Lecture 8: Depth perception Flashcards
Name primary visual cortex akas
Area v1
Striate cortex - stripped pattern when stain and look at
Name secondary cortex akas
Area v2
Extrastriate cortex
Describe retinotopic mapping
Spatial organization of visual field preserved in brain
Neighbouring areas of retina mapping onto neighbouring areas of primary visual cortex - v1
Point by point manner
Spatial relationships preserved in lgn and v1 = corresponds to specific regions in v1
How much cortical space in v1 dedicated to central vision
More cortical space dedicated to central vision than peripheral vision
What does fovea have
Higher density of ganglion cells than peripheral retina
Descrbe what lgn shows
Reflects fovea’s asymmetry of ganglion cells
= 2x as manny parvocellular layers as magnocellular layers
Describe cortical magnification
V1 further amplifies total number of neurons devoted to central vs peripheral vision where the central 10 degrees of visual field (1% of visual field) occupies 50% of v1s surface
Asymmetry further enhanced
More spaced allocated for processing visual info in Center of visual field
What are retinal ganglion cells sensitive to
Frequency
Not orientation
Like on center, off surround
Cannot tell is vertical or diagonal
Cells will fire the same - same activity, light and darkness produce same effect regardless of orientation
Descrbe response properties of v1 neurons
By combining info from several retinal ganglion cells= possible to detect orientation of lines
Hubel and Wiesel
Pattern diff for reach cell - since there are many = combine activity
Ap firing rate related
Describe how v1 neurons can detect orientation
Area v1 neuron connected to lgn neuron connected to retinal ganglion cell
Area v1 neurons have preferred orientation
If light perfectly aligns with center of cells = all fire= repsond vigorously
If receptive field organized differently = will be different and respond less
When are v1 neurons most active
Best around 70 degrees = preferred orienttaion
Describe tilt after effect
Have diff populations of neurons that specialize in certain frequencies and orientation
V1 neurons = turned to that orientation and d naturally become tired = less and less response
Describe tilt after effect- ADAPTATION
Looking at pattern fo stripes for a certain time will tire neurons and shift the balance to opposite direction
Then shift to opp direct = neurons habituate - become tired after fire for a while
WHAT is orientation encoded by
orientation encoded by a population of neurons
What does perception of objects rely on
Activity across whole pop of neurons in v1
Describe illusion effect = titled lines
Neurons tuned to vertical lines = will repsond most to this pattern
Neurons around 45 degrees exhibit decrease firing rate comapred to before habituation occurred
Ones not close to 45 degrees = wont change bc not habituated
= percieve orientation of lines as to the right
Hubel and Wiesel = thought could explain how see objects but actually more complicated
What is a hyper column
1 mm block of striate cortex - contains all machinery Jesse ray to look after everything the visual cortex is responsible for - in a certain small part of the visual world
V1 neurons organized into hyper columns
Describe hyper columns
Each contains cells corresponding to every possible orientation - 0-180 degrees, with one set preferring info from left eye and one set preferring info from right eye
Associated with specific area in visual field
Describe hyper columns = image
Insert electrode perpendicular to cortex and went through and looked at layers
Each column = have same orientation preference
Each neighbouring column = orienttaion close to other but diff angle
Each has 3 hyper columns per eye
Describe simple cells
Respond to light with increasing intensity when orientation matches receptive field
Lines with certain orientation
Orientation selectivity represented by tuning curves (firing rate vs line orientation)
Describe complex cells
Some detect motion in their receptive fields
Selective for line orienttaion and often for movement direction
Not simple
Describe receptive field properties
Of complex cells
Larger and less well defined than simple cells
Some span whole hemifield responding to specific orientation/movements across area
Describe specialized complex cells
End stopped cells
Respond best to moving bars of specific length that end within their receptive fields - responsive to lengths of lines
Do not fire if bars exceed their receptive field
Involved in detecting angles, corners, and boundaries
What forms basis of depth perception
Visual system good at keeping track of left and right fields = forms basis of depth perception
Describe depth perception generall
We naturally perceive depth - see 3d
But visual info reaching retina = 2d
Bc retina flat memrbane
Describe depth perception and Kant
Space = a priori form
Mind already knows space exists before receiving sensory info
Uses a priori knowledge to interpret sensory info
Sensory info does not tell us about depth
Know depth exists
Nervous system pre arranged to extract 3d from 2d
How do we infer depth
Use 2 cues =
Monocular cues
Binocular vision and stereopsis
Describe monocular clues
Do not need 2 eyes
Implicit inference - mostly logical
Ex = Tricked into thinking all stairs go up or down
Describe binocular vision and stereopsis
Seeing with 2 eyes
Refers to perception of depth we get from binocular vision
Much more hard wired
What are pictorial clues
Depth cues Inferred from static image
Using monocular vision
Help brain interpret depth in 2d scenes
Name types of pictorial cues
Occlusion
Relative size
Familiar size
Linear perspective
Texture gradient
Aerial perspective
Shading
Describe occlusion - pictorial cues
A depth cue wehre one object partially blocks another = indicates that blocked object is father away
One of most reliable depth cues
Desire accidental viewpoint - pictorial cues
Specific line fo sight that creates an ambitious or misleading depth interpretation
Ex = object behind another, could be that that object has cut out but looks whole bc of accidental viewpoint, also like an art structure = look at certain Angle and looks like a face
Describe relative size - pictorial cues
Comparison of size between items without knowing absolute size of either one
All things being equal = assume smaller objects farther away than large ones
Describe familiar size - pictorial cues
Comparison of size when knowing absolute size of one of the items
Can infer the absolute distance because we know exact size of one of teh objects
Ex=Know approx how big chair is in relation to humans
Describe relative height - pictorial cues
For objects touching ground = those higher in visual field, appear further away, looks like occlusion but if look at higher = far
For objects floating in air = shadows cna help infer their relative heights
Why does relative height infer depth - pictorial cues
Objects further away = seen with a wider angle relative to our body
When look at something close = decrease viewing angle
When sight parallel to ground - perpendicular to body = experience greatest feeling of distance - horizon
Describe texture gradient - pictorial cues
Pattern that repeats
Depth cue based on geometric fact that items of teh same size form smaller, closer spaced images father away they get
Further = appears smaller, for chairs, where are bird appears larger when further
Describe linear perspective - pictorial cues
Lines that are parallel in the 3d world wil appear to converge in a 2d image as they extend into distance
Lines project = to vanishing point = apparent point at which parallel lines receded in depth convergence
What can linear perspective also result from - pictorial cues
Combo of relative size and relative height = produces strong impression fo depth
Describe aerial perspective - pictorial cues
Depth cue based on implicit understanding that light scattered by atmosphere
More light scatted when look through more atmosphere
More distant objects appear fainter, bluer and less distinct - Bluer bc Rayleigh scattering and whiter bc mie and non selective scattering
Describe shading - pictorial cues
Variations in light and shadow on surface provide info about its shape, depth, orientation
Objects with gradual shading changes = appear curved
Sharp contrasts = suggest edges or discontinuities in depth
Brain assumes light comes form above - influencing depth perception
What are dynamic cues
Strong
Related to movement
Movement of images on retina
How is depth perceived when observer stationary
Depth perceived through relative movement of objects at diff depths (kinetic depth effect)
Effect strong for rotating spheres = where luminance variations father enhance depth perception - sphere videos
Describe motion parallax - dynamic cues
Images closer to observer move faster across visual field than images further away
Brain uses info to calculate distances of objects in environment = infer disatcne of object when moving in parallel fashion to scene
Describe motion parallax - dynamic cues - in real situations
Observer can move eyes to track object - which will give observer impression that objects beyond their fixation = moving in same direction as them
Objects located beyond fixation point look like following you/going in opposite direction
Seems like object beyond fixation point moves in same direction as you, not true tho bc moving eyes at same time you are moving
Describe optic flow - dynamic cues
Apparent motion of objects in visual scene produced by relative motion between observer and scene
Objects closer to observer will move more rapidly
Objects in focus of expansion wont move (closer to focus of expansion = wehre we headed to = move slower)
Describe binocular cues - oculomotor depth cues = vergence
Eye convergence for near objects - move eyes inwards
Divergence for distant ones - move eyes outwards
Describe binocular cues - oculomotor depth cues = Accomdoation
Ciliary msucles contract and adjust lends shape for focu
Nervous system keeps track of degree of accommodation and uses it to infer depth
Accommodation = not binocular cue
Describe binocular cues - oculomotor depth cues = Absolute depth perception
Unlike pictorial cues
Both fo these allows use to tell wehre object located better = accommodation and vergence
Describe binocular cues - oculomotor depth cues = Vergence limit
Reliable up to 2m
Perceive degree of divergence and use as depth cue
Describe binocular cues - oculomotor depth cues = Accommodation limit
Active up to ~6m (beyond = optical infinity)
Cannot tell difference in distance of objects beyond 6m
Describe stereopsis - binocular cues
Rich impression fo depth we get from seeing with both eyes
Relies on binocular disparity
Describe binocular disparity
Slight difference in images perceived by each eye due to their horizontal separation
Which brain uses to calculate depth and percieve a 3d view of environment
2 eyes do not receive same image = helps infer depth
Describe Vieth muller circle - gen
Imaginary geometric circle in visual space
Passes through point of fixation and optical centres of both eyes
Represents theoretical locations fo objects that produce corresponding retinal points in both eyes = result in no binocular disparity = appear at same depth as fixation point
Concept used to explain basis of binocular vision and depth perception
Describe Vieth muller circle - explain
Q = on circle established by fixation point
So Q = projected on corresponding points
Inversely = red point lands on disparate points on left and right eyes retina
= green point is perceived as a single image, but red = double
Corresponding/non corresponding retinal points = helps infer depth of object = distance
Describe Vieth muller circle - image
R = not projected to corresponding retinal points bc not on circle
Fovea = same point
Q = always to left of fovea = same location on both eyes
But r is not
What is horopter
Spatial region in visual space wehre objects project images onto corresponding retinal points in both eyes
Objects on horopter appear to be at same depth as point of fixation and are perceived without binocular disparity
Horopter cna take diff shapes depending on viewing conditions - like distance of fixation
Includes vieth muller circle under idealized conditions
What is panums fusional area
Region in visual space around horopter where slight binocular disparities can still be fused by brain to produce single unified perception of depth
Objects within area = seen as single despite differences in images projected onto retinas of each eye
Objects outside panums fusional area = may appear double - diplopia
Area critical for maintaining binocular vision and depth perception
Fused into one unitary percept - provides rich depth perception
Describe image of steropsis - horpter and fusional area
Panum fusional area = area where vision capable fo integrating input from left and right visual fields despite binocular disparity
Horopter = space wehre objects fall on corresponding retinal points - Oval shape, bc eyes not fully sphere
Name 2 kinds of disparities
Crossed disparity
Uncrossed disparity
Crossed disparity - definition
Occurs when object closer to you than point of fixation
In front of horopter
Crossed disparity - Explanation
Image of object falls on outer (temporal) side of retina in both eyes
From left eyes view = located to right of fixation point
From rigth eyes view = located to left of fixation point
Crossed disparity - Result in perception
Indicates object in front of fixation point and closer to you in 3d space
Uncrossed disparity - defintion
When object father away from you than point of fixation = past horopter
Uncrossed disparity - Explanation
Image of object falls on nasal side of retina in both eyes
From left eye view = located to left of fixation point
From right eye view = located to right of fixation point
Uncrossed disparity - Result in perception
Brain interprets as indicating object behind fixation point = father away in 3d space
Describe v1 binocular neurons
Different binocular neurons in v1 encode all categories of retinal disparity (zero, crossed - near, uncrossed - far)
Neurons allow perceiving depth within panums fusional area
Neurons tuned to diff degrees of disparity
Other columns that receive input form both eyes = allow stereopsis
Describe v1 binocular neurons = graph
Zero disparity = tuned zero , no differences, fire if no disparity, more neurons tuned to 0 disparity = when on horopter
Tuned near = bit of crossed disparity
Number of neurons decrease with increasing disparity = point where so much disparity that image cannot be resolved- gradual= perception of depth decreases as get away from horopter
What is stereoscope
Device for presenting one image to one eye and another image to other eye
Popular in 1900s
Many children in modern days have view master = also a stereoscope
Get sense of depth that comes from binocular rivalry
How do movies look 3d
Each eye must receive a slightly different view of teh scene - just like in real life
Early method = anaglyphic glasses = red lens and blue lends = colour filters
Current methods= use polarized light and polarizing glasss to ensure that each eye sees a slightly different image= use polarity/ orienttaion of light wave
Describe correspondence problem
Visual system has to match images from left and rigth eyes
In binocular vision - problem of figuring our which bit of image in left eye should be matched with which bit in right eye
Like for objects not in panums= projected to non corresponding retinal points so how does system know to connect points or that it is same object
Do we need to recognize objects to have stereopsis? Explain 2 hypotheses
1 = r and l eye —> each do object recognition then —> binocular integration —> stereopsis
2 = right and a left eye —> both do binocular integration - automatically before object recognition —> BOTH stereopsis —> both object recognition
Describe what stereogram proves
Object embedded in random dot stereogram = can only see object with both eyes. =
Proves hypothesis 2
Both eyes need to recognize object together
What is random Dot stereogram
Visual pattern consisting of 2 images made up of random dots
When viewed with both eyes - using stereoscopic vision - the 2 images create illusion of 3d shape or object floating in space
Describe generation of random dot stereogram
2 similar random dot patterns created
Section of one pattern shifted slightly horizontal - left or right - relative to corresponding section in other pattern
Describe disparity of random dot stereogram
Horizontal shift creates binocular disparity when the 2 images viewed together
Disparity mimics differences in the images seen by the left and right eyes in real world 3d viewing
Describe perception of random dot stereogram
When each eye views one image - using a stereoscope or by crossing/uncrossing eyes = brain fuses 2 patterns
Shifted section appears as a 3d object or surface due to the brains interpretation of disparity
Describe crossed disparity stereogram
Start with random grey shoes - subset dots that are different = shift positions for right eye = moves left
Can play with this = object will be perceived as closer
Describe uncrossed disparity stereogram
Subset dots move to rigth for right eyes view
Can play with this = object will be perceived as farther
What is binocular rivalry
Competition between the 2 eyes for control of visual perception
Evident when complete diff stimuli presented to 2 eyes
Why do we not experience diploid for objects outside panum fusion area
Bc visual system represses image coming from one eye - prevents you from seeing double
Randomly switches every few seconds