Overview

1
Overview

• Theories of categorization
• Classical view
• Prototypes
• Exemplar theories
• The role of similarity
• The role of beliefs and implicit theories
• Category learning

2
Concepts

• We use concepts in everyday thinking and
  interacting with objects
• Classifying objects
• Referring to objects in communicating
• Applying our general knowledge
• Dealing with novel items -- we look for
  similarities to items that we already know about
• Making inferences

3
Classical view

• Categories have boundaries. An item is either a
  member or not a member.
• Items must satisfy certain conditions to be
  within the category
• dog has 4 legs, born of dog parents, barks, has
  a tail
• city has tall buildings, has paved roads, people
  work there

4
Problems for classical view

• Concepts are hard to define. Example what is a
  game?
• Easy to find category members that do not fit the
  rule
• dog with three legs
• fictional cities (Atlantis)
• Cannot explain typicality effects a German
  shepherd is a better example of a dog than a
  Chihuahua

5
What causes typicality effects?

• One explanation is that people compare each
  instance to a prototype for the category
• A prototype is a representation that combines the
  average features for that category
• protoypical dog has 4 legs, barks, has a tail,
  weighs around 45 pounds, is brown

6
Prototypes - examples

• More examples of prototypes
• The prototypical pizza has tomato sauce, has
  cheese, is 12 in diameter, is baked
• The prototypical faculty member is...?
• The prototypical family is... ?
• Categories have fuzzy boundaries - not clear what
  are members and what are not

7
Evidence of typicality effects

• Categories have graded membership Some members
  of a category are reliably rated as better
  members than others
• rating task
• production task
• picture identification task
• induction task

8
Rating task for fruit

• On a scale of 1 to 7, with 7 as highest
  typicality
• Apple 6.25
• Peach 5.81
• Strawberry 5.00
• Watermelon 4.06
• Fig 3.38
• Olive 2.25

9
Production task

• When asked to make up a list of fruits, people
  will more generate typical fruit names earlier on
  the list (apple, orange, peach, grapefruit,
  apricot, grapes, blueberries, honeydew)
• Suggests that memory search starts with the
  prototype and works outward to less typical
  members of the category

10
Picture identification task

• Participants are asked to make a yes/no judgment
  - Is this item an example of the category?
• People are quicker to say yes to pictures of
  more typical examples (e.g., German shepherds
  apples) than less typical examples (e.g.,
  Chihuahuas grapes)

11
Induction task

• Participants are asked to learn a new fact about
  one item A robins heart has three chambers.
  People were more willing to agree that A ducks
  heart has three chambers than in the reverse
  situation
• A ducks heart has three chambers.
• Does a robins heart have three chambers?

12
Induction task (cont.)

• People will make inferences from the typical to
  the whole category
• People will not make inferences from the atypical
  to the whole category
• What you know to be true of a typical bird is
  likely to be true of most birds
• What you know to be true of an atypical bird
  (e.g., penguin) might not be true of most birds

13
Basic level categories

• Proposed by Eleanor Rosch
• Basic level categories are the natural level of
  categorization
• not too specific
• not too general
• usually requires only one word
• may be acquired earliest by children

14
Basic level categories (cont.)

• Evidence that we use basic level categories
• developmental studies
• more likely to be produced in naming tasks
• easier to explain what members of basic level
  categories have in common

15
Exemplar theories

• Exemplar a specific remembered instance
• Your concept of dog is made up of all the
  examples of dogs that you have encountered
• Typical items are encountered more frequently,
  so people have many stored representations of
  them, and these memories will be well primed

16
Exemplar theories (cont.)

• Exemplar theories can explain typicality effects
• Recognition task typical items are more quickly
  recognized because memory search for a matching
  exemplar will be fast
• Production task when asked to list items in a
  given category, typical items are more frequently
  represented in memory

17
Advantages of exemplar over prototype theories

• 1. The set of exemplars can represent how
  variable the category is
• Information about variability is used by people,
  and is not represented by a prototype
• 2. Categories can shift with recent experience -
  this suggests that we use specific examples, not
  an average that is very stable

18
Advantages of Exemplar over prototype theories
(cont.)

• 3. We can generate categories that have no
  prototypes
• things to take on vacation
• things to see in Paris
• these are examples of ad hoc categories
• The evidence suggests that conceptual knowledge
  relies on more than prototypes

19
The role of similarity

• Similarity (or resemblance) is crucial to both
  prototype and exemplar theories of categorization
• Categorization by similarity is a useful
  heuristic (fast and efficient strategy)
• However, we judge more than similarity when
  judging category membership
• A painted, flattened lemon is still a lemon
• A well-done counterfeit bill is not a 20 bill

20
Similarity (cont.)

• Similarity depends on context
• W hen asked to rate the similarity between Italy
  and Switzerland, people gave moderate ratings
• When the people were asked to compare Italy,
  Switzerland, and Brazil, the similarity rating
  for Italy with Switzerland went up

21
The role of beliefs

• Which features we pay attention to depends on our
  general beliefs. Example
• When judging whether an animal is a raccoon, we
  pay attention to deep properties (genetic).
  Superficial properties (e.g., fur color) are less
  relevant

22
The role of beliefs (cont.)

• Our judgments are guided by implicit theories
  about how the world works
• Can an airplane be made of glass?
• Can an airplane be made of ice cream?
• Even though neither of these is close to the
  prototypical airplane, or to airplanes people
  have encountered (exemplars), people tend to say
  yes to glass and no to ice cream

23
The role of beliefs (cont.)

• Alfred is an unmarried adult male. He has been
  living with his girlfriend for the past 23 years.
  Is Alfred a bachelor?
• Bernard is an unmarried adult male and does not
  have a partner. Bernard is a monk living in a
  monastery. Is Bernard a bachelor?
• Charles is a married adult male, but he has not
  seen his wife for many years. Charles is
  earnestly dating, hoping to find a new partner.
  Is he a bachelor?
• Donald is a married adult male, but he lives in a
  culture that encourages males to take two wives.
  Donald is earnestly dating, hoping to find a new
  partner. Is he a bachelor?

24
Concept learning

• When first exposed to a concept, people refer to
  exemplars
• As they learn more about the category, they rely
  more on a prototype
• However, people continue to be influenced by
  cases they have seen recently
• Dermatologists were more likely to make a correct
  diagnosis if they had recently seen a similar case

25
Summary

• Categorization is complex! Peoples judgments of
  category membership are influenced by
• prototypes
• exemplars (especially recent ones)
• beliefs about what features are relevant
• implicit theories about what features are
  essential