computed in the visual brain?

Direct method:
detects motion automatically

specialised detectors compute motion directly from intensity variations in the retinal image without feature tracking

Motion perception remains possible even when the spatial or temporal displacement is so small and we cant resolve it
(sub-threshold spatial and temporal
displacement)

-brief, high density displays which preclude cognitive tracking of features

How well did you know this?

Not at all

Perfectly

Direct methods of detecting motion automatically works best under conditions where the indirect method doesn’t work properly such as…

high density displays

these don’t have any obvious features that we can track over time, but when moving we can tell
(these preclude cognitive tracking of features)

How well did you know this?

Not at all

Perfectly

Limitations of local motion analysis:
Explain Aperture problem

1- due to V1’s small receptor fields, local motion detectors in V1 only ‘see’ a small part of the image and respond to motion that is only orthogonal to luminance edges

leading to ambiguous estimates (unclear of true action of motion of objects sometimes)

solution: combine the responses of detectors with receptive fields located in different regions of space

How well did you know this?

Not at all

Perfectly

Limitations of local motion analysis:

speed is the ratio of temporal frequency and spatial frequency (speed = TF / SF)

Explain Speed selectivity

There are issues with judging how fast objects are moving

V1 neurons tend to respond selectively to a specific combination of TF and SF (rather than to all stimuli moving at a particular speed)

Solution to this:
combine the output of detectors that respond to the same TF/SF ratio across v1 neurons

How well did you know this?

Not at all

Perfectly

What is the purpose of an internal delay in the Reichardt model of motion detection?

-It matches time it takes for stimulus to move between receptive fields, detector (M) will signal rightwards motion
-Movement in opposite (leftwards) direction gives no response

How well did you know this?

Not at all

Perfectly

Motion computed in the visual brain:

Motion can be represented by its orientation of pattern (1D image indicates no motion, but staggered 2D image mimics motion)

What is this due to?

Motion-sensitive receptive fields

Luminance profiles of moving contours (image) are oriented in space-time

Result: detecting motion (direction and speed) is analogous (has similarities) to extracting spatial orientation

How well did you know this?

Not at all

Perfectly

Some V1 cells have receptive fields which are oriented in space-time. What does this mean?

They respond strongly to:
oriented edges
moving in a preferred direction

But NOT at all in the opposite direction

How well did you know this?

Not at all

Perfectly

3 dimensional displays of dots which are perceived to form objects is known as the…

Structure from motion

How well did you know this?

Not at all

Perfectly

Name the properties of MT neurons:

-large receptive fields Receive input from: many V1 neurons from both eyes are tuned to different binocular disparities (ratio of RF dimensions ~10:1) -located in middle temporal visual area (MT or V5) -responsible for motion

superior temporal sulcus (STS) is associated with which type of motion?

Biological

Finish the sentence: V1 neurons only respond to a particular orientation (edge or bar) moving in a particular direction, but MT neurons respond to ...

a preferred direction independent of pattern

Which model is used to explain how the brain detects motion via:

Reichardt model of motion detection -trying to pick up motion from a bar moving left to right -PVC has simple cells (A) with spatially orientated receptor fields (only respond to stimuli in a certain position of the receptive field) -moving bar stimulates a first response in neuron A, then followed by a second response in B due to non-motion orientated selective neurons To compare 2 outputs: -add in an extra stage which requires simultaneous stimulation from each input of A+B (image 2) via an internal delay

Sate 2 limitations of local motion analysis?

1-Aperture problem 2- Speed selectivity

Which 2 regions of higher-order motion processing have been implicated in the extraction of structure from motion by fMRI studies? A- intra-parietal sulcus (IPS) B- medial superior temporal (MST) C- superior temporal sulcus (STS) D- lateral occipital sulcus (LOS)

A- intra-parietal sulcus (IPS) and D- lateral occipital sulcus (LOS)

Which area of higher-order motion processing has particularly large receptive fields and has neurons that respond selectively to complex optic flow patterns such as rotation, contraction, expansion? A- superior temporal sulcus (STS) B- medial superior temporal (MST) C- intra-parietal sulcus (IPS) D- lateral occipital sulcus (LOS)

B- the medial superior temporal (MST)

Beyond MT/V5 this is known as...

Higher-order motion processing -the further we go in the visual system, the more complex the characteristics of motion are selective to

True or false: Many more neurons in MT also show proper speed tuning than in V1

True

True or false: MT neurons receive input from both eyes and are tuned to different monocular disparities

False, different binocular disparities

Which area of higher-order motion processing has neurons which respond selectively to complex biological motion (motion of people/animals)? A- superior temporal sulcus (STS) B- medial superior temporal (MST) C- intra-parietal sulcus (IPS) D- lateral occipital sulcus (LOS)

A- superior temporal sulcus (STS)

Name the 4 types of motion associated with each hierarchal processing area: MT/V5, V3a MST, V6 STS IPS

MT/V5, V3a- Global motion MST, V6- Complex motion STS- Biological motion IPS- Structure from motion remember this order!

medial superior temporal (MST) is associated with which type of motion?

Complex

intra-parietal sulcus (IPS) and lateral occipital sulcus (LOS) are associated with which type of motion?

Structure from motion

Neural consequences of motion adaptation: What happens when we respectively start to stimulate a neuron that's motion selective in PVC V1? Repeated stimulation in the preferred direction...

reduces the responsivity (stimulus response) of direction-selective V1 neurons (direction of motion) -adaptation reduces the response of some neurons to stimuli at all contrast levels (response gain), but shifts the contrast response function laterally for others (contrast gain)

Neural consequences of motion adaptation: When adapting to motion-selective neurons in MT The contrast gain shifts in function after adaptation, reducing the neuron response (black is post adaptation) As the changes in MT and V1 are similar, we question whether the effects are actually happening in MT, or they are inherited from changes of V1 Experiment: to test this, used smaller stimuli presented in sub-regions of receptive field Found: effects disappeared What does this suggest?

suggests that effects of adaptation are likely inherited from earlier levels (e.g. V1)

Neural consequences of motion adaptation: Researchers measured adaptation of MT cells over shorter timescale following the onset of motion, MT neurons display an initial transient peak (spike) in firing rate that decays to a sustained level over ~100ms this decay is thought to be a form of motion adaptation this reduction in responsiveness is maintained when stimuli are moved to different sub-regions of the receptive field, what does this suggest about where motion processes are localised?

suggesting that it is not inherited from V1 inputs

Perceptual consequences: The motion aftereffect is also known as...

waterfall illusion

Perceptual consequences of motion adaptation: Motion adaptation produces changes in sensitivity and biases that mirror those seen following adaptation to static stimuli What 2 things does this refer to?

Contrast sensitivity -contrast sensitivity is reduced for gratings moving in the adapted direction, but not in the opposite direction (eg. if we look at a stationary stimulus and adapt to a moving stimulus, our ability to detect a similar moving stimulus will be impaired). Direction aftereffect (tilt aftereffects) -adaptation to a given direction of motion produces repulsive biases in the perceived direction of subsequently viewed stimuli (eg. if we adapt to dots moving upwards to the left, then view dots moving upwards, it appears to be moving upwards to the right) shows tuning of effect between adaption and test direction (repulsive effect is being pushed away)

Perceptual consequences: What is it called when we adapt to a moving stimulus for a period of time (its given direction of motion), and the static or moving objects appear to move in the opposite direction?

The motion aftereffect stare at top photo and then the bottom will look like its moving

Which model explains the motion aftereffect?

Distribution shift model

Which model explains that neurons are first tuned to all different directions some respond a lot if stimulated by upwards motion, but if its direction is changed, its response starts to decrease? Prior to adaptation, a static test pattern produces similar responses from motion selective neurons tuned to all directions but then adaptation selectively reduces neural responsivity to the adapted direction, causing a shift in the distribution of responses to the test pattern

Distribution shift model

Properties of the static motion aftereffect: Traditionally, investigations of motion aftereffects use static test patterns, and measure the duration of the illusory motion percept (P sees the motion aftereffect in the opposite direction and you ask people how long that effect lasts) Strong aftereffects will persist for longer and are measured by duration This static motion aftereffect is...

Name the 4 main properties of the static motion aftereffect:

1-Location specific 2-Spatial frequency tuned 3-Partial interocular transfer 4-Temporal frequency tuning

What property of the static motion aftereffect is this referring to? Adapting to motion at one retinal location, but then view a static image somewhere else in the receptive field, it doesn’t result in a MAE at other locations (to fit the neurons with small receptive fields in V1)

Location specific

What property of the static motion aftereffect is this referring to? The strongest MAE is elicited when the adapting and test stimulus have similar spatial frequency (most neurons in V1 are tuned to particular spatial frequencies)

Spatial frequency tuned

What property of the static motion aftereffect is this referring to? Strongest MAE obtained at a constant temporal frequency (~8Hz), regardless of stimulus spatial frequency (Pantle, 1974). (neurons in V1 are tuned to S+T frequencies instead of speed)

Temporal frequency tuning

What property of the static motion aftereffect is this referring to? Switching eyes between adapting and test periods reduces the MAE, but does not abolish it (compared to testing with both eyes due to having a mixture of neurons that respond to one/ both eyes)

Partial interocular transfer

MAE is known as...

Motion Aftereffect

There is an issue with having people self report/ indicate how long their motion aftereffect lasts due to subjectivity as some people may have a tendency to report subtle effects than others. What is an alternative way to measure the MAE?

To test with a moving or flickering test stimulus and try to null the percept of illusory motion -this image doesn't necessarily depict a direction of motion as it could be either, thus P equally likely to say up or down -if there is more contrast downwards> most P will say downwards direction of motion vice versa, but when matched its 50/50 -post adaption to upwards contrast, P will view original contrast ratio as moving downwards (opposite direction) HOWEVER, if we add in enough upwards contrast we can make the motion image ambiguous in its direction of motion again. Ie- The strength of the MAE can be quantified by measuring the shift in this null point after adaptation in a given direction

The dynamic MAE shows several characteristics consistent with adaptation at a higher stage of processing, where local motion signals are integrated. Name the main properties of a dynamic motion aftereffect: (looks like what you would expect at that later stage in MT, more complex)

-Low positional specificity (can detect motion regardless of position in space) -Complete interocular transfer (doesn't care which eye it adapts to, effect is just as strong thus having binocular neurons which respond to both) -Speed tuning neurons in MT respond to actual speed, regardless of the orientation of their T+S frequencies

Neurons in MT are binocular or monocular?

monocular - respond to visual stimuli from both eyes as the stage of visual processing is more complex at MT

Neurons in MT have Complete interocular transfer. What does this mean?

neuron doesn't care which eye it adapts to, effect is just as strong, it responds to info from both eyes thus having binocular neurons, complex part of visual processing

Which property of a dynamic motion aftereffect is this referring to? (can detect motion regardless of position in space)

-Low positional specificity

Which property of a dynamic motion aftereffect is this referring to? neurons in MT respond to actual speed, regardless of the orientation of their T+S frequencies

Speed tuning

Evidence card Evidence for high-level motion adaptation: observers adapted to biological motion stimuli which distinguished males from females adaptation to the gate of one gender biased judgements of subsequently viewed gates towards the other gender (so if you view a male first, post adaption you are more likely to identify the neutral image as female opposite gender) What about when randoming the dots?

Randomising the phase of each individual dot’s motion -significantly reduced the size of the effect, (P estimates of gender is unclear) -suggests that it was not due solely to local adaptation (ie- not the local movement of dots in the exact fashion of genders walking) -now it is unclear biological motion

What are the 3 theoretical ideas about what the function of motion adaptation is?

1 Improved discriminability ! -Shown to improve speed discrimination in humans/ monkeys 2 Improved efficiency -Improves the efficiency of the neural code by equalizes response levels over time and reducing redundancy 3 Normalization -Motion adaptation acts to recalibrate the zero-velocity point (face coding relative to prototype or not)

Fill in the sentence: Motion can be inferred ... , by tracking the position of objects/features over time. However, the visual system also contains ... for computing motion directly from intensity variations in the retinal image.

indirectly, specialised detectors

Which processing does this describe: V1 contains some motion selective neurons, which have receptive fields that are oriented in space-time. These neurons respond to motion orthogonal to luminance edges and tend to be tuned to a particular combination of SF and TF. A-Higher-order motion processing B-Global motion processing C-Local motion processing

C- Local motion processing

Which processing does this describe: MT neurons receive input from many V1 neurons. They respond to a preferred direction of motion, independent of the type of pattern. Some MT neurons also show true tuning for image speed. Neurons in area MST display preferences for complex motion stimuli such as optic flow fields. A-Higher-order motion processing B-Global motion processing C-Local motion processing

B- Global motion processing

Which processing does this describe: Processing of biological motion and structure from motion has been associated with dorsal stream areas such as the superior temporal sulcus (STS) and intra-parietal sulcus (IPS) A-Higher-order motion processing B-Global motion processing C-Local motion processing

A- Higher-order motion processing

Fill in the sentence: In V1 adaptation to drifting patterns produces similar effects to that seen with static stimuli. That is, adaptation reduces the neural responsivity, either by reducing overall ... or shifting the ... function

firing rate, contrast response

Fill in the sentence: Comparable changes to the contrast response function are also observed in MT, but likely reflect changes ... from V1 rather than ... in MT itself. Adaptation over a short timescale reveals adaptation effects that are intrinsic to MT.

inherited, occurring

Fill in the sentence: Adaptation to moving stimuli results in ... changes in contrast sensitivity and repulsive biases in perceived direction.

direction-selective

What is a supporting example of aftereffects of adapting to higher-order motion properties which have been found?

gender of biological motion

Fill in the sentence: After adaptation to motion in a given direction, static or directionally ambiguous stimuli appear to ... (the motion aftereffect).

move in the opposite direction

Location specificity, partial inter-ocular transfer, sf and tf tuning all describe which stage of processing?

Local processing

Low location specificity, full inter-ocular transfer, speed tuning all describe which stage of processing?

Global processing

Fill in the sentence: Measured with a static test stimulus, the MAE displays properties consistent with adaptation at a ... stage

local processing

Fill in the sentence: The dynamic MAE displays properties consistent with adaptation at a ... stage

global processing