2.1 Cognition: Thinking & Remembering
Problem 6 – How do we represent language?
Literature: Robinson-Riegler, Matlin
Mechanical aspects of reading:
- Top-down processing: from expectations and knowledge
- Bottom-up processing: from word/data
Eye movements
- Saccades = movement of eye from one point to another; continuous; 6-8 letters; 20ms
Saccadic suppression = little to no intake of visual info (too fast)
o Fixations = brief pauses (200-300ms; length depends)
Gaze = consecutive fixations; summed = gaze duration
o Some words not fixated; others fixated multiple times
o Influences on gaze time:
Content words; unfamiliar words; predictions; expectation; age of learning word; word length/spacing
o Fixation not necessary for identification (e.g. function words)
- Word skipping (high frequency & short words)
- Regressive saccades
= backwards movements when saccade was too long
o Poor readers do more backtracking
- Perceptual span
= amount of text covered effectively to the right of fixated point
o English (3 characters left; 15 right); may differ
o Parafoveal information = slightly out of perceptual span; aids word recognition process
Detect word length; identification of length → skipping
o Space → words are distinguishable; target for eyes; cues following word by constraining length
Visual object recognition
Template-matching theory
- Template of pattern stored in memory → compare stimulus to pattern
- Limitation:
o Inflexible
Even recognize different fonts
o Only for 2D, isolated letters, numbers, objects
Recognize objects even if you only see part of it
o Too simplistic
Feature-analysis theory
- Objects composed of number of characteristics/components
- Distinctive features of letters (e.g. “R” has curve, diagonal line, vertical line)
o Compare features → still recognize different fonts
- Support: slower distinction between similar letters (e.g. R and P vs. L and O)
- Limitations:
o Not simply list of features, but also physical relationships
o Other complex objects have too many lines/curves features to compare (e.g. a moving horse!)
Recognition-by-components theory
- Three dimensional shapes
- Objects represented by arrangement of simple 3D shapes (→ “geons”)
Word recognition
- Orthography = look of written word
Direct access view = assumption that orthography → recognition
Indirect access view = assumption that orthography → phonological representation → recognition
- Supported
- Homophome (word pairs with same sound components but different spelling; e.g. reed, read)
Problem 6 – How do we represent language?
Literature: Robinson-Riegler, Matlin
Mechanical aspects of reading:
- Top-down processing: from expectations and knowledge
- Bottom-up processing: from word/data
Eye movements
- Saccades = movement of eye from one point to another; continuous; 6-8 letters; 20ms
Saccadic suppression = little to no intake of visual info (too fast)
o Fixations = brief pauses (200-300ms; length depends)
Gaze = consecutive fixations; summed = gaze duration
o Some words not fixated; others fixated multiple times
o Influences on gaze time:
Content words; unfamiliar words; predictions; expectation; age of learning word; word length/spacing
o Fixation not necessary for identification (e.g. function words)
- Word skipping (high frequency & short words)
- Regressive saccades
= backwards movements when saccade was too long
o Poor readers do more backtracking
- Perceptual span
= amount of text covered effectively to the right of fixated point
o English (3 characters left; 15 right); may differ
o Parafoveal information = slightly out of perceptual span; aids word recognition process
Detect word length; identification of length → skipping
o Space → words are distinguishable; target for eyes; cues following word by constraining length
Visual object recognition
Template-matching theory
- Template of pattern stored in memory → compare stimulus to pattern
- Limitation:
o Inflexible
Even recognize different fonts
o Only for 2D, isolated letters, numbers, objects
Recognize objects even if you only see part of it
o Too simplistic
Feature-analysis theory
- Objects composed of number of characteristics/components
- Distinctive features of letters (e.g. “R” has curve, diagonal line, vertical line)
o Compare features → still recognize different fonts
- Support: slower distinction between similar letters (e.g. R and P vs. L and O)
- Limitations:
o Not simply list of features, but also physical relationships
o Other complex objects have too many lines/curves features to compare (e.g. a moving horse!)
Recognition-by-components theory
- Three dimensional shapes
- Objects represented by arrangement of simple 3D shapes (→ “geons”)
Word recognition
- Orthography = look of written word
Direct access view = assumption that orthography → recognition
Indirect access view = assumption that orthography → phonological representation → recognition
- Supported
- Homophome (word pairs with same sound components but different spelling; e.g. reed, read)
