A. Nature of perceptual information
B. Problem of perceptual constancy -- pattern recognition
A. General ideas about pattern recognition
1. Data-driven vs. concept-driven models
a. Constructivist approaches (e.g., Gregory, 1980)
b. Direct perception (e.g., Gibson, 1979)
c. Compromise: Perceptual cycle (Neisser, 1976)
2. Segmentation vs. holism
a. Featural approach (see below) segments patterns into parts
b. Gestalt approach: principles of organization (proximity, similarity, good continuation, Prägnanz: closure, common fate) segregate figure and ground
B. Holistic (data-driven) theories about pattern representation
1. Template (analogical match)--visual scanners
2. Prototype (average of images)--Posner et al. (1967)
C. Segmental (data-driven) theory: Featural approaches
1. Definition: distinctive features (Gibson, 1969)
2. Pandemonium (Selfridge, 1959)
3 Neurophysiological evidence: e.g., striate cortical cells
4. Psychophysical evidence: e.g., sensory afterimages
5. Experimental psychological evidence
a. confusions (by children or learners or under poor conditions)
b. reaction time (to judge same or different)
c. judgments of similarity
D. Recent approaches (coordinating data- and concept-driven)
1. Problems with the featural approach
a. what are relevant features?
b. relations between features
c. ambiguity or various interpretations
2. Interactive, relational approach to patterns
a. "global" vs. "local" features (Navon, 1977)
b. recognition-by-components (Biederman's geons, 1987)
c. context effects (Gregory, 1980; Rock, 1983)
d. memory effects (subjective contours)e. evidence from visual agnosias (e.g., prosopagnosia)
f. evidence from hemispheric asymmetry
3. Interaction of attention and perceptual processes
Click here for: