Chapter 3-Feature and Pattern Detection Flashcards
Hyper Columns
-Contain all the neurons necessary to process any pattern in the visual field
-Orientation columns include V1 neurons sensitive to the same orientation
-Ocular Dominance shows preference toward one eye
Visual Acuity
Smallest detail that we can recognize
Visual Angle
Area on retina that an image occupies
Cones and visual acuity
Distribution of cones across the retina affects acuity (Acuity decreases with increased distance from fovea)
Spatial Frequency
of cycles per degree of visual angle (Dark to light cycles)
Perceptibility of a pattern is a function of spatial frequency and contrast (More sensitive to moderate frequencies)
Spatial Frequency Channels
Coarse- Respond to low frequencies
(Identifies object in visual field)
Medium- Respond to moderate frequencies
(Aids in pattern recognition and detecting contours)
Fine- Respond to high frequencies
(Processes fine details and specific examples of objects)
Extrastriate Cortex(V2-V5)
V2- Orientation and movement
V3- Complex and Dynamic form
V4- Color processing
V5- Motion perception
Processing Streams
Dorsal- “Where” Stream
Ventral- “What” Stream
Edge Detection
Most basic feature the visual system has evolved to detect is an edge or contour
-Receptive fields of striate and extrastriate neurons detect boundaries
Illusory contours
Illusory contours can be created when features align, giving the appearance of a boundary
Modal v. Amodal completion
Modal-Perception of illusory contours, like in the kanizsa figures
Amodal-Perception of complete but occluded objects
Lateral inhibition and edge detection
Lateral inhibition allows for the visual system to detect edges
Inverse Projection Problem
An image on the retina can be creates by an infinite number of feature combinations from the environment
Accidental Viewpoint
Gives perception of an object that is not present
Global v. Local level
Local level- Features
Global Level- Features put together to make a unified whole
Gestalt Psych
“The whole is greater than the sum of all parts”
Law of good continuation
Two elements will be grouped together if they lie on the same contour
Law of closure
We tend to form illusory contours to ‘close’ a figure
Law of proximity
Items near each other tend to be grouped
Law of similarity
Similar looking items tend to be grouped
Law of parallelism
Parallel contours likely to belong to the same group
Law of symmetry
Symmetrical regions are likely to be grouped
Law of common region
Items will be grouped if they appear to be part of the same, larger region
Law of connectedness
Items will be grouped if they are connected
Law of pragnanz
We perceive patterns in the simplest way possible
Perceptual Ambiguity
Addressed by yate’s thesis:
1. The visual environment is inherently ambiguous
2. Visual system must select ONE interpretation and stick with it
3.For efficiency, perception is the simplest interpreation
Priming
Presentation of something will influence how you respond to something later
-Helps overcome perceptual ambiguity
Expectancy
Expectancy through regularity does not affect perceptual ambiguity
Figure-ground Differentiation
Visual system must determine which part of the visual field is an object and which is the background
Affected by:
-Surroundedness
-Size
-Symmetry
Template theories
-Mental blueprints are stored in memory
-Recognition occurs by matching image to templates
-Problems:
-Lots of templates necessary
-Cannot explain how we recognize occluded/obscured patterns
Pandemonium model
-Occurs across several processing stages
-Each stage has perceptual committees that determine what features or patterns are present
(Bottom-up)
Structural Theories
-All objects can have a structural description
-Patterns are recognized by processing structural arrangement of parts
-Recognition based on assemblage of parts
Principle of componential recovery(Biederman)
Corners are critical for recognition
Recognition by Components (RBC) Theory
-Proposes geons are basic units of recognition
(Geons are 3d pieces)
-When we recognize patterns we detect geons and how they are arranged
-Geons are viewpoint-invariant
(We recognize them at any orientation)
