Chapter 2: Typicality and the Classical View of Categories G. - - PowerPoint PPT Presentation

Chapter 2: Typicality and the Classical View of Categories

• G. Murphy (2002) The Big Book of Concepts, The MIT Press.

Noortje Venhuizen

ILLC/ UvA

October 11, 2010

Contents

1

Introduction

2

Classical View of Concepts Problems with the Classical View

3

Typicality Posner and Keele (1968, 1970) Rosch and Mervis (1975)

4

Conclusion

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 2 / 13

Introduction

Concepts and Definitions

Question: What is the mental representation of concepts/ categories? First answer: definitional. Dogs: have four legs, bark, have fur, eat meat, sleep, ...

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 3 / 13

Introduction

Concepts and Definitions

Question: What is the mental representation of concepts/ categories? First answer: definitional. Dogs: have four legs, bark, have fur, eat meat, sleep, ... The conditions in a definition should be: Necessary: conditions hold only for members of the category. Sufficient: if all conditions are fulfilled, the entity should belong to the category.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 3 / 13

Classical View of Concepts

Classical View of Concepts

Proponents of concepts as definitions: Hull (1920), Smoke (1932), Piaget (1964) Main claims:

1 Concepts are mentally represented as definitions. 2 Every object is either in or not in the category. 3 There is no distinction between category members. Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 4 / 13

Classical View of Concepts

Classical View of Concepts

Proponents of concepts as definitions: Hull (1920), Smoke (1932), Piaget (1964) Main claims:

1 Concepts are mentally represented as definitions. 2 Every object is either in or not in the category. 3 There is no distinction between category members.

Problems: Theoretical/ empirical problems

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 4 / 13

Classical View of Concepts Problems with the Classical View

Theoretical Problems

Categories difficult to define using necessary and sufficient features. Fuzziness of categories:

science, law, artificial domains

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 5 / 13

Classical View of Concepts Problems with the Classical View

Theoretical Problems

Categories difficult to define using necessary and sufficient features. Fuzziness of categories:

science, law, artificial domains Well, I’ll tell you something. You really don’t know what a metal is. And there’s a big group of people that dont know what a metal is. Do you know what we call them? Metallurgists! . . . Here’s why metallurgists dont know what metal is. We know that a metal is an element that has metallic properties. So we start to enumerate all these properties: electrical conductivity, thermal conductivity, ductility, malleability, strength, high density. Then you say, how many of these properties does an element have to have to classify as a metal? And do you know what? We can’t get the metallurgists to agree. Some say three properties; some say five properties, six properties. We really don’t know. So we just proceed along presuming that we are all talking about the same thing. (Pond, 1987 pp. 62-63)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 5 / 13

Classical View of Concepts Problems with the Classical View

Empirical Problems

Category membership not discretely determined Borderline cases between categories.

Hampton (1979), McCloskey and Glucksberg (1978)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 6 / 13

Classical View of Concepts Problems with the Classical View

Empirical Problems

Category membership not discretely determined Borderline cases between categories.

Hampton (1979), McCloskey and Glucksberg (1978)

Non-logical use of concepts Non-compositionality of categories, failure of transitivity.

Hampton (1982, 1987)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 6 / 13

Classical View of Concepts Problems with the Classical View

Empirical Problems

Category membership not discretely determined Borderline cases between categories.

Hampton (1979), McCloskey and Glucksberg (1978)

Non-logical use of concepts Non-compositionality of categories, failure of transitivity.

Hampton (1982, 1987)

Typicality effects Typical category members are the good examples - what you normally think of when you think of the category. The atypical objects are ones that are known to be members but that are unusual in some way.

Rosch (1975), McCloskey and Glucksberg (1978), Rips, Shoben and Smith(1973)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 6 / 13

Typicality

Typicality Effects

Examples of typicality effects: Judgment of borderline cases (McCloskey and Glucksberg (1978)) Ease of membership judgments (Rips, Shoben and Smith(1973)) Production of items from a category (Battig and Montague (1969)) Learning artificial categories (Rosch, Simpson and Miller (1976)) Inferences about category members (Rips (1975)) ...

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 7 / 13

Typicality

Revision of the Classical View

Distinction between two aspects of category representation: core and identification procedures Characteristic (non-definitional) features part of identification procedure; core definitional. Typicality effects result from the identification procedures.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 8 / 13

Typicality

Revision of the Classical View

Distinction between two aspects of category representation: core and identification procedures Characteristic (non-definitional) features part of identification procedure; core definitional. Typicality effects result from the identification procedures. But: Hampton’s (1988, 1995) results show that typicality ratings are the best predictor of untimed category judgments, the ones that should involve the core.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 8 / 13

Typicality Posner and Keele (1968, 1970)

Typicality as a Phenomenon

Prototype structure of concepts: illustrated by Posner and Keele’s (1968, 1970) dot-pattern studies:

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 9 / 13

Typicality Posner and Keele (1968, 1970)

Typicality as a Phenomenon

Prototype structure of concepts: illustrated by Posner and Keele’s (1968, 1970) dot-pattern studies: Each category has a most typical item (actual or non-actual) Items less similar to prototype more likely to be excluded from category Typicality is a graded phenomenon (extremely typical/ moderately typical/ atypical/ borderline category members)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 9 / 13

Typicality Rosch and Mervis (1975)

What Makes Items Typical and Atypical?

Frequency?

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 10 / 13

Typicality Rosch and Mervis (1975)

What Makes Items Typical and Atypical?

Frequency? Counterexamples: chicken, car racing.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 10 / 13

Typicality Rosch and Mervis (1975)

What Makes Items Typical and Atypical?

Frequency? Counterexamples: chicken, car racing. Family resemblance Typical items...

1

tend to have the properties of other category members.

2

tend not to have properties of category nonmembers.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 10 / 13

Typicality Rosch and Mervis (1975)

What Makes Items Typical and Atypical?

Frequency? Counterexamples: chicken, car racing. Family resemblance Typical items...

1

tend to have the properties of other category members.

2

tend not to have properties of category nonmembers.

Rosch and Mervis (1975) study: Evaluation of typicality of items in natural categories. Listing the attributes of each of the items. Judge-amending process for correctness of attributes. Weighting of attributes on basis of occurrence.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 10 / 13

Typicality Rosch and Mervis (1975)

What Makes Items Typical and Atypical?

Frequency? Counterexamples: chicken, car racing. Family resemblance Typical items...

1

tend to have the properties of other category members.

2

tend not to have properties of category nonmembers.

Rosch and Mervis (1975) study: Evaluation of typicality of items in natural categories. Listing the attributes of each of the items. Judge-amending process for correctness of attributes. Weighting of attributes on basis of occurrence. Result: Items with most common features in the categories have the highest score.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 10 / 13

Typicality Rosch and Mervis (1975)

Examples of categories

Categories are ordered by typicality:

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 11 / 13

Typicality Rosch and Mervis (1975)

Results

Correspondence between high typicality rating and high score. ⇒ first part of hypothesis confirmed.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 12 / 13

Typicality Rosch and Mervis (1975)

Results

Correspondence between high typicality rating and high score. ⇒ first part of hypothesis confirmed. Non-correlational study also confirms second part of hypothesis: Items with greater overlap with other categories are harder to learn and rated less typical.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 12 / 13

Typicality Rosch and Mervis (1975)

Results

Correspondence between high typicality rating and high score. ⇒ first part of hypothesis confirmed. Non-correlational study also confirms second part of hypothesis: Items with greater overlap with other categories are harder to learn and rated less typical. Additional variables suggested by Barsalou (1985): Central tendency: corresponds to first part of Rosch and Mervis’ hypothesis. Frequency of instantiation: frequency as a member of the category. Ideals: degree to which each item fits the primary goal of each category.

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 12 / 13

Conclusion

End of Classical View?

Extremely difficult to find definitions for most natural categories. Typicality and unclear membership unpredicted by classical view. Existence of intransitive category decisions. “A theory should not be held just because the criticisms of it can be argued against - the theory must itself provide a compelling account of the data.” (Murphy, 2002)

Noortje Venhuizen (ILLC/ UvA) Chapter 2: Typicality and the Classical View of Categories October 11, 2010 13 / 13