PSY 341 Lecture Notes - Lecture 14: Mental Representation, Eleanor Rosch, List Of Fallacies
4/12/17
Categorization
•Fundamental to almost all cognitive processes
•Top-down feedback from categorized knowledge can sharpen perception
•Bottom-up: Senses -> mental representation
•To p - D o w n : M e n t a l r e p r e s e n t a t i o n - > s e n s e s
•Categories: Abstract mental representations that stand for things that you perceive or conceive
• Much of our knowledge is organized in terms of categories
Basic Model of Categorization
•From structural description of percept
•Search for a match in memory
•Select best match
•Draw inferences
•Store new results in memory
Classical Model: category membership is based on set of necessary and sufficient properties - all or none
•Problem: Not all categories concepts have necessary and sufficient features
•Strengths and Weaknesses
•Strengths:
•Clearly defined representation based on necessary and sufficient features, object either meets or
does not meet criteria
•odd/even #s
•Geometrical shapes
•Biological terms
•Weaknesses (fuzzy boundaries)
•Can’t account for similarity effects
•Can’t account for graded nature concepts
•Can’t account for context effect
Prototype Theory (Rosch) -avg / most experienced info
•Control insights of Prototype Theory
•Feature of most examples of categories are usually NOT necessary or sufficient (eg. Baby birds and
penguins can’t fly)
•We seem to think that some # of a category are more representative than others
•Prototypes are the avg of all relevant features of all exemplars in a category
•May NOT correspond to an actual object
•Ex: Avg American Family has a father, mother, and 2.5 children
•Categorization Process
•Compare structural description to all prototypes on memory
•Select prototype of category that structural description is most similar to
•Strengths and Weaknesses
•Strengths
•Accounts for graded structure effects (eg. Some birds better members of bird category)
•Variation in degree of closeness examples have to prototype
•Account for similarity judgement (robin, sparrow, flamingo)
•Exemplars sharing more features considered more similar
•Predicts Prototype Effects (unstudied Prototypes recognized better than studied Exemplars)
•Weaknesses:
•Can’t account for exemplar memory (YOUR dog/cat//mother)
•Not clear constraints on features - What features are extracted? What features are important?
•Doesn’t capture correlated constraints
Exemplar Model - based on the individual
•Says we don’t create prototypes and store then in memory
•We don’t use prototypes for classification
•Instead, we store and search through exemplars in memory
•Exemplars are structural description of EACH encountered object
•Must store a structural description of EVERY instance encountered
•Exemplars that are similar to each other are stored closer together - this closeness forms categories (eg.
Speech sounds)
•Categorization process
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Top-down feedback from categorized knowledge can sharpen perception. Categories: abstract mental representations that stand for things that you perceive or conceive. Much of our knowledge is organized in terms of categories. Classical model: category membership is based on set of necessary and sufficient properties - all or none. Problem: not all categories concepts have necessary and sufficient features. Clearly defined representation based on necessary and sufficient features, object either meets or does not meet criteria odd/even #s. Prototype theory (rosch) -avg / most experienced info. Feature of most examples of categories are usually not necessary or sufficient (eg. baby birds and penguins can"t fly: we seem to think that some # of a category are more representative than others. Prototypes are the avg of all relevant features of all exemplars in a category. Ex: avg american family has a father, mother, and 2. 5 children. Compare structural description to all prototypes on memory.
