perceptions of objects motion and depth

Question 1

what was evolutionary benefit of vision?

Answer 1

Early primates started foraging for reach source of calories: fruits and
nuts that usually grow at the ends of tree branches → require
reaching and grasping.
 Need to identify shapes and 3D positions of object, then move to get
them: object recognition, motion and depth perception.
 1/3 to 1/2 of the primate brain cortex is involved in visual perception.
 Human brain further exploits this specialization: e.g. creating new
tools, reading.

Question 1

what is ventral strream for and what is it comprised of?

Answer 1

Ventral stream: object
identification
Cortical areas: V1 → V2 → V4
→ Lateral Occipital Complex
(LOC) & Infero-temporal cortex
(IT)

Question 2

What is dorsal stream for and what is it comprised of?

Answer 2

Dorsal stream: motion and
position in space.
Cortical areas: V1 → Middle
Temporal (MT or V5) → Medial
Superior Temporal (MST) →
Posterior Parietal Cortex (PPC)

Question 3

How do these ususally work to encode vision and perception of things?

Answer 3

 T**here is a Hierarchical organization: **early visual areas encode image
features, at the later stages features are combined into global
structures.
(V1 → V2 → V4 → Infero-temporal cortex)

 Predictive coding: prediction of the input by the higher order areas
sent to the early visual areas; prediction error signal (mismatch
between predicted and actual input) is sent to the higher order
areas to update representation.
prediction
error signal
input
prediction
error signal

Question 4

How do we group features into objects?

Answer 4

Principle of proximity, closure, similarity, connection and common fate.

Question 5

What do the gestalt principles refer to?

Answer 5

Proximity: seeing all these circles right next to each other enables us to see lines (made of circles)
Common fate: when features are moving together we see them as an object.

Similarity: combinign similar features together to make an object example, red circles in square of blue ones.

Connection: features observed connected together get percieved as one object.

closure:The principle of closure states that when we look at a complex arrangement of visual elements, we tend to look for a single, recognizable pattern. In other words, when you see an image that has missing parts, your brain will fill in the blanks and make a complete image so you can still recognize the pattern.
circle being drawn with broken lines is still percieved as a circle.

Question 6

Role of V1 in vision?

Answer 6

Detection of edges and their orientation – early stage of object processing

 Direction selectivity – early stage of motion processing

 Binocular cells – early stage of depth
perception

 Lesion – no conscious vision

 Late (>100ms) signal processing in
V1 is sensitive to global organization
of a scene due to feedback from
higher-order areas (V4, IT, or MT) →
V1 response modulation. Modulated by context dependent feedback.

Question 7

What contributes to contour saliency?

Answer 7

Factors that contribute to contour
saliency include the number of contour elements
(compare the first and second frames), the spacing of
the elements (third frame), and the smoothness of the
contour (bottom frame). When the spacing between
contour elements is too large or the orientation
difference between them too great, one must search
the image to find the contour.

Question 8

How does contour integration in V1 reflext Gestalt principles?

Answer 8

Contour integration reflects the perceptual rules of
proximity and good continuation. Each of the four
images here has a straight line in the center, and all
four lines have the same oblique orientation. In some
images the line pops out more or less immediately,
without searching.

Question 9

What is V2?

Answer 9

Visual area V2, or secondary visual cortex, also called prestriate cortex,[24] is the second major area in the visual cortex, and the first region within the visual association area. It receives strong feedforward connections from V1 (direct and via the pulvinar) and sends strong connections to V3, V4, and V5. It also sends strong feedback connections to V1
In terms of anatomy, V2 has many properties in common with V1: Cells are tuned to simple properties such as orientation, spatial frequency, and color. The responses of many V2 neurons are also modulated by more complex properties, such as the orientation of illusory contours,[25][26] binocular disparity,[27] and whether the stimulus is part of the figure or the ground.[28][29]

Question 10

Figure ground segration and how that is enabled by perception of illusory contours?

Answer 10

You can see a nice response to the real rectangle.
Then a response still quite nice to theilliterate rectangle
and then nothing to the parts of the same image.
So um what is that for?
This is very helpful in segregating um figure from the
ground, especially if you think about thenatural images.
Well, one like one item against the ground isbasically
is a highly correlated um representationagainst very low correlation.
Uh say say uh this chair against that wall, the
chair has a large flat surfaces.
So there is a lot of correlations of, ofindividual features.

Question 11

How do Illusory contours further help us in perception?

Answer 11

Detection of illusory contours allow us to detect/percieve obejcts/ shapes partially occluded in nature and environment.

Question 12

What does V4 do?

Answer 12

Can detect shapes. Different cells have different shape preferneces irrespective of position in receptive field (THIS IS CALLED POSTITION INVARIANCE).Receptive fields are usually bigger in V4 neurons comapred with V2. **Integrates local cues into global shape object based representation. **

Question 13

What would a V 4 lesion result in?

Answer 13

V4 lesions in primates lead to severe disruptions of objects discrimination.

Question 14

How are objects detected by LOC?

Answer 14

V4 → LOC & IT (LOC possibly = TE in
monkeys)
* Lateral Occipital Complex – representation
of complex shapes. Results from human
fMRI; presented real-life objects, degraded
images, textures; LOC responded
selectively to objects, both familiar and
unfamiliar, and has size invariance. (Malach
et al., PNAS 1995) .
* LOC demonstrates form-cue invariance (apple will be apple regardless of colour and how it is presented cartoon, real life etc.

An object recognition system should be insensitive to the precise physic

Question 15

How are Ventral vs dorsal streams involved in categorisation of objects?

Answer 15

Whereas objects, houses and scenes (as compared with textures) activate both ventral and dorsal regions, faces and animals (as compared with textures) activate mainly lateral and ventral regions (Figure 1). Interestingly, faces and animals seem to activate similar regions, but houses and scenes activate different regions from those activated by faces and animals.

Question 16

Structures responsible for categorisation include?

Answer 16

V4 → LOC & IT
Categories of objects have different
representations: faces an animals
houses and landscapes
word shapes
tools

Question 17

Aperceptive agnosia?

Answer 17

Being able to dectect certain features of object but not see it as a whole. Will need to depend on other cues.

Question 18

Associative agnosia?

Answer 18

Being able to percieve object as a whole but unable to remember it. cannot assign meaning or function to it.

Question 19

Prosopagnosia

Answer 19

• In humans face recognition depends on the
  region of IT called fusiform face area (located
  within fusiform gyrus)
   Due to damage to the fusiform area ability to
  recognise faces is impaired, but perception of
  other objects & cognition are intact
   Prosopagnosia can be acquired or congenital
   Compensatory strategies are possible
  (recognise by voice, or shapes eg: big teeth)
   Can identify individual objects that make up a
  face, but can’t integrate into a complete facial
  percept

Question 20

Perceptual inference of motion in illusions.

Answer 20

Apparent motion has the attribute of motion due to perceptual inference arising from activation of directioanlly selective cells.The results indicate that the responses of strongly directionally selective cells were significantly higher for element pairs presented such
that the direction of motion in the illusion was in the cell’s preferred direction.

Question 21

Illusions

Answer 21

a perception of
motion arises from presentation of
stationary images/objects
demonstrated at different spatial
position as long as the spatial and
temporal frequencies of presentation
are appropriate (e.g., movie or a
sequence of lights in a neon sign)

Question 22

biologically important forms of motion

Answer 22

perceive optic flow (resulted from a relative motion between an
observer and a visual scene, e.g. objects changing size when we
move closer) and complex biological motion – specific movements
that results from the actions of animated objects (e.g. people)> Imoprtant for navigation, hazard perception and object segmentation.

Question 23

What is the motion perception hierarchy in dorsal stream?

Answer 23

V1 → Middle Temporal (MT or V5) → Medial Superior Temporal (MST) → Posterior Parietal Cortex (PPC)

Question 24

What cells in V1 are the ones that respond nicely to direction?

Answer 24

the ones that projected to MT. Usually in layer 4.

Question 25

What is the neural mechanisms of directionally selective cells in motion?

Answer 25

The selectivity of a neuron to the direction of
movement depends on the response latencies ofpresynaptic neurons. The response latencies of presynaptic neurons a and b relative to the onset of a stimulus are somewhat longer than those of neurons d and e. When a stimulus moves from left to right, neurons a and then b are activated first, but
because their responses are delayed their inputs arrive simultaneously with inputs from neurons d and e and therefore sum at the target neuron, causing it to fire. In contrast, stimuli moving leftward produce responses that arrive at different times and therefore do not reach threshold.