EXAM 2: Concepts and Categories Flashcards
Concepts
The mental representation of a category
Can be concrete (clear rules for category membership) or abstract (vague rules that don’t apply to all members)
Category
A set of things that are grouped together
Concepts: What is Instance/Exemplar
A member of a particular concept
Concepts: What is a feature?
Specific attribute that members share
What are the 3 levels of categorization?
Superordinate, Basic, Subordinate
What are some examples of superordinate levels of categorization?
Clothes, Furniture
What are some examples of Basic levels of categorizations
Coat, Chair
What are some subordinate levels of categorization
Pea coat, Folding chair
How do we form concepts?
(1) We encounter instances
(2) Get feedback (from person or environment)
(3) Store info in long-term memory (the info that is stored differs depending on the theory
Recognition by components: What is the structure of a concept?
Geons!
Recognition by components: Ho do we classify new instances?
Assess the spatial relationship of the geons, search memory for a match
Recognition by components: What info is stored in LTM?
The specific arrangement of the Geons
Recognition by components: Limitations
Works well for objects that can be defined by their shape, but fails with many other concepts
Classical View: What is the structure of this concept?
List of necessary and sufficient features
Classical View: How do we classify new instances?
See if instance matches the list for a specific concept
Classical View: What info is stored in LTM?
The list of features (once you know them, you’ll never categorize them incorrectly) (works really well for well-defined, concrete categories)
Problems with classical view
Some categories aren’t so easy to define. Between-category boundaries can be “fuzzy”, assumes all exemplars are equally good instances
Problems with classical view: Typicality effect
Some instances are more typical
Problem with classical view: Family resemblance
The degree of overlap between members of a category
Problems with classical view: Classical view cannot account for typicality effect because
- Assumes all exemplars are equally good instances
- The feature list is all you need
Prototype View: What is the structure of a concept?
- Idealized, “average” example that includes typical features
- One prototype for each concept/category
Prototype View: How do we classify new instances?
- Compare new instance to all prototypes closest match wins
Prototype View: What info is stored in LTM?
- The individual’s ONE prototypical example of the concept
- Prototype will MORPH over time with each new instance
Problems with prototype view
(1). How do we form these prototypes? Not well understood yet
(2). Not good at capturing knowledge at boundaries (EX: baby puppy looks like a guiniea pig but we know its a dog… why?)
(3). Context effects: These can influence typicality ratings (if you are always comparing to your prototype, you should always give the same rating, but people dont.
Exemplar View: What is the structure of a concept?
All instances of that concept
Exemplar View: How do we classify new instances?
Compare new isntance to every instance for all concepts - If new instance is most similar to instances of one concept, gets categorized as that concept
Exemplar View: What info is stored in LTM?
All instances
Strengths of exemplar view
(1) No need for concrete list of features
(2) Can account for the typicality effect (more stored instances of some cateogory members)
(3) Can account for context effects (context may heighten specific instances)
Problems with exemplar view
Comparing to every instance in LTM seems pretty inefficient
