Categories • My dog sleeping. My dog. All golden retrievers. All dogs. All canines. All mammals… • Each of these is a category. • Categorization is the process of deciding which details matter, and which don’t, for some purpose.
Bruner, Goodnow, & Austin (1956) • Categories • reduce complexity of environment • allow us to generalize lessons • guide choice of response • make hierarchical knowledge available
Two questions about categories • 1. What is the structure of natural categories like? • This is a question about the world. • 2. How is information about natural categories represented in memory? • This is a question about your mind.
1. The structure of categories in the world • Is the question below a question about the world, or about us? “Which two are most similar: sheep, goats, cows?” • To some extent, the structure of natural categories is given by the world. • To some extent, it is impressed upon the world by human cognition
1. The structure of categories in the world • Most important work done by Eleanor Rosch. • Rosch argued that our categorical knowledge is organized in a hierarchy: • Superordinate, basic level, subordinate. • This is a relation of containing.
Mammal Dogs Cats Horses Collie Airedale Persian Siamese Arabian Superordinate Basic Level Subordinate
The Basic Level • The basic level falls between superordinate and subordinate levels. • It’s the one we use when we name an object. • It’s the one children learn first. • Things in a basic level category look like other things in the same category but not like things in other categories. This quality is not true at other levels.
Table Chair Lamp Bookcase This is asuper-ordinate category Dining room table Patio table Coffee table Picnic table This is a subordinate category Furniture Types of chair
Review: Hierarchy according to Rosch • Things in the world present themselves in a hierarchy of levels of categorization • Basic level, items in a category look like each other but not like members of other categories. • Basic level is first one learned and one used spontaneously in naming objects.
Rosch’s second contribution - Typicality • Rosch argued that some members of a category are “better” than others – that is, more typical. • such members have ‘family resemblance.’ • typical members can be verified fastest (implying fast access to their repns.) • typical members are similar to other members, unlike non-members of category
Things in the world vs. in the head • So far, we’ve been talking about things in the world. We asked, how do these things assemble into categories? • Now, we turn to the question of things in your head. • What is the nature of your knowledge about categories? How are they represented?
2. Mental representation of categories • Three kinds of categories: • Natural (e.g., Mammals) • Artificial (e.g., Animals that weigh > 100 pounds) • Functional (e.g., Things to bring out of the house in case of fire.) • We’ll consider only natural categories
Mental representation of natural categories • There are four competing models. Each specifies how a decision on category membership is made (how do you decide if this is an X?). Note – they might all be wrong. • 1. Prototype • 2. Feature frequency • 3. Nearest neighbour • 4. Average distance
Prototype models • Aprototype is a typical member of a category • Prototype theories say that, through experience, we create a central example of each category, and store that example. • A prototype captures what is typical of a category • A prototype may exist only in your mind (e.g., not as an actual object in the world).
Feature frequency models • Categorization is based on how many features the to-be-classified object shares with each of the available categories. • E.g., a whale shares ‘breathes air’ and ‘gives birth to live young’ with mammals. It shares ‘lives in the ocean’ and ‘moves by tail and flipper action’ with fish. • So a whale could be a fish or a mammal… • Models predict confusion about whales
Nearest neighbour models • New object is compared with each exemplar of each stored category. • Difference (on any dimensions) between object and each exemplar in each category is computed. • New object is classified in same category as object it is most similar to (smallest difference).
Has feathers Has feathers Exemplar: Duck New object: Hat Quacks Textile Swims Worn on the body Category: Water fowl Has feathers Exemplar: Goose Long and thin Exemplar: Tie Honks Textile Swims Worn on the body Category: Clothing
Average distance models • Comparison of new object to all stored exemplars of categories as in N. Neighbour. • Object goes in category with smallest average distance (exemplars to object). • Compare with Nearest Neighbour model – here, it is average distance for the category, not just which exemplar is closest, that counts.
Review: models of representation • These models reflect 2 very different views of category representation: • Prototype model • ‘what is generally true’ about something is stored and available when needed • this view emphasizes abstract representations (that is, not much detail)
Review: models of representation • N.N. and A.D. models • ‘what is generally true’ is not stored, but computed when needed. • these views emphasize storage of individual experiences with objects rather than storage of abstract ‘essences.’