Lec 3/ TB Ch 3 Flashcards
1
Q
- Spatial vision
- Gutenberg printing press analogy
- Independent component analysis (ICA):
A
- Spatial vision: refers to our ability to resolve or discriminate spatially defined features.
- Gutenberg’s printing press
- Language is represented as text; text can be pieced together from letters
- each letter has it’s own fx
- visual images can be represented by visual elements combined together
- Neurons detect the visual elements; brain combing info to form an image of the world
- Language is represented as text; text can be pieced together from letters
- How can we identify the elements of visual images?
- Independent component analysis (ICA): reveals independent components in a signal (ex. audio of 2 ppl speaking; tiny parts in image) factors (components) that underlie signals.
- When you have the tiny parts, you can create infinite # of images
- Our visual system use concept to represent real life
- neurons collect the components of reality
2
Q
- 3 main types of gratings
- 4 characteristics of gratings
- 2 units for f
*
- 2 units for f
A
- 3 main types of gratings
- Rectangular grating
- Sine wave (sinusoidal) grating
- GaborsL Sine grating w/ bull’s eye
- These are building blocks of vision (& audition)
- 3 (+1) characteristics of gratings
- Frequency
- Cycles per second = Hz
- Cycle = pattern repeat
- Cycles per visual degree = cpd
- Stretch out your hand, width = 10 visual degree
- Cycles per second = Hz
- Amplitude
- Phase
- Blue and red wave
- Red wave is shifted
- Blue and red wave
- (Orientation) – later
- Y-axis can be many things: sound amplitude, luminance, visual contrast
*
- Y-axis can be many things: sound amplitude, luminance, visual contrast
3
Q
- spatial f
- cycles per degree
- diff b/w CVA vs fournier transform
- Fournier transform
- Sound example
- Image example
- how do we see high spatial f; what do we see?
- how do we see low spatial f; what do we see?
- What parts of eyes detect high vs low spatial f?
- Describe Lincoln illusion
- Describe Mona Lisa Illusion
A
- Spatial Frequency: The number of cycles of a grating per unit of visual angle (usually specified in degrees)
- Cycles per degree: The number of dark and bright bars per 1 degree of visual angle
- CVA vs Fourier
- CVS: dunno what are the components from the start
- Fourier: we assume the basic components are gratings
- Fourier transform: an operation that breaks down a function/an image into sine waves of different frequencies.
- Ex. Music can be decomposed into sine waves with different temporal frequencies.
- Ex. Visual images
- Close up → we see High f: details
- far away → we see Low f: gist
- Center fovea: high f
- Periphery: low f
- Lincoln illusion
- close up → activate fovea → only see high F info (squares); interference w/ low f
- far away → activate periphery → see low f info (gist of Lincoln)
- Mona Lisa’s Secret
- stare at face (use fovea/ fine details) → no smile
- stare ar her cheek (you look away; use periphery/ gist) → smiles
- Sfumata: layering thin pains
- Lower layers = blur version of Mona Lisa smiling
- Upper layers = not smiling
4
Q
- how does visual system code images
- What happens if stripes are too thin?
- Contrast
- Acuity
- 2 ways to measure acuity
- min visual angle?
- Cycle in visual acuity
- 3 factors that limit acuity
- convergence: fovea vs periphery photoreceptor
- Why is central vision slower than peripheral vision?
A
Visual acuity
- The visual system codes images in terms of oriented stripes.
- Limits:
- narrowness of the stripes = visual acuity
- too thin -> can’t see diff -> gray patch
- Contrast: The difference in illumination between a figure and its background
- narrowness of the stripes = visual acuity
- Acuity: The smallest spatial detail that can be resolved w/ 100% contrast
- Measuring visual acuity: - diff ways
- 1 Ophthalmologists use distance (20/20, 20/30, 20/15)
- What you see in x ft/ what avg person see this at ? ft
- 2 Vision scientists use the smallest visual angle of a cycle of a grating
- Minimum visual angle of a cycle: 0.017 deg
- Around the size of indiv photoreceptors
- Minimum visual angle of a cycle: 0.017 deg
- Cycle: one repetition of a black and white stripe in a grating
- 3 factors that limit acuity
- 1 spacing of photoreceptors (Can’t further cram them together)
- 2 Each photoreceptor codes luminance info of a patch of the visual field; the output of 1 photoreceptor = avg of whatever stimulated by visual field
- 3: convergence (e.g., multiple photoreceptors projecting onto 1 bipolar cell etc.)
- In fovea, 1 photoreceptor is connected to 2 bipolar -> connect to 1 ganglion
- IOW: no info loss
- Periphery: many photoreceptors packed together less tight + connected to bipolar -> many bipolar connect to 1 ganglion
- There’s many convergence -> info loss → lower acuity
- In fovea, 1 photoreceptor is connected to 2 bipolar -> connect to 1 ganglion
- 1 Ophthalmologists use distance (20/20, 20/30, 20/15)
- Peripherical cones respond 2x faster to light than foveal cones (30 vs 60 ms)
- Foveal cones hv longer axons than peripheral cones to allow dense packing in the central fovea
- The longer axons transmit slow signals better than fast ones
- The slow response allow foveal cones to increase their reliability by integrating their input over a long time
*
5
Q
- Left vs right image (what you can see)
- Ailiasing
A
- Left images: each photoceptor has 1 output (light vs dark) -> mind reconstruct reality
- Right: gratings hv higher spatial f -> each photoreceptor sees light and dark(entire cycle is on a single cone); gives only 1 output -> grey
- IOW: Aliasing refers to artifacts that result when the signal reconstructed from samples is different from the original continuous signal.
- Produce diff textures (ex castle image)
- IOW: Aliasing refers to artifacts that result when the signal reconstructed from samples is different from the original continuous signal.
6
Q
- Snellen test
- Otto Schade - study method
- contrast sensitivity function (CSF)
- What is it based on?
- Sensitivity?
- What cycle/degree to we have best vision?
- Left = low spatial f (levelling off) - 2 reasons?
A
- Snellen test – check vision at eye dr
- X
- Additional factor: contrast
- •Otto Schade showed people gratings with different SFs (spatial f) and had them adjust the contrast until the gratings were at the point where they disappeared.
- • The contrast sensitivity function (CSF) describes the relationship between contrast and spatial frequencies.
- High spatial f+ high contrast = can see clearly; vv
- Red curve: you see it when you are close; green = far away
- The contrast sensitivity function CSF is based on many individually measured (absolute) contrast thresholds (smallest amount of contrast to detect a pattern)
- Sensitivity: inverse of a threshold.
- MP: best vision (peak) at 7 cycle/degree
- At 60 cycles per degree = can’t see shit (on the right)
- IOW: 0.017 def?
- Left = low spatial f (levelling off) – 2 reasons
- 1 Nothing interesting
- 2 Low spatial f have lots of energy and contrast -> don’t need to be so sensitive
*
- At 60 cycles per degree = can’t see shit (on the right)
- MP: best vision (peak) at 7 cycle/degree
7
Q
- 2 main features of visual corte
- Why can’t left eye see visual field 9?
- Solution
- 4 step process how info travel from ganglion cells?
A
- 2 main features of visual cortex: topography and magnification
- Topography
- Ex. women’s right eyebrow (her right; #3,4) appears on position 3,4 in visual field and striate cortex
- Magnification: info is scaled
- The amount of cortical area devoted to a specific region in visual field
- Fovea is represented on #5 in retina
- Cortical magnification:
- Objects on fovea are processed by a large area in striate cortex (magnified)
- Objects on periphery -> tiny portion
- Each eye see both visual fields, there’s a lot of overlap
- Left eye can’t see all the right visual field (#9) (b/c nose is blocking)
- Not an issue as you can turn your head
- Left eye can’t see all the right visual field (#9) (b/c nose is blocking)
- Each eye see both visual fields, there’s a lot of overlap
- Process
- 1 Ganglion cells near nasal
- 2 Partially crossover -> LGN
- The lateral ganglion cells (don’t crossover)
- 3 Sorting of visual field is AT chiasm
- 4 After chiasm
- RH = left visual field; vv
- Topography
8
Q
- lesion on right eye/optic nerve → name?
- lesion on optic chasm → name?
- lesion after optic chiasm
- 3 types
- damage of rH → damage on ?? visual field
A
Functional deficits of the visual system
- Lesions in lower visual areas cause (retinotopic) visual field defects???
- If there’s lesion on optic nerve (blue cross) = right eye
- Right visual field gone/ blind
- Damage optic chiasm (ex. tumor)
- Heteronymous hemianopia
- Hemi = half
- An = lose
- Opia = sight
- Hetero = different side
- Left eye left side gone; opp
- Homonymous hemianopia (half gone)
- Lesion AFTER optic chiasm
- Homonymous hemianopia (half gone)
- Quadrantanopia (quarter gone)
- Scotoma (little part)
- NOTE: damages on the right H -> can’t see left visual field
- Opp is true
- Blindspot is natural scotoma
- NOTE: there’s a dent in the centre; fovea and macula are spared
- Often the foveal part of the visual field is spared, large cortical representation for fovea; need to destroy the whole thing to destroy fovea (unlikely)
- NOTE: there’s a dent in the centre; fovea and macula are spared
- Summary
- Heteronymous: different visual field defects for the two eyes
- Homonymous: same defects for the two eyes
- Hemianopia: half the visual field is blind
- Quadrantanopia: a quarter of the visual field is blind
- Scotoma: a small visual field defect
9
Q
- what does Mexican hat fx tell us?
- 4 properties ganglion cells are sensitive to?
- Response activity when there’s low vs godilock vs high f?
- reason?
- tiny vs large receptive field
*
- Response activity when there’s low vs godilock vs high f?
A
- (On-centre) ganglion cells like spots of light
- Can visual this in 3D plot = Mexican hat fx
- Color = activation level
- X, y = where on visual field
- Can visual this in 3D plot = Mexican hat fx
- 1 Retinal ganglion cells “are tuned” to certain spatial frequencies
- Low f = weak response
- b/c some light is in peripheral = inhibit
- (NOTE: each line = AP)
- Response peaks at Goldilocks
- Low f = weak response
- 2 Depends on size of receptive field
- Tiny = fine tuned for high f
- Big = tuned for low f
- 3 Retinal ganglion also tuned to certain phases
- 0 light in (+), no light in (-) -> activated
- 90 activation and inhibition are in equal amounts -> no response
- 180 light are (-) -> lots of inhibition
- 270 same as 180
- IOW: position of grating on visual field -> diff activation of ganglion cell -> gives out diff info
- 2 Depends on size of receptive field
- 4 Ganglion cells care about spatial frequency; don’t care about orientation (receptive field = circle)
*
10
Q
- left LGN → which visual field?
- LGN layers
- magno
- layer #
- input?
- What does it detect
- Parvo
- layer #
- input
- What does it detect
- Middle layer name?
- What does it detect
- magno
- Layer # & ?? lateral
- 2 types of info from top down processing
- Topographical mapping
A
From the eye to the brain
The lateral geniculate nucleus (LGN)
- Contains 6 layers, each contains a map of the visual field
- Left LGN -> right visual field; vv
- Layers 1 & 2: magnocellular system (high luminance sensitivity, motion/flicker).
- High lum: low l -> can see
- Input from M ganglion cell
- Respond to large, fast-moving objects
- Layers 3-6: parvocellular system (high spatial frequency, colour).
- Smaller receptive field (tinier cells) -> high spatial f
- Input from P ganglion cell
- Respond to details of stationary targets (suggest visual system split image input by types)
- Interlaminar system (b/w the 2 LGN): koniocellular system (color)
- x
- Both LGNs receive input from both eyes (green = right eye)
- Layers 2, 3 & 5: ipsilateral eye (same side)
- Layers 1, 4 & 6: contralateral eye
- Receive important input from cortex (top down)
- Attention
- Arousal
- Ipsilateral: Referring to the same side of the body (or brain)
- Contralateral: Referring to the opposite side of the body (or brain)
- Each LGN layer has a [EL1] map of a complete half of visual field
- Objects in the right visual field (objects to the right of where our gaze is) are mapped onto diff layers of left LGN
- IOW: RS of the world falls on LS of retina
- Topographical mapping: neural basis of where things are in space
- IOW: RS of the world falls on LS of retina
[EL1]LS of retina goes to LGN
11
Q
- where does LGN project to?
- other names (3 other names)
- how many cells?
- 2 main features of striate cortex?
A
Striate cortex
- LGN projects to the Striate cortex, aka
- primary visual cortex (LGN sends visual info here 1st)
- V1
- Brodmann area 17
- Major transformation of visual information takes place in striate cortex
- It has about 200 million cells!
- Two important features of striate cortex:
- Topographical (retinotopic) mapping: orderly mapping of the visual field onto a neural structure (LGN, V1, …) such that neighbouring points in the visual field are projected onto neighbouring patches of neural tissue.
- Ex. Fovea and sides of fovea are located close together in V1
- Ex. Auditory cortex – topographic mapping by f
- 2 Dramatic scaling of information from different parts of visual field: cortical magnification
- Ex. Fovea is represented way bigger than the periphery
- Topographical (retinotopic) mapping: orderly mapping of the visual field onto a neural structure (LGN, V1, …) such that neighbouring points in the visual field are projected onto neighbouring patches of neural tissue.
- Two important features of striate cortex:
- X
- Retinotopy (in macaque): A flickering stimulus (left) and its retinotopic representation in layer 4C of V1 (right), revealed through CO staining. Reproduced from Tootell et al (1988a). Note the magnification.
- Ex. F and 1 on LS vs F an 1 on RS
The mapping on the Fovea and 1 is projected into a larger area on RS (Va
12
Q
- Receptive Fields in Striate Cortex
- Orientation tuning
- Hubel & Wiesel - study with the cat procedure
- Describe process in red and. blue circles
- Cortical cells respond will to 3 things?
- 4 other things they respond to
A
Properties of the Striate Cortex
- Functional Properties of the Striate Cortex
- Receptive Fields in the Striate Cortex
- Columns and Hypercolumns
- Selective Adaptation: The Psychologist’s Electrode
- X
- Receptive Fields in Striate Cortex
- Selective responsiveness to orientation:
- Orientation tuning: tendency of neurons in striate cortex to respond optimally to certain orientations, and less to others
- Selective responsiveness to orientation:
- – Hubel & Wiesel
- Kuffler -> discovered concentric receptive field of Ganglion cells
- Show images to cuts
- Record neuron activity at V1
- Found that V1 is activated when changing slides
- V1 was not responsive to dots, but lines
- Respond to specific orientation -> orientation tuning
- X
- Orientation tuning function of a cortical cell (firing rate vs degree of orientation)
- How are the circular receptive fields in the LGN transformed into the elongated receptive fields in V1?
- Hubel and Wiesel’s explanation
- Red circles = LGN receptive field
- Blue circles = activated
- A bar of light shining on those fields
- A line of cells are activated in the center field (yellow); their periphery are inhibited (grey)
- Receptive field (yellow) projects to 4 LGN neurons; 4 LGN neurons project to 1 V1 cell
- V1 gets the sum of the V1 center activation, and the inhibition on the periphery
- Other mechanisms: lateral inhibition
- X
- Many cortical cells respond especially well to:
- Bars, Edges, Gratings
- Cells also respond to Moving lines, Direction, End stoppings, One eye (respond to 1 eye more)
13
Q
- visual crowding
- where does it happen
- define
- solution
A
Some Perceptual Consequences of Cortical Magnification
- Major obstacle in periphery is visual crowding
- Deleterious effect of clutter on peripheral object recognition
- Objects are easily identified in isolation may seem jumbled when in presence w/ other stuff
- Crowding impairs discriminating objects, and ability to recog and respond to objects in clutter
- Mitigation: eye movements
14
Q
- 3 things simple cells respond to
- 4 things for complex cells
- End stopping
- where is it found?
- Occular dominance
- LGN vs striate cortex
- Black vs white stripes
A
- Simple cells respond to
- edges or (bright or dark) bars
- of a certain orientation
- in a certain part of the visual field (receptive field).
- Complex cells also have orientation tuning
- Respond more generally than simple cells (phase- insensitive)
- As long as it is in correct orientation -> I am activated/happy
- Larger receptive fields.
- Prefer moving stimuli
- Respond more generally than simple cells (phase- insensitive)
- *** Many simple cell info feed into complex cell
- End Stopping: Process by which cells in the cortex first increase their firing rate as the bar length increases to fill up its receptive field, and then decrease their firing rate as the bar is lengthened further
- Found in both Simple & complex cells
- It may be used to measure length of lines
- Ocular dominance
- Each LGN cell responds to one eye or the other, but never to both, but:
- Each layer is either ipsilateral or contralateral only
- In striate cortex, cells may respond to input from both eyes though they retain a relative dominance of one eye.
- Black eyes -> only respond to left eye (IOW: left eye dominance)
- White stripes –> respond to R eye
*
- Each LGN cell responds to one eye or the other, but never to both, but:
15
Q
- Column
- 3 common property in the same column
- Difference across columns horizontally
- Hypercolumn
- 4 characteristics
- # of cells in ganglion vs striate cortex
- Circle in rainbow image?
- White circle/CO blob
- What is it?
- Method that discovered it?
- Receive input from?
- What does it do?
A
- How are neural response properties organized?
- Column: A vertical arrangement of neurons
- Hubel and Wiesel found the same preferred orientation when testing neurons at different depths (=layers) of V1
- IOW: same orientation, spatial f and dominance throughout
- Column: A vertical arrangement of neurons
- However, they found systematic, progressive change in preferred orientation when testing neurons laterally; all orientations were encountered in a distance of about 0.5 mm
- Horizontal: orientational tuning systematically change
- Hypercolumn: A 1-mm block of striate cortex containing “ all the machinery necessary to look after everything the striate cortex is responsible for, in a certain small part of the visual world” (Hubel, 1982)
- IOW: the whole purple green block
- Take all the columns and bundle them together
- All orientation; spatial f combination
- There’s ocular dominance
- These neurons all receive the same image from the visual field; but each analyze the image in diff ways
- Explains the increase of # cells
- Ganglion 100 million
- Striate cortex: 200 million
- Optical imaging tells a similar story
- Explains the increase of # cells
- Used fMRI in microscope lv
- Diff areas respond to diff orientations
- Circle = Pinwheels of orientation in V1
- aka hypercolumns
- Can detect all diff combinations (orientation tuning, spatial f, L vs R dominance)
- The white circle/ CO blob: neuron has no orientation tuning/spatial (similar to receptive fields of LGN and ganglion cells)
- Vertical cylinders in the layers
- It is a circular receptive field that is receptive to anything
- Vertical cylinders in the layers
- There are Regular array of “CO blobs” in systematic columnar arrangement
- (discovered by using cytochrome oxidase staining technique)
- Receive input from koniocellular system and parvocellular system
- IOW: they process color perception only
*
- (discovered by using cytochrome oxidase staining technique)