Cognitive Lecture 5

Northumbria University: Knowledge and Categories

Recognition of Objects and Abstraction

• Recognizing objects, faces, words, and sentences is fundamental in cognitive processing.

• Abstraction: Involves generalizing beyond direct perception of details.

  • Facilitates forming expectations and making decisions in new situations.

  • Categorization is a method of applying knowledge to real-world situations, building abstract knowledge.

Defining Categories

• Formal categories often have specific definitions, but day-to-day (natural categories) can be more complex.

• Family Resemblance: Describes that natural categories lack strict definitions but capture similarities among members.

  • Examples of how to judge category membership are discussed, highlighting the subjective nature of categorization.

Hierarchical Structure of Categories

• Categories can exist in hierarchical relationships:

  • Superordinate Category: The broadest category (e.g., 'Vehicle').

  • Basic/Entry-Level Category: Most commonly used label (e.g., 'Car').

  • Subordinate Category: More specific (e.g., 'Toyota Prius').

• Basic categories offer a balance between specificity and generality, enhancing understanding of features.

Hierarchical Semantic Network Model

• Proposed by Collins & Quillian (1969).

• Illustrates how semantic memory is structured, emphasizing:

  • Cognitive economy: Knowledge is stored with minimal connections.

  • Information retrieval occurs through spreading activation among nodes within the network.

• Example nodes:

  • Animal: Bird > Canary, Pigeon

  • Vehicle: Car > Prius, 747

Hierarchical Distance Effect

• The verification of relations at similar levels is faster than at different levels.

• Example: It’s quicker to verify that a canary can sing than that it has feathers due to closer relational links.

• Activation spreads more slowly across distant nodes (greater distance in the hierarchy).

Familiarity in Hierarchical Networks

• Familiarity may influence verification speed (Conrad, 1972):

  • More frequent encounters with certain phrases strengthen associations, leading to quicker responses.

• Typicality effects explain variations in response speeds for different category members, as some members (like canaries) are perceived as more typical than others (like penguins).

Feature-Set Models and Categorization

• Feature-Set Models: Categorization based on shared features.

  • Prototype Approach: Every category has a prototype representing shared characteristics (not always an actual instance).

  • Exemplar Approach: Categories are represented as lists of stored exemplars; comparisons are made with all members rather than a fixed prototype.

Knowledge in the Brain

• Brain damage can lead to specific deficits, affecting knowledge about either living things or tools.

• Embodied Cognition: Predicts that knowledge retrieval occurs through simulation of experiences; for instance, reading about actions can activate related brain areas.

Language and Comprehension

• Understanding language goes beyond simplistic word recognition.

• Inferring meaning is crucial for full comprehension, especially with non-literal language (metaphors, sarcasm).

• Pragmatic Model (Grice, 1975): Sequence for processing language: 1) process literal meaning; 2) assess sense; 3) search for alternatives if necessary.

  Challenges to model:

• Non-literal meanings (e.g., metaphors) are not always more slowly understood

• Alternative view:

• Difficulties with non-literal meanings due to competition between literal and non-literal meanings (both processed)

Discourse Processing

• Language comprehension spans multiple sentences.

• Necessary inferences can be categorized as:

  • Logical: Direct conclusions (e.g., birds have wings).

  • Bridging: Linking previous and current information.

  • Elaborative: Drawing on world knowledge to infer information not stated directly.

Schema Theory in Comprehension

• Story comprehension depends on understanding more than mere language; knowledge of characters, events, and their interaction is vital.

• Schemas(Bartlett,1932): Integrated knowledge packets that aid memory and comprehension, helping to fill gaps during information processing.

  • Example: Waking up schema comprises steps like alarm, coffee, shower.

Comprehension Without Schema - Bransford and Johnson 1972

• Illustrates cognitive difficulty when prior knowledge (schema) is not used in processing information, resulting in confusion.

Strengths and Weaknesses of Schema Theory

• Strengths:

  • Highlights how prior knowledge facilitates understanding of language and memory recall.

  • Also explains many memory errors and distortions

    • Participants recall events not mentioned in stories, but that are schema-consistent

    • An important consideration in eyewitness testimony

• Weaknesses:

  • Abstract concepts can be hard to define precisely.

  • Unclear when specific schemas become activated, which complicates understanding.

    Alternatives

  • Models that specify processes involved in linking text to LTM (Construction-integration model)

  • Models that specify key details readers try to identify in text (Event-indexing model)

  • Models that link comprehension to simulation (Experiential-simulation approach)

Summary of Key Points to Review

• Knowledge structures: categories, family resemblance, and types of hierarchies.

• The role of hierarchical semantic networks and the importance of typicality in categorization.

• Embodied simulation and its relevance to comprehension and knowledge retrieval.

• Understanding language and discourse requires more than sentence processing; it necessitates a grasp of schemas and the ability to make inferences.