Word Recognition

Psy315 - Class 8

(This page last updated 2 October, 2006.)

 

 Basic Paradigm

 

Word Frequency

Repetition Priming

Word Superiority Effect

Display Sequence
  • Word displayed briefly.
  • Mask at word position.
  • Subject reports letter at position indicated by arrow by selecting one of two choices.

     

Feature Net Model of Word Recognition

Detectors connected to one another and arranged in hierarchy.
  • feature detectors
  • letter detectors
  • word detectors

 

How does the Feature Net explain the effects of word frequency?

  • Before a word is seen, different word detectors start out with different levels of activation.
  • A high frequency word starts with a high level of activation.
    • Only a little additional activation from the stimulus is needed to trigger recognition.
  • A low frequency word starts with a low level of activation.
    • Much additional activation from the stimulus is needed to trigger recognition.

 

How does the Feature Net explain the effects of repetition priming?

  • After a word is seen the first time, its activation remains higher than normal.
  • If the word appears again soon afterward, it will require less activation to trigger recognition.

 

The simple Feature Net cannot explain why words and pronounceable nonwords are easier to perceive than nonpronounceable nonwords.

  • Add extra layer of detectors for bigrams (letter pairs) between letters and words.
  • Pronounceable nonwords will have many of the same bigrams that appear in real words.
  • Nonpronounceable nonwords will not have as many of these bigrams.
  • When a pronounceable nonword (HICE) appears, there will be bigram detectors ready to respond to many of its letter pairs (HI, CE).
  • A nonpronounceable nonword (RSFK) will not generate as much activity in the bigram detectors, and it will thus not be recognized as easily.

 

next class: Object Recognition


Psych 315: Cognitive Psychology Kyle Cave Psychology Dept. U. Mass.