Lecture 12: Semantic Memory and Categorization

Lecture 12: Semantic Memory Distinctions Within Explicit Memory

1. Explicit/Declarative Memory

  • Definition: Explicit or declarative memory is characterized as conscious and verbalizable, allowing individuals to articulate knowledge about facts and events. It encapsulates the idea of 'knowing that.'

A. Types of Explicit Memory

  1. Episodic Memory

    • Definition: Refers to the memory of personal episodes tied to specific times and places.

    • Examples:

      • Dinner last night

      • Your first kiss

  2. Semantic Memory

    • Definition: Represents general knowledge and factual information that is not tied to specific times or places.

    • Examples:

      • Understanding what dinner is

      • Basic arithmetic, such as $2 + 2 = 4$

2. Categorization in Semantic Memory

  • Organizational Structure: Semantic memories are organized around categories, which facilitate a more systematic approach to understanding the knowledge base.

  • Knowledge Groups: Individuals often discuss categories rather than specific instances, allowing for easier cognitive processing of similar objects or concepts.

A. Example: Categorization by Pigeons (Wasserman, 1987)

  • Study Design:

    • Pigeons trained to peck one of four keys based on stimuli.

    • Trained over 30 days with rewards for correct responses.

  • Results:

    • 81% accuracy with familiar (old) examples.

    • 64% accuracy with new exemplars.

    • Conclusion: Pigeons were able to abstract a category from the given examples, suggesting a level of conceptual understanding.

B. Categorization by Conceptual Knowledge

  • Developmental Insights:

    • Humans, starting from 4 years old, can initiate categorization.

    • Children may categorize based on features perceived as essential rather than merely perceptual.

  • Example: A child might assert that a blackbird feeds its young with mashed food rather than milk, based on perceived essential characteristics.

3. Defining a Mental Category

A. Classical View

  • Definition: The classical view of categorization determines membership through defined criteria.

  • Features:

    • Necessary Features: Characteristics an object must possess to belong in a category.

    • Sufficient Features: Traits that guarantee category membership if present.

  • Examples:

    • For ‘bird’: wings, beak, ability to fly, etc.

  • Note: Many scientific classification systems utilize this framework.

B. Problems with the Classical View

  • Challenges include defining features that are both necessary and sufficient to encompass everyday categories.

  • Counterexample: Not all birds possess wings or can fly.

C. Modern View

  • Probabilistic Categories: Characteristic properties are considered likely rather than strictly necessary or sufficient.

    • Enables acknowledgment of fuzzy boundaries in category definitions.

    • Example: Flightless birds still exhibit traits common to birds, like having feathers.

4. Evidence for Probabilistic (Fuzzy) View

A. Typicality Effects

  • Sentence Verification Study:

    • Individuals verify more typical category members faster than atypical ones.

    • Examples:

    • "A robin is a bird" elicited quicker confirmations than "A chicken is a bird."

  • Hedges:

    • Describing less typical members using language that reflects deviation from prototypical features.

    • Example: "A whale is technically a mammal," implying it deviates from typical land-dwelling characteristics.

B. Characterization Based on Similarity

  1. Exemplar Theories

    • Individuals store multiple examples of a category in memory, categorizing new items based on similarity to these examples.

  2. Prototype Theories

    • A single average or idealized representation (prototype) of a category is stored in memory, with categorization driven by similarity to this prototype.

C. Geometric Representation of Similarity

  • Similarities among concepts can be visually represented in geometrical space based on ratings of similarity.

  • Methodology:

    • Participants rate the similarity of various objects.

    • Spatial arrangement reflects the distance implied by those ratings (closer objects are perceived as more similar).

5. Metric Axioms and Their Violations

A. Minimality

  • Axiom: The distance between a concept and itself should always equate to the smallest possible value (zero).

  • Violation: Familiar concepts tend to be rated as more similar to themselves than unfamiliar concepts.

  • Example: An apple is rated more similar to itself than a pomegranate is to itself.

B. Symmetry

  • Axiom: The similarity rating between two concepts should remain consistent regardless of the order in which they are compared.

  • Violation: Unfamiliar concepts are often rated as more similar to familiar ones than vice versa.

  • Example: A pomegranate compared to an apple is rated more similarly than the reverse.

C. Triangle Inequality

  • Axiom: If concept A is similar to concept B and B to C, then the distance from A to C should be less than the combined distances of (A + B) + (B + C), denoting that A and C maintain a level of similarity.

  • Violation: Similarity judgments can contradict this axiom.

  • Example: Chocolate and coffee may be rated similarly, but coffee and tea show no such similarity despite their proximity in categorical context.

6. Geometric Approaches

  • Conceptual Framework: Geometric theory summarizes how individuals perceive similarity ratings among various concepts.

  • Inconsistency with Theory: Individual similarity ratings do not always align with geometric theory.

A. Tversky’s Featural Contrast Model

  • Framework: A feature-based approach aiming to explain similarity through shared and distinct characteristics between items.

  • Mechanism: The approach evaluates similarities based on feature overlap without requiring the metric axioms established in geometric theory.