Visual Word Recognition

25 important questions on Visual Word Recognition

Feature detection in the nervous system

Feature detectors are groups of neurons that code for perceptually significant stimuli. Early in the sensory pathway feature detectors have simple properties, late more complex detectors respond to features that are more specific

  • Hubel & Wiesel: edge detectors in the visual cortex of the cat. Edges do not occur often in the background 'noise' of the visual environment
  • Deel learning: multiple layers of non-linear processing units

Process of visual word recognition

Looking up words in the mental lexicon could proceed:
  • One-by-one (sequentially)
  1. Forster's autonomous serial search model
  • In parallel (simultaneously) = now assumed
  1. Morton's logogen model
  2. Connectionist model (localist/distributed) --> Interactive activation (IA) - McClelland & Rumelhart + PDP model - Seidenberg & McClelland

Pseudoword superiority effect

Letter in legal (pronounceable) non-words (CRAF) can be recognized faster than letter in series is consonants (CRGF)
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

IA model: word frequency explanation

Via differences in resting activation

IA model: word superiority effect explanation

Via top-down feedback

IA model: neighborhood effects explanation

Via coactivation of candidates and lateral inhibition

IA model: orthographic priming effects explanation

By presenting the prime a short time and then the target item, relevant candidates already become active

IA model: nonword effects explanation

Can be explained after formulating a task account

Problems of the IA model

  • Model functions only words of one and the same length
  • 'Position coding': words with a wrong letter order should not be recognized but they are
  • Phonology and meaning are not included

Phonological recoding (sounding out)

During reading, phonological representations may also be activated:
  1. Direct route: directly into the lexicon
  2. Indirect route: through sounding out
  3. Meaning-route: via meaning to sound


The Dual-route / 'house' model has all three routes

Types of scripts: abrades / Consonant Alphabets

Only represent consonants

Arabic, Hebrew

Types of scripts: Alphabets

Represent consonants and vowels

Latin, Greek

Types of scripts: Syllabic Alphabets / Abugidas

Have symbols for consonants and vowels. Consonants have an inherent vowel that can be changed by another vowel or diacritics. Vowels can also be written with separate letters.

Bengali, Bilanese

Types of scripts: Syllabaries

Represent syllables (consonants + vowels, or merely vowels)

Hiragana, Katakana

Types of scripts: Logographic writing systems

Represent (part of) words

Ancient Egyptian, form of Chinese

Speech of word recognition (ERP data)

Linear regression analysis of EEG recorded in a visual lexical decision task. Principal component analysis.

Word's surface form and meaning are first accessed at different times in different brain systems and then processed simultaneously. This supports interactive processing models with different stages.

Principle component analysis

4 orthogonal variables reflecting separable processing in visual word recognition:
  1. Word length
  2. Letter n-gram frequency
  3. Lexical frequency
  4. Semantic coherence of a word's morphological family


  • Starting round 90 ms: word length and letter n-gram frequency effects
  • Starting at 110 ms: lexical frequency effect
  • Starting at 160 ms: word - pseudo word differences
  • After 200 ms: all variables exhibit simultaneous EEG correlates


Source estimates indicate parieto-temporo-occipital generators for length, letter n-gram, frequency and word frequency. Widespread activation with foci (focus) in the left anterior temporal love and inferior frontal cortex related to Semantic coherence.

Types of priming: Repetition priming

Prime = target. Facilitation effect also appears when there is a longer time between the two presentations of the target.

Types of priming: Semantic priming

Prime and target are related in meaning (Bread and Butter), facilitation

Types of priming: Orthographic priming

Prime resembles target in spelling (Couch and Touch). Results are mixed, mostly facilitation has been found

Types of priming: Phonological / homophone priming

prime has the same pronunciation as the target (Knows and Nose), facilitation

Automatic or controlled priming?

Controlled processes are affected by expectations and are slower than automatic processes

Lexical ambiguity: Homographs

One spelling, multiple pronunciations and meanings



Tear

Lexical ambiguity: Homophones

Multiple speling and meanings, same pronunciations

Knight - night

Speed of context effects on visual word recognition (ERP data)

  • Hypothesis: sentence comprehension depends on continuous prediction of upcoming words
  • Effects of word frequency and contextual predictability on N1 (early visual processing), P200 (perceptual decoding), and N400 (semantics)

  • N1 --> low predictable words has a more negative anterior N1 than high predictable words - only for high frequency words
  • P200 --> low predictable words has a less positive P200 than high predictable words
  • N400 --> low predictable words had a greater N400 than high predictable words.

Context facilitates visual-feature and orthographic processing and semantic integration later

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo