Lec 6- Flashcards
In general as we move up the hierarchy what happens to receptive fields retinotopic maps.
Unlike p and m cells do cells in higher visual areas respond to linear sum so spots of light?
And what is the current thinking. And just explains diff parts so v1, and as we go higher specialised
In general, as we move up the hierarchy, receptive fields become larger, retinotopic maps become less precise or even absent, and the type of stimulus that visual neurons prefer becomes more elaborate.
In particular, unlike P-cells and simple cells, many cells in higher visual areas do not respond to the linear sum of (spots of) light presented within their receptive fields but respond strongly only when stimulated with specific stimulus configurations.
As you can imagine, classifying these cells is a very difficult thing to do because it is difficult to discover what the appropriate stimuli are.
Thus, current thinking is that V1 is a preliminary area, where information about all visual sub-modalities (luminance, colour, movement, stereopsis) is represented. From here, the cortex becomes more specialised, with separate ‘streams’ projecting to different areas, each of which specialises in a different type of information (e.g. colour in V4, motion in MT).
V1 revision AREA 17
And what is blindsight and what does this mean for seeing and v1
Evidence from single cell recordings, optical imaging and various anatomical staining and labelling techniques indicate that: 1) V1 is retinotopically mapped, 2) has columnar organisations for both orientation and size (spatial frequency), 3) has cells selective for wavelength (blob-cells interspersed amongst interblob cells), 4) has some cells selective for direction of motion, 5) is organised for ocular dominance, 6) has cells classified as simple, complex and end-stopped (hypercomplex). Cytochrome oxidase stains for blobs in v1 and and thin stripes in v2.
5 Blindsight
Individual case studies of patients with brain damage to V1 suggest that intact V1 is required for conscious perception. Because such patients claim that they cannot see, but usually have intact retina and LGN, they are sometimes described as ‘cortically blind’.so consciously cant see but subconsciously they respond
However, this does not mean that V1 is necessarily the site of ‘seeing’ (indeed, growing evidence suggests it is not), but that for conscious percepts, visual information must first pass through V1 and perhaps then on to further areas.
But not all visual information must pass through V1 to get into visual cortex. There are minor routes to other visual parts of the brain that project directly from sub-cortical areas. These other routes are probably the substrate for the ‘seeing without visual awareness’ phenomenon known as blindsight. These pxs have no v1 no conscious vision but are good at discrimination tasks so can tell you if light flashed was upper or lower eventhough they insist they cant see it. So v1 involved in processing visual data but not necessarily involved in seeing bc they tech can see subconsciously.
V2- where does this area receive most input from so what does this mean for retinotopic mapping. Rfs vs v1 and what are most of the cells.
Illusory contours and an example of an illusion kanizsa
This area is the second visual area in the hierarchy (also called area 18 or pre striate cortex) and receives the most substantial part of its input from V1. Its point-to-point connection with V1 means that, like V1, it is retinotopically mapped, though with slightly less precision than V1. Receptive fields are typically larger than those found in V1, and most cells are complex cells.
Many cells in v2 respond to illusory contours.
In the kanizsa triangle a white upward pointing triangle is seen superimposed on background elements yet there is no luminance cue for the boundary of the figure so the contours are illusory. There is also a weak brightness illusion here so the triangle in the foreground appear lighter than the background. In the vertical ones you can see that the colours are all the same white as now the illusion of the vertical bar is abolished when gaps in inducing elements are closed so it isnt whiter than white.
This is bc the brain tends to perceive incomplete figures as complete or closed and prefers to perceive continuous and smooth contours rather than fragmented or discontinuous ones. Fills in the missing parts.
Cytochrome oxidase strips
What do v2 stain for
What are cells witin thin and thick stripes important for
Where do they all receive input from
What are interstripes important for
Like V1, V2 also stains for cytochrome oxidase (CO). In V2, CO rich areas form ‘thick stripes’ and ‘thin stripes’ with intermediate unstained ‘interstripes’.
Cells within thin stripes appear to be important for colour and receive projections from the blobs in V1
whereas cells in thick stripes appear to be important for stereo vision (see later lecture).
The interstripes receive input from the interblob regions of V1 and are thought to be important for spatial form.
Visual streams= DORSAL/ VENTRAL PATHWAY= where and why
What and where stream
Evidence- are these two streams linked or segregated
Functional aspects of these streams how do we know that where is__ and what is__
pair of parallel processing streams through visual cortex. One of these (the ‘what’ steam) projects to the temporal lobe (IT) and deals with object structure. The other (the ‘where’ stream) projects to the parietal lobe (e.g. Area 7a, which also receives input from the pulvinar) and deals with object location and spatial relations.
There is good evidence for segregation between these two streams. Baizer et al (1991) injected large amounts of two different retrograde tracers1 into monkey brains. One tracer was injected into part of the parietal lobe and the other into part of the temporal lobe. Upon inspection of earlier visual areas, they found almost no neurons that contained both tracers.
Evidence for the functional aspects of these streams comes from clinical case studies of damage to the relevant brain regions. For example, damage to the parietal lobe (the ‘where’ system) can cause patients to have problems orienting towards or reaching for objects (a condition known as optic ataxia) but does not impair performance in object discrimination and recognition tasks.
On the other hand, damage to the temporal lobe (the ‘what’ stream) can result in patients having problems with form perception and recognition, and visual memory, but spares performance on visuospatial tasks. So parietal= tasks temporal= what is it.
Once again ab the dorsal/ vernal
What are they where do they project to
Damage here
Where what pathway
Dorsal= where so parietal lobe. Ventral= what so temporal lobe
Dorsal- Projects from v1 to v2 to mt= v5. And extends to area 7a
Ventral= projects from V1 to V2 to V4 and to IT
Parietal=dorsal
Damage to parietal lobe so damage to some of the dorsal pathway can result in optic ataxia like in last flashcard. Unable to grasp orientated postcard but able to report its orientation bc temporal lobe still works. See world but cant act upon the world. The dorsal stream guides how we do actions.
Damage to ventral= temporal
Visual agnosia. Inability to recognise objects
Or prospagnosia
V1 vs IT
And damage to the IT or ventral stream.
IT and faces, how are faces encoded
V1 cortical amps. Doesn’t know anything about top down processing it looks at diff edges. Orientation distributions it just encodes info about local orientation or features. Cant tell you where parts of that flower are.
IT- don’t have a retinotopic map anymore, this tells us more about the structural components so the edge is there or the stem is there etc. columnar organisation of IT= elaborate cells. region of IT responded only to faces in an experiment. Faces encoded by distribution of activity across population of cells and it is that pattern of activity that is specific to one individual face so 100 cells are needed for one individual.
Prosopagnosia typically arises from damage to regions within the ventral stream of the brain, in the inferior temporal cortex (IT). This condition is characterized by an inability to recognize faces, including those of familiar individuals, despite intact visual perception in other areas. not solely due to damage in the IT cortex; other areas within the ventral stream. Prosopagnosia can result from various factors, including brain injury, stroke, or neurodevelopmental conditions.
-within class identifications cant tell what it is eg type of frog or rope of car. Some people may recognise smth better than others so need fair comparison. Some people cant identify their cows and birds as well as they used to so maybe this over faces.
Area 7a
This area is located in the posterior part of the parietal lobe and receives information about head and body position as well as visual input.
RFs are large and often span both visual hemifields. So called gaze-locked cells have been reported in this area. The key requirement for these cells is that the animal must be gazing in a particular direction. If the direction of gaze is inappropriate for a cell, then no RF can be found. Thus, these cells will respond to objects only when an animal is gazing in the ‘right’ direction, with different cells preferring different directions.
V3
Is it retin mapped, describe rfs.
What are cells in v3 selective for
Is it involved in colour processing and why
V3 is retinotopically mapped (though less precise than V2) and receptive fields are larger than those typical in V1 and V2. Cells found that are selective for isoluminant contours.
Cells in V3 are markedly selective for orientation and motion, receiving input via the transient magnocellular pathways from V1 and V2. In V3, the form is defined by luminance variations that just happen to be moving. Found pattern selective cells in v3 similar to those in MT.
Early reports suggested that V3 was not involved in processing colour, though this view has since been challenged. For example, Gegenfurtner found cells (similar to those found in V4) that responded to so called ‘isoluminant’ stimuli containing chromatic modulation alone. Such stimuli might contain, say, red & green stripes, but if photographed using black and white film would appear featureless so uniform grey. If cells responded to these kinds of stimuli they must be seeing the chromatic modulation which means they must have wavelength selective properties.
So isolumiannt and boundary defined by colour but v3 can respond as luminance is the same it differed only in colour.
MT- middle temporal cortex
What are these sensitive to
Rfs
What does damage here lead to
What is mt also involved in
Unlike v1 difference
All of the cells in this area are sensitive to direction of motion.
They have fairly large receptive fields and are not concerned with the colour of the stimulus.
Unlike V1, the cells in this region have a columnar organisation for motion. Some cells in this area appear to be sensitive to relative motions and could be used in computing form and depth from motion. Furthermore, some of these same cells are also sensitive to stereo disparity, suggesting that depth cues from motion and stereopsis may be combined here.
MT is probably also involved in grouping together individual motion elements to achieve a representation of object motion. In v1 only orientation selective So, in V1 there are no ‘vertical’ cells that will respond to the compound stimulus described above. However, in MT, some of the cells that prefer vertical component motion will also respond to the compound stimulus in our example. These cells are called ‘pattern’ cells and correspond more closely with our actual perceptual experience of motion than do the motion selective cells in V1.
Clinical lesions in MT lead to a condition known as akinetopsia (motion blindness). In this very rare condition, patients appear to have no visual representation of motion. They can recognize objects and understand spatial layout but cannot see things move.
MST- medial superior temporal cortex vs mt
MT is located in the medial temporal lobe of the brain.
Mt also known as V5, is primarily involved in processing visual motion. It is sensitive to the direction and speed of moving visual stimuli. Neurons in MT respond preferentially to motion in specific directions and speeds, making it crucial for motion perception and tracking moving objects.
MST is located adjacent to MT, also in the medial temporal lobe.
MST is involved in processing complex motion information, including the perception of optic flow (the apparent motion of visual elements in a scene as an observer moves through it) and the perception of three-dimensional motion. MST integrates information from MT and other visual areas to provide a more comprehensive understanding of visual motion and spatial orientation. Some cells selective for complex 2d motion eg: expansions, contractions, rotations, spirals.
MT is primarily responsible for processing basic visual motion, such as direction and speed, while MST is involved in processing more complex aspects of motion perception, such as optic flow and three-dimensional motion.
- Specificity: Neurons in MT are highly tuned to specific directions and speeds of motion, whereas MST neurons are more broadly tuned and respond to a wider range of motion stimuli, including complex patterns of motion.
V6 parieto occipital visual area
V6, also known as the dorsal medial area (DM), is a region within the visual cortex of the brain. near the junction of the parietal and occipital lobes.
V6 is part of the dorsal visual stream, which is involved in processing spatial information and guiding actions based on visual input. Specifically,
Partial occipital visual area= PO
Gaze locked cells
V6 is thought to play a role in processing motion and integrating visual information with other sensory inputs to aid in spatial perception and navigation. While not as extensively studied as some other areas of the visual cortex, V6 contributes to our understanding of how the brain processes visual information to navigate and interact with the environment.
What have some cells in v6 been called
Please explain this properly below
Real position cells
Real position cells are neurons in the brain’s visual cortex, specifically in the V6 region. Unlike other neurons whose receptive fields move with eye movements, real position cells maintain their receptive fields fixed in the real world. This means they respond to objects in a consistent position relative to the observer’s body, rather than relative to the position of the object on the retina. They help provide information about the stability of objects in the environment as the observer changes gaze direction.
V6 responsible for representing space in egocentric coordinates as opposed to far less useful retinal coordinates of v1 and v2
V4
Retinal map and columnar organisations or colour
Non selectivity for lgiht wavelength
Damage to v4
Specific visual loss
Chromatopsia from co posioning why may cv remain intact
Why are blob cells in v1 more resistant to damage from carbon monoxide poisoning
Is v4 just colour processing what else is colour processing
- Retinal Map and Columnar Organization for Color: This refers to the structure of the visual cortex, where cells are organized in columns and process specific aspects of visual information. In V4, there’s a particular arrangement for color perception.
- Non-selectivity for Light Wavelength: Some cells in V4 aren’t picky about the color of light hitting the retina, which is unusual because one might expect them to be highly specific.
- Achromatopsia from V4 Lesions: When V4 is damaged, patients experience achromatopsia, where they see the world only in shades of grey, indicating the critical role of V4 in color vision.
- Specific Visual Loss: Unlike damage to other visual areas like V1, which would result in complete blindness, lesions in V4 lead to selective visual deficits, highlighting the specificity of each visual area’s function.
- Chromatopsia from Carbon Monoxide Poisoning: In some cases of carbon monoxide poisoning, color vision remains intact, but patients struggle with fine detail and object recognition. This suggests damage to specific regions like the interblob areas in V1, while sparing other regions like the blob areas.
- Metabolic Activity and Blood Supply: Blob regions in V1, which are metabolically active and receive a larger blood supply, may be more resistant to damage from carbon monoxide poisoning compared to other regions.
- V2 and V4 Connectivity: Thin-stripe regions in V2, involved in color processing, project to V4. However, V4 is not solely dedicated to color; it also processes complex spatial patterns and forms.
Colour constancy
Color constancy is the ability to perceive the color of an object as relatively constant despite changes in lighting conditions. Essentially, it’s the phenomenon where the perceived color of an object remains the same under different illuminations. For example, a red apple will still appear red whether it’s viewed under natural sunlight or artificial indoor lighting, even though the wavelengths of light reflecting off the apple may vary. This perceptual stability allows us to recognize and identify objects accurately in varying environments.
Wavelengths that reach our eye are dependent on: spectral composition of illumination, and spectral reflective properties for the surface. Mainly interested in second. Human vision appears to be able to disregard the illumination and deliver results relating to just the spectral reflect acne of properties of surfaces.
A bit more on colour constancy mistakes and tungsten light
Spectral composition of tungsten lighting is much more biased towards the long wavelengths so red than its natural daylight.
Vision can make small mistakes due to illuminant eg shop artificial lighting . Maybe colour constancy is bc of top down processing so our brain telling us bananas are yellow. On other hand it is the average estimate of the wavelength over the visual scene if constant then cancel out. Estimate illuminant and analyse object.
Cells in v4 large rfs 30 times larger v1= colour constancy effect.
The ability to discount illuminant= colour constancy.
How is colour constancy achieved
Top down
Bottom up= comparison of wavelengths across large areas of retina and uses this to estimate spectral composition of illuminant and then discounts it by subtraction.
Some cells in v4 have this property large rfs pulling info from large area but delivers it about a small patch.
Destroying colour constancy gets rid of that averaging so you see it insolation and you see the true wavelength. So wavelength we see is a combination of averaged illuminant and the wavelength of the object. The wavelength we need to block the averaging and look in isolation.
Why is colour constancy not perfect
Cloth under artificial shop light greens and blues
Extreme or unusual lighting eg nightclubs
We have it to deliver a constancy about the perception of the world so surfaces look same colour irrespective of the illumination. But visual system does have to make some assumptions ab the world so can be misled.
Diff areas and names
V1= area 17
V2= area 18
V5= mt
Parietal lobe= 7a
It= ventral stream. V6 is Dorsal stream or po
What is simultanagnosia
10 distinct response levels for 5 visual neurons how many cells code for how many different faces
Condition where people are unable to see more than one object amongst a group of two or more closely placed objects.
Each neuron has 10 distinct response levels
5 so each one multiples it by 10.
10 times 10 times 10 times 10 times 10= 100 000 different faces
As we move up the cortical visual hierarchy which is false about visual neurons
And which are true
Their retinotopic maps become less precise- true as go from highly ordered in v1 to less as higher complex processing
Their rfs become larger true- as integrate more info from larger regions of vf
Their stimulus selectivities become more difficult to characterise- complex and varied responses more abstract cant really characterise
Their stimulus sens= more refined- trye as more selective
False- it becomes possible to determine their responses to compelx stimuli by knowing their responses to small spots of light-
False as insight but cant get the full complex response from simples stimulus shown.
Ungurlieder and mishkin proposed number of major visual streams in visual cortex is
Milner and Goodall proposed which dichotomy for primate vision
2= dorsal and ventral
Action/ perception which is a reworking of the what/ where dichotomy.
Zeki argued that what and where doesnt provide good insight into function of brain.
Action vs perception
Milner and Goodale suggested radical re interpretation of cortical visual pathways. Where should be relabelled action and what should be labelled perception. Enabling organism to rapidly interact w environment.
Perception- what it needs to do so ooh hole dont go in there
Action- your legs moving away.
Visual illusion called titchener circles
Evidence for disassociation between action and perception
Apparent different sizes but they are the same left and right.
When a circle is around smaller circles it looks bigger and when circle around bigger circles it looks smaller.
Spatial contrast effect and the surrounding ones make middle ones change.
Perceptual effect so the ventral system. Perceptual guides illusion and action system is one that guides actions.
Ventral= what= action
Dorsal=what=perception
Which of the following is true about the two pathway model of vision beyond v1
Does dorsal stream make rapid decisions
The ventral stream is thought to be involved in the perception of motion- no this is dorsal system where motion
The ventral system is thought to be involved in directing visually guided actions on the world- no ventral is identification and recognition of objects
The dorsal system is linked to inability recognise faces- ventral
The main distinction between dorsal and ventral stream as is that ventral streams is concerned with stationary objects but dorsal is concerned with moving objects- dorsal is motion= where. Ventral is object identification and recognition.
True- the dorsal stream is thought to be involved in rapid decision making processes that do not necessarily link to conscious visual perceptions.
Which is generally false
Different parts of primary visual cortex separated by few mm or so encode image data from diff regions of retina= true
Different parts of it separated by few mm or so encode diff types of image features true
Neighbouring columns of cells in mt are selective for diff directions of motion= true
Neighbouring columns of cells in v1 are selective for diff orientations of luminance contour= true
Different cells in v4 selective for each one of the hundred of diff colours tht we can see-= false
