Sensation and
perception
2022-2023
,Motion perception
Lecture 8
Motion sensitivity
Motion common mechanism: early visually guided organism detects motion, but not color, stereo
(depth) and form.
- Not for colour
o Colour blind
o Different for different colours
- Stereo: need for two eyes + feeling of normal stereo perception, but we don’t notice stereo
on different points on the retina
- Form: prosopagnosia, limitation of form processing
Not the case for motion!
Why so few motion-perception related problems?
- Robust because, many areas are motion sensitive: alle of them are invited somehow
o V1: extract motion
o MT: extract motion
But others are still responsive to motion -> need a lot of damage to impair motion
Is there one motion system? / one way of motion sensation? No -> answer: how deal different
animals deal with motion?
- Motion processing models: started looking at the retina -> found (first in fruitfly): all in the
retina, while in humans it’s not.
What is motion?
Intuitive: change of an object through space and time -> but not in the brain!
- Example: can’t see an object, only if it moves = objects can be defined by motion
o So, not only the retina, but higher processing needed.
- Example: perceiving left or right movement based on the assumption about the background
- Example: illusory rotation, while there is no rotation.
Motion isn’t straightforward object processing
Motion after effect: looking at motion -> picture seems to move = an object doesn’t have to move to
perceive motion.
Summary: objects can be based on motion + motion perception is not necessarily based on objects.
Bilocal Correlator: input from one location – which is send to a collector side X – the other box gives
delayed input = given the right speed, the signals will end at x in the same time.
- Velocity = span/delay.
- X requires input from two cells to give an response
o So, it only excites when something moved!
- But: simplified version, critic
o it will also respond to an on-off flicker (when it’s not apparent
during the span!)
,more complex version solves a bit of the problem: reichart detector (but
the other one is sufficient to understand)
- The x needs input two times: both from point 1 & 2 -> two
incoming signals: excitation
- The second x input is inhibitory and needs also two inputs,
becomes important when:
o Inhibition if it doesn’t move (activation of both stimuli at
the same time)
o Inhibition for flicker (on/off)
Signal cancels out due to inhibition when the stimulus doesn’t
move
Speed selectivity:
- Change the delay: longer delay means that the time between the two collector points is
longer = slower speed
- Change the distance between the two input sizes: change the distance between inputs –
larger distance, but same amount of tine = faster speed
There is no all-or-nothing principle
Realise: there is also a direction perforation -> wrong direction = not arriving of signals at the same
time. (image)
- Change direction preference: change the delay
Detector failure: signals can’t be matched based on luminance / color anymore -> the failure to see
motion (to the same degree) suggests the bilocal correlator will not be activated at both parts of it’s
receptive field.
Detector problems: moving an object with different spatial frequency over a different spatial
frequency -> this shouldn’t work, because all objects are moving at the same time.
- The ability to thee this kind of motion suggests we use additional mechanisms beyond the
reichart detector -> 2nd Order Motion
First order motion models = based on luminance changing through space and time. BUT 2 nd Order
Motion models are not based on luminance.
Second order:
1. 1st order motion: luminance based (Reichart detector) – example: motion based on motion
2. Temporal texture: texture information to build second order motion effect – no background
temporal information, but the foreground has.
3. Spatial frequency texture: spatial frequency in back- and foreground changes
Motion defined by motion: connect bicolar receptive fields (= idea of model building).
- 1&2 combined = signal (first order)
- 2&3 combined = signal (first order)
When both signalled, the second order X will be activated.
Are these 1 order and 2 order motion different in the brain?
, - Interocular transfer: look with one eye at red and the other at white -> no green after effect
= lack of interocular transfer (activation in one eye is not transferred to the eye)
o Higher in the hierarchy – more binocular cells: doesn’t matter which eye the
information comes from
Good pin point if something is early in the visual field (early means interocular transfer)
- Found evidence: more interocular transfer of 2 nd compared to 1st order motion
What do we know about the different areas? Importance of receptive field sizes.
Local motion: every object has it’s own way of moving -> small receptive fields.
- V1
o But receives input from LGN, V2, MT (so it also receives feedback from later visual as
well as non-visual areas)
o Known for it’s orientation selectivity
o About 20% show direction selectivity
- V2-V3
o Many direction selective responses
o Receives much of their input from V1
o Direction selective properties similar to V1
Tuning properties are much the same, so doing the same thing in exactly the same way. But
they have cells sensitive to motion.
Aperture problem: grating moving behind the foreground – both able to perceive it as different
parts (local motion) or as a whole (global motion).
- Small receptive fields: provide local information
o but local motion signal different from global motion signal
large receptive fields capture global motion signal
perceiving as moving in different directions. Because of limited information, the motion stays
ambiguous.
- Small receptive fields donot give you the full information and can cause imbigeous stimulus
Global motion: partly the objects are moving in the same direction
- V1 local motion -> integrating -> enough in the same direction? = global motion!
o By integration causing a good global motion percept
- Medial Temporal (MT), also known as human V5
o Input from Superior Colliculus, Pulvinur and V1-4 (even from the cortex)
o 90% of the cells are direction selective = motion processing cells
o Large receptive fields = integration of small receptive fields
o Associated with the perception of motion (not only processing!)
Complex motion: integrates local and global signals to create more complex versions of motion.
- Expansion & contraction & rotation
- Medial superior temporal area: even larger receptive fields
o Also integration of speed and direction needed
Biological motion (=complex motion): white dots as a representation of soccer players who beat each
other -> only the movement pattern = still visible that it’s biological (human) motion.
perception
2022-2023
,Motion perception
Lecture 8
Motion sensitivity
Motion common mechanism: early visually guided organism detects motion, but not color, stereo
(depth) and form.
- Not for colour
o Colour blind
o Different for different colours
- Stereo: need for two eyes + feeling of normal stereo perception, but we don’t notice stereo
on different points on the retina
- Form: prosopagnosia, limitation of form processing
Not the case for motion!
Why so few motion-perception related problems?
- Robust because, many areas are motion sensitive: alle of them are invited somehow
o V1: extract motion
o MT: extract motion
But others are still responsive to motion -> need a lot of damage to impair motion
Is there one motion system? / one way of motion sensation? No -> answer: how deal different
animals deal with motion?
- Motion processing models: started looking at the retina -> found (first in fruitfly): all in the
retina, while in humans it’s not.
What is motion?
Intuitive: change of an object through space and time -> but not in the brain!
- Example: can’t see an object, only if it moves = objects can be defined by motion
o So, not only the retina, but higher processing needed.
- Example: perceiving left or right movement based on the assumption about the background
- Example: illusory rotation, while there is no rotation.
Motion isn’t straightforward object processing
Motion after effect: looking at motion -> picture seems to move = an object doesn’t have to move to
perceive motion.
Summary: objects can be based on motion + motion perception is not necessarily based on objects.
Bilocal Correlator: input from one location – which is send to a collector side X – the other box gives
delayed input = given the right speed, the signals will end at x in the same time.
- Velocity = span/delay.
- X requires input from two cells to give an response
o So, it only excites when something moved!
- But: simplified version, critic
o it will also respond to an on-off flicker (when it’s not apparent
during the span!)
,more complex version solves a bit of the problem: reichart detector (but
the other one is sufficient to understand)
- The x needs input two times: both from point 1 & 2 -> two
incoming signals: excitation
- The second x input is inhibitory and needs also two inputs,
becomes important when:
o Inhibition if it doesn’t move (activation of both stimuli at
the same time)
o Inhibition for flicker (on/off)
Signal cancels out due to inhibition when the stimulus doesn’t
move
Speed selectivity:
- Change the delay: longer delay means that the time between the two collector points is
longer = slower speed
- Change the distance between the two input sizes: change the distance between inputs –
larger distance, but same amount of tine = faster speed
There is no all-or-nothing principle
Realise: there is also a direction perforation -> wrong direction = not arriving of signals at the same
time. (image)
- Change direction preference: change the delay
Detector failure: signals can’t be matched based on luminance / color anymore -> the failure to see
motion (to the same degree) suggests the bilocal correlator will not be activated at both parts of it’s
receptive field.
Detector problems: moving an object with different spatial frequency over a different spatial
frequency -> this shouldn’t work, because all objects are moving at the same time.
- The ability to thee this kind of motion suggests we use additional mechanisms beyond the
reichart detector -> 2nd Order Motion
First order motion models = based on luminance changing through space and time. BUT 2 nd Order
Motion models are not based on luminance.
Second order:
1. 1st order motion: luminance based (Reichart detector) – example: motion based on motion
2. Temporal texture: texture information to build second order motion effect – no background
temporal information, but the foreground has.
3. Spatial frequency texture: spatial frequency in back- and foreground changes
Motion defined by motion: connect bicolar receptive fields (= idea of model building).
- 1&2 combined = signal (first order)
- 2&3 combined = signal (first order)
When both signalled, the second order X will be activated.
Are these 1 order and 2 order motion different in the brain?
, - Interocular transfer: look with one eye at red and the other at white -> no green after effect
= lack of interocular transfer (activation in one eye is not transferred to the eye)
o Higher in the hierarchy – more binocular cells: doesn’t matter which eye the
information comes from
Good pin point if something is early in the visual field (early means interocular transfer)
- Found evidence: more interocular transfer of 2 nd compared to 1st order motion
What do we know about the different areas? Importance of receptive field sizes.
Local motion: every object has it’s own way of moving -> small receptive fields.
- V1
o But receives input from LGN, V2, MT (so it also receives feedback from later visual as
well as non-visual areas)
o Known for it’s orientation selectivity
o About 20% show direction selectivity
- V2-V3
o Many direction selective responses
o Receives much of their input from V1
o Direction selective properties similar to V1
Tuning properties are much the same, so doing the same thing in exactly the same way. But
they have cells sensitive to motion.
Aperture problem: grating moving behind the foreground – both able to perceive it as different
parts (local motion) or as a whole (global motion).
- Small receptive fields: provide local information
o but local motion signal different from global motion signal
large receptive fields capture global motion signal
perceiving as moving in different directions. Because of limited information, the motion stays
ambiguous.
- Small receptive fields donot give you the full information and can cause imbigeous stimulus
Global motion: partly the objects are moving in the same direction
- V1 local motion -> integrating -> enough in the same direction? = global motion!
o By integration causing a good global motion percept
- Medial Temporal (MT), also known as human V5
o Input from Superior Colliculus, Pulvinur and V1-4 (even from the cortex)
o 90% of the cells are direction selective = motion processing cells
o Large receptive fields = integration of small receptive fields
o Associated with the perception of motion (not only processing!)
Complex motion: integrates local and global signals to create more complex versions of motion.
- Expansion & contraction & rotation
- Medial superior temporal area: even larger receptive fields
o Also integration of speed and direction needed
Biological motion (=complex motion): white dots as a representation of soccer players who beat each
other -> only the movement pattern = still visible that it’s biological (human) motion.
