PSY280: Lecture 5 Notes
1. Describe and recognize the parameters of middle vision
a. Middle vision is the stage of processing that comes after basic
features and before object recognition or scene understanding.
b. Its goal is to organize the elements of a scene.
2. Summarize some of the challenges with finding edges
a. Connectedness won’t work and isn’t even necessary.
b. We can differentiate important edges.
c. We don’t need edges to perceive segregation.
i. We can have perception of edges without the sensory input
of edges as shown in the illusory contours.
3. Explain the structuralism failed to explain some perceptual phenomenon
and inspired the Gestalt Approach
a. Structuralism: perceptions result from the summation of many
elementary sensations. Perceptual experience is combination of
lower components, e.g. color, size…
b. Structuralists could not explain illusory contours or apparent
motion. It talks about the components of the whole.
4. Define and recognize illusory conjunctions and apparent motion
a. Apparent motion, e.g. stationary light flickering on and off gives you
the appearance of motion even though all spots are stationary.
5. Explain the general Gestalt Approach
a. Gestalt: the whole is different than the sum of its parts.
b. The perceptual whole is different from the sum of its sensory parts.
c. Gestalt principles are like individuals with opinions about how the
information should be organized.
d. When two principles are operating, it’s hard to predict which one
6. Recognize various Gestalt principles that apply to perception
7. Explain the Pandemonium model and its levels
a. Image demons get the sensory information (starting at the lowest
b. Feature demons like specific features (a feature demon for each
single feature, which creates noise).
c. Cognitive demons like specific letters (taking evidence from feature
d. The decision demons listen to the committee.
8. Describe ambiguous figures and explain their implications a. Ambiguous figures that give rise to 2 different perceptions.
ii. Multiple interpretation of the 3D cube
9. Summarize the 2 rules that tend to be used to resolve ambiguity and
recognize novel applications
a. Rule 1: Honor physics
i. E.g. solid, opaque objects block light, so we interpret the
blockage of light/ absence of shades as the solid object
b. Rule 2: Avoid accidents
i. We reduce unsolvable problem and turn it into easily
solvable problem. There are many interpretations for each
object that are only briefly considered by visual system.
10. Describe the principles of figure-ground segregation
a. To determine what is the object and what is the background, there
are some specific Gestalt rules that just apply to this figure-ground
b. Border ownership is associated with the figure. (The border that
lines the figure)
c. Surroundedness: the one that is surrounded is the figure
d. Parallelism: the parallel contours are the figures
e. Size: the smaller size is the figure
f. Relative location: the bottom is the figure
g. Symmetry: the symmetric one is the figure
h. Meaningfulness: the one that has meaning is the figure
11. Compare and contrast naïve template theory and structural descriptions
a. Naïve template theory: it matches every single pixel of what you
see to the representation in your memory. It is about the matching
of image to the mental template. The brain looks at the basic
structure of the image but not the specific details, so no
overwhelmingly many templates are needed. Object recognition is
by matching to the templates.
b. Structural description: it is about how do the objects relate to each
other or not. Object recognition by components.
c. Recognition by components model exploits the concept of non-
accidental properties and take advantage of structural description.
12. Define and recognize non-accidental features
a. E.g. curves, the only time that you do not see a curve is when you
rotate the object in depth as in a coin, but when you rotate a ball,
there is always curve.
b. E.g. parallel contours, when you rotate the object in depth, parallel
lines do not change and they are always parallel to each other
constantly. c. E.g. t-junctions (occlusion)
d. E.g. y-junctions
e. E.g. arrow junctions
13. Describe the kinds of NAPs that define geons and recognize novel
a. Geons are geometic ions, the basic units of objects.
b. There are 36 geons, each composed of unique sets of non
accidental properties (NAPs)
c. It has viewpoint invariance, that is geons can be identified when
viewed from most viewpoints.
d. Edges: can be straight or curved
e. Axes: going through the body with or without bending it
f. Contour: changing how far apart they are in distance
14. Summarize recognition by components and its major criticism
a. Principle of componential recovery
b. If you are viewing from orientation that you cannot see the geon,
one cannot identify the object.
c. We can also identify an object if we are able to see a few