Please enable JavaScript.

Coggle requires JavaScript to display documents.

Pattern Recognition: The study of how people identify objects in their…

- - - - Criteria for selecting features for uppercase letters: Critical, unchanged (under changes in brightness, size, perspective), yield unique pattern for each letter, number of features should be reasonably small
      - Evaluated by how well it can predict perceptual confusions (two patterns that are mistakenly identified as each other); generated by presenting letters very rapidly and asking observer to identify them
        
        Model proposed by Geyer and De Wald trumps Gibson
    - - Identifying faces within a category can be difficult when we emphasise features that allow us to distinguish between categories (e.g. race) rather than within categories (e.g. facial features)
      - Emphasising distinctive features
  - - - It is easier to distinguish two different geons than to distinguish between two variations of a geon
      - Removed 65% of contour from drawing of objects, some allowing observers to see how the segments were related, some making it more difficult to do so. When drawings were presented for 100msec, subjects correctly named 70% if contours deleted at midsegments, but correctly named less than 50% if contours deleted at vertices
- - - - Sensory memory: Iconic memory (visual), echoic memory (auditory), haptic (tactile), olfactory (smell), gustatory (taste)
- - - - Visual perception involves parallel processing. There is simultaneous processing of all letters in a word. Recognition also occurs simultaneously at three different levels of abstraction (word level, letter level, feature level). There are two connections between the levels: excitatory connections (positive evidence) and inhibitory connections (negative evidence).
      - All letters of a word are in their correct order (critique)
    - - Nodes: Represented by features, letters, and words in the interactive activation model
      - Nodes are connected to one another by excitatory and inhibitory connections that differ in strength
      - Activation rules: How a node combines its excitatory and inhibitory inputs with its current state of activation
      - State of activation: Nodes can be activated to various degrees
      - Output functions: Relate activation levels to outputs (e.g. what threshold has to be exceeded for conscious awareness)
      - Learning rule: Learning occurs by changing the weights of the excitatory and inhibitory connections between the nodes; learning rule specifies how to make these changes
      - Many psychologists believe that this model more accurately portray how the b rain works than other more serial models of behaviour, may capture how people learn, allow for a different kind of computing in which many weak constraints can be simultaneously considered
      - Huber and O'Reily (2003)
        
        Connections temporarily habituate, resulting in lower firing rates when viewing a stimulus for longer time. Inhibitory connections of model are within levels rather than between levels, resulting in masking effects and competition between alternative interpretations of the same stimulus
        
        Viewing a word for longer durations creates neural habituation for that word, making the word more difficult to identify when it occurs again as the target