Task 5: Mechanisms of middle and high-level vision Flashcards
Goal of middle vision
organise elements of visual scene into groups
Perceptual organisation
process by which elements in environment become perceptually grouped to create our perception of obejcts
Components of perceptual organization
- Grouping (putting together into units)
2. Segregation (separating one area from another)
Structuralism
- accumulated parts create the whole
- opponent to Gestalt approach
Gestalt approach
- an objects is a whole independent of the parts
- approach based on heuristics and not laws
Demonstrations against adding up sensations (structuralism)
- apparent movement
- illusory contours
Gestalt organizing principles about perceptual grouping
- Law of Pragnanz = structure is as simple as possible
- Law of Similarity = similar things appear to be grouped together (colour, shape, size and orientation)
- Law of good continuation = objects partially covered by other objects are seen as continuing behind the covering object
- Law of Proximity = things that are near each other
- principle of common region (overpowers proximity)
- principle of uniform connectedness (connected regions)
- principle of common fate (moving in the same direction)
Factors that determine perceptual segregation of figure and ground
- border ownership: even when figure and ground share contour, the border is associated with the figure
- we perceive an object as figure because of its meaningfulness, symmetry and region (lower region is better)
- the ground lack shape
Reversible figure-ground
patterns that can be perceived differently in the context
Relative motion
perceiving objects because they move
Gist of a scene
meaning of a scene that can be recognised within a fraction of a second
- is perceived first, followed by perception of details
Inference in perception
- knowledge of physical and semantic regularities
- some perceptions are the result of unconscious assumptions about the environment
- likelihood principle
Bayesian inference
- perception is combination of current stimulus and knowledge about the conditions of the world
- prior probability
- consistency of hypothesis
What are the different visual areas in the temporal lobe responsible for low, middle and high - level vision?
Low-level vision = V1 = Area 17 = Striate cortex
- lines and edges
Middle-level vision = V2 = Area 18
- border ownership
- real and illusory contours
- combines lines to form objects
High-level vision = V4
- complex attributes
Naive template theory
- is inaccurate because too many templates would be needed
- matching neural representation with a stored representation of the same shape in the brain
Solution for the naive template theory
structural description = objects perceived as relationship between parts, therefore we would keep in memory only one representation (recognise Jennifer Anniston with glasses and without)
Biederman’s Model of Recognition-by-Components theory
- objects are recognised by identities and relationships of their component parts
= usually recognise objects by the use of geons
Geon
- parts of which the perceptual objects is built
- easy and quick to recognise, regardless of orientation in space
- viewpoint invariant
Viewpoint invariance
representations of an object don’t change when observer’s viewpoint changes
Template descriptions
2D memory representations from different viewpoints
Another way by which objects are recognised is by
entry-level category (label)
=> subordinate (specific) or superordinate (broader) level
Illusory contours evoke response in
area 18 of the visual cortex (V2)
Area 17 is
unable to see contours
Other factors that pay a role in the formation of contours are
corners and line ends
Illusory contours are generated by
the amount of contrast at luminance borders
Experiment I with the Kanisza square concluded that
children of 3-4 months are able to extract subjective contours
Visual word form area
activated when we read
Pure alexia
- impariment in VWFA
- inability to recognise a single word, faces, objects, digits and numerals
- ability to speak and even write
We are able to read because
of a preexisting circuit that links left ventral visual pathway to left-hemispheric language areas
Left hemisphere lateralisation
of VWFA, the same as spoken language
Neurons in visual pathway respond to
simple shapes => they signal properties that tent to be viewpoint invariant
Acquisition of reading induces
reorganisation of ventral visual pathway (features for letter recognition are incompatible with those for faces) => one area pushes the other away, towards the right hemisphere