Week 5 - Lecture 9 - Categorisation vTB

Page 1: Introduction to Categorisation

  • Course: PSYC201 - Cognitive Psychology

  • Instructor: Dr. Tom Beesley

  • Lecture 5 focuses on categories and their importance in cognitive psychology.

Page 2:

  • Empty Page

Page 3: Objectives

  • Understand categories/concepts and their usefulness.

  • Appreciate the limitations of classical views on categorisation.

  • Grasp the two main theories of categorisation:

    Exemplar Model

    Prototype Model

Page 4: What is a category?

  • Definition: A category groups instances that share common attributes (e.g., 'bird', 'animal').

  • Categories allow psychological learning and memory representation of these groupings, referred to as concepts.

Page 5: Functions of Concepts

  • Classification: Treats diverse entities as equivalent.

  • Understanding and Prediction: Facilitates anticipation of characteristics and initiates appropriate behaviors for new instances.

  • Communication: Enhances societal communication through shared knowledge about categories.

Page 6: How do we categorise?

  • Theories on human cognition's categorization process reflect key inquiries:

    • Are all category instances stored in memory?

    • How are instance relationships encoded?

    • Is category-specific knowledge stored?

Page 7: Classical view (“feature-based” theory)

  • Proposed by Bruner, Goodnow & Austin (1956).

  • Mental representations consist of defining features necessary for category membership.

    • Example: A triangle has three sides and a closed geometric form.

Page 8: Classical view limitations

  • Members of a category are either included or excluded without gray areas.

  • All category members are viewed equally, leading to confusion in application (e.g., are all musicians defined equally?).

Page 9: Additional Limitations of the Classical View

  • All non-members are equally poorly defined.

  • The process of category evaluation against defining features lacks clarity.

Page 10: Failings of the Classical View

  1. Typicality Effects:

    • Some instances are more typical than others, demonstrated by comparing members (e.g., differences between a sparrow and a penguin).

  2. Unclear cases:

    • Ambiguities can arise in classification (e.g., defining 'sport').

Page 11: Continued Failings of the Classical View

  1. Failure to Specify Defining Features:

    • Simple shapes like triangles are easy, but what defines a bird or musical instrument is complex and hard to pin down.

Page 12: McCloskey & Glucksberg (1978) Study

  • Participants classified items based on typicality. Repeat tests showed that

    • Typicality correlates with category assignments.

    • Greatest changes appeared in instances with intermediate typicality, indicating fuzzy category boundaries.

Page 13: Rosch & Mervis (1975)

  • Investigated what defines something as typical in a category.

  • Frequency of occurrence (e.g., sparrows are seen more often than penguins) influences judgement.

Page 14: Further Findings from Rosch & Mervis (1975)

  • More common attributes among members signify stronger typicality.

  • Fuzzy categories are further evidence against strict classical definitions of concepts.

Page 15: Typicality and Family Resemblance

  • Items viewed as prototypical yield higher family resemblance to other category members while showing lower resemblance to other categories (e.g., comparing tomatoes to usual fruit categories).

Page 16: Prototype Theory

  • Suggests that a summary representation is formed based on experiences with instances.

  • This representation reflects an ideal or average member of the category, termed the 'Prototype'.

Page 17: Dynamic Nature of Prototype

  • As more instances are experienced, the prototype evolves reflecting the central tendency or average.

Page 18: Advantages of Prototype Theory

  • Addresses the difficulties of defining features while allowing for ambiguity in boundary definitions.

Page 19: Evidence for Prototype Abstraction

  • Posner & Keele (1970):

    • Used random dot patterns in tasks to assess classifications of prototypes vs. distortions.

Page 20: Outcome of Posner & Keele's Research

  • Participants performed well with prototypes despite not training on them directly.

  • Better performance seen with small distortions over larger ones after a week.

  • Suggests learning involves abstract representation of prototypes during initial exposure.

Page 21: Challenges with Prototypes

  • Natural categories can present strange prototypes (e.g., defining both flying and non-flying birds).

  • More likely representation involves ideal features rather than a single item.

Page 22: Exemplar Theory

  • Proposes storing all exemplars rather than an abstract prototype.

  • Classifications rely on the similarity of new items to stored instances.

Page 23: Similarity Computation in Exemplar Theory

  • Classifications function on computed similarities from all stored exemplars, assessing if similarity crosses a threshold for categorization.

Page 24: GCM Model Overview

  • Nosofsky's Generalised Context Model (1984, 1986) elucidates exemplar theory's mechanisms involving attributes for classification.

Page 25: GCM Decision Making

  • Decisions are made based on the probability that an exemplar belongs to a category by comparing similarities among group memberships.

Page 26: Prototype Model Relations

  • Similarity computations compare new instances to the category's prototype rather than all exemplars collectively.

Page 27: Typicality and Classification

  • Exemplar theory can easily account for variations in typicality and category boundaries, thus explaining much of human cognition in categorisation.

Page 28: Further Research on Prototypes

  • Evidence supporting prototype theory includes tasks showing a direct correlation between item similarity to prototypical representations.

Page 29: Summary of Findings

  • The dialogue surrounding category learning emphasizes contrasting exemplary and prototypical theories, with implications for understanding cognitive flexibility and categorization depth.

Page 30: Implications for Learning

  • The likelihood of individuals grasping strict definitions or sets of features is low, highlighting the propensity towards prototype and exemplar approaches in cognitive processing.

Page 31: Suggested Reading

  • Murphy, G. (2002). The Big Book of Concepts. Cambridge: MIT Press. Available in the library for deeper understanding.