lecture 2- categorization Flashcards
cognitive economy
-by dividing the world into classes of things we decrease the amount of information we need to learn, perceive, remember and recognize
superordinate category
- high coverage, low predictive value
- often applicable
- yields little information
- example: animal
basic level category
- trade-off usefulness/information
- example: dog
- basic level shifts downward as expertise grows -> in subordinate categories
subordinate category
- low coverage
- high predictive value
concept formation
- abstraction of feature set
- e.g. child acquires representation of concept ‘apple’
concept learning
- by applying concept and getting feedback
- e.g. child learns that/why tomato is not an apple
similarity based categorization (3)
1) classical theories
2) prototype- or probabilistic theories
3) exemplar- or instance-based theories
classical theory
-category is represented by a series of defining features that specify category boundary
flaws classical theory
- categorization would require perfect match between object and definition
- some objects are more typical members than others
- > these are learned and recognized faster
Prototype-or probabilistic theories
- category is represented by a series of characteristic features
- prototype specifies category center
prototype flaws
- list of attribute values ignores feature correlations?
- no information saved on the range of attribute values ?
- similarity to earlier example influences categorization
exemplar-based/instance-based model
-category consists of set of examples
exemplar based
flaws
- defining features ?
- memory capacity
prototope or specific examples?
research on language acquisition and expertise development suggest:
- exemplars are more important during early learning
- abstract information (prototype) becomes more important as experience grows
experts keep using examples
-physicians are more likely to diagnose correctly when they have recently seen a specific case that presents like the current one
