Semantic Memory
I. Introduction to
Semantic Memory
A. What is semantic memory?
1. Permanent memory store of our world
knowledge
2. Different from episodic memory – no
representation of when or where we learned
the information
3. Examples
• What is the capitol of Colorado?
• How many legs does a horse have?
• What color is a canary?
I. Introduction to
Semantic Memory
B. Key questions:
1. How is information stored?
2. What is the nature of the
representation?
3. How is information learned?
4. How is information retrieved?
I. Introduction to
Semantic Memory
C. New terminology
1. Concepts
• Mental representations
• Often “the fundamental unit of thought”
• “An idea that includes all that is characteristically
associated with it” (Medin, 1989)
2. Proposition
• The relationship between concepts
• E.G.
– A canary is yellow
– A canary is a bird
– A bird has wings
– A bird is an animal
II. The Collins & Quillian
Hierarchical Model
A. Key Properties
1. Network: An interrelated set of concepts
/ body of knowledge.
2. Node: A point or location in the network
representing a single concept.
3. Pathways: associations between
concepts (propositions) that are
directional.
II. The Collins & Quillian
Model
A. Key Properties
3. Pathways: associations between
concepts (propositions) that are
directional.
• “ISA” pathways denote category
membership – “Canary is a bird”
• Property pathways describe properties of
concepts “Bird has feathers”
Sample Portion of the
Semantic Network
II. The Collins & Quillian
Model
B. Spreading Activation
1. The mental activity of accessing and retrieving
information from the network.
2. Takes passive concepts (those not currently in
working memory) and activates them (puts
them in working memory).
3. Activation then spreads to related nodes (e.g.,
activation to the doctor node would also
spread to the nurse node).
II. The Collins & Quillian
Model
C. Key Features of the Model
1. Hierarchical – concepts are arranged in
a hierarchy
2. This allows for cognitive economy –
removes any redundancy
3. Takes time for activation to spread
across inactive nodes
4. Activation spreads quickly across
recently activated nodes
Sample Portion of the
Semantic Network
III. Smith’s Feature
Comparison Model
A. General Structure
1. Information about concepts are
represented as feature lists
2. These lists include both defining
features and characteristic features
• Defining features are essential to the
meaning of a concept
• Characteristic features are common, but not
essential to the meaning of a concept
III. Smith’s Feature
Comparison Model
2. These lists include
both defining
features and
characteristic
features
• Defining features are
essential to the
meaning of a concept
• Characteristic features
are common, but not
essential to the
meaning of a concept
Robin
Physical object
Living
Animate
Feathered
Red breasted
III. Smith’s Feature
Comparison Model
2. These lists include
both defining
features and
characteristic
features
• Defining features are
essential to the
meaning of a concept
• Characteristic features
are common, but not
essential to the
meaning of a concept
Robin
Eats worms
Roosts in trees
III. Smith’s Feature
Comparison Model
2. These lists include
both defining
features and
characteristic
features
• Defining features are
essential to the
meaning of a concept
• Characteristic features
are common, but not
essential to the
meaning of a concept
Leos
Born between July 23
and August 22
III. Smith’s Feature
Comparison Model
2. These lists include
both defining
features and
characteristic
features
• Defining features are
essential to the
meaning of a concept
• Characteristic features
are common, but not
essential to the
meaning of a concept
Leos
Loyal
Self-assured
Charming
Generous
Opinionated
Overbearing
Proud
III. Smith’s Feature
Comparison Model
A. General Structure
1. Information about concepts are
represented as feature lists
2. These lists include both defining
features and characteristic features
3. Features are stored starting with most
defining followed by most characteristic
Sample Feature Lists
III. Smith’s Feature
Comparison Model
B. Feature Comparison
1. Models are often tested using sentence
verification tasks – e.g. a ROBIN is a
BIRD
2. The model begins with Stage 1 – global
feature comparison
• ‘Fast yes’ responses occur when there is a
large number of shared features
• ‘Fast no’ responses occur when there are
few shared features
A ROBIN is a BIRD…
III. Smith’s Feature
Comparison Model
B. Feature Comparison
2. The model begins with Stage 1 – global
feature comparison
• ‘Fast yes’ responses occur when there is a
large number of shared features
• ‘Fast no’ responses occur when there are
few shared features
• Intermediate comparisons (some shared,
some not shared) move to a Stage 2
Comparison of Defining Features
Feature Comparison
III. Smith’s Feature
Comparison Model
B. Feature Comparison
2. The model begins with Stage 1 – global
feature comparison
• Intermediate comparisons (some shared,
some not shared) move to a Stage 2
Comparison of Defining Features
• When the defining features match, a ‘slow
yes’ response is given
• When there is a feature mismatch, a ‘slow
no’ response is given
Feature Comparison
IV. Direct Comparisons of
Models – Central Themes
A. The Principle of Cognitive Economy
1. A primary assumption of the
Hierarchical Model is cognitive economy
– only non-redundant facts are stored
2. The members of a category inherit the
properties of the category itself – the
principle of inheritance
IV. Direct Comparisons of
Models – Central Themes
B. Property Statements
1. Problem with the
Feature Comparison
model, it is unable to
account for property
statements
2. Can handle “Robin
has wings” but…
3. The model assumed a
feature list that
corresponded to the
properties – i.e.
“THINGS WITH
WINGS”
4. Hierarchical models
handle the same as
any other statement
IV. Direct Comparisons of
Models – Central Themes
C. Typicality Effects
1. Typicality refers to the degree to which items
are viewed as typical, central members of a
category
2. Battig & Montague (1969) category
membership norms
3. Less frequent category members have lower
feature overlap than common members
• Robin is a bird – common
• Chicken is a bird – uncommon
IV. Direct Comparisons of
Models – Central Themes
C. Typicality Effects
3. Less frequent category members have
lower feature overlap than common
members
• Robin is a bird – common
• Chicken is a bird – uncommon
4. Typical members can be judged more
rapidly than atypical members
IV. Direct Comparisons of
Models – Central Themes
C. Typicality Effects
3. Less frequent
category members
have lower feature
overlap than common
members
• Robin is a bird –
common
• Chicken is a bird –
uncommon
4. Typical members can
be judged more
rapidly than atypical
members
IV. Direct Comparisons of
Models – Central Themes
C. Typicality Effects
4. Typical members can be judged more
rapidly than atypical members –
typicality effect
IV. Direct Comparisons of
Models – Central Themes
D. Semantic Relatedness
1. A direct prediction of feature
comparison relates to how similar
concepts are
2. Rips (1975) – demonstration
IV. Direct Comparisons of
Models – Central Themes
D. Semantic Relatedness
1. A direct prediction of feature
comparison relates to how similar
concepts are
2. Multi-dimensional Scaling Solutions
IV. Direct Comparisons of
Models – Central Themes
D. Semantic
Relatedness
1. A direct prediction of
feature comparison
relates to how similar
concepts are
2. Multi-dimensional
Scaling Solutions
IV. Direct Comparisons of
Models – Central Themes
D. Semantic Relatedness
1. A direct prediction of feature
comparison relates to how similar
concepts are
2. Multi-dimensional Scaling Solutions
3. Modern semantic network models
IV. Direct Comparisons of
Models – Central Themes
D. Semantic
Relatedness
4. Modern semantic
network models
• Redundant
information
• Distance
represents how
similar concepts
are
Learning and Motivation,
Concepts and
Categories
I. Types of Categories
A. Natural Categories
1. Occur naturally in
the world
2. Essentially define
themselves
3. Naturally occurring
concepts are labeled
after discovery
I. Types of Categories
A. Natural Categories
1. Occur naturally in
the world
2. Essentially define
themselves
3. Naturally occurring
concepts are labeled
after discovery
I. Types of Categories
B. Artifact Categories
1. Objects or
conventions
designed by humans
to serve particular
functions
2. Category
membership is
primarily determined
by function /
intended use
I. Types of Categories
B. Artifact Categories
1. Objects or conventions
designed by humans to
serve particular functions
2. Category membership is
primarily determined by
function / intended use
3. Should herding dogs be
an artifact category?
I. Types of Categories
C. Nominal Categories
1. Linguistic conventions that involve the
arbitrary assignment of a label to things that
fit a particular set of conditions
2. Often defined as a matter of convenience
I. Types of Categories
D. Ad Hoc Categories
1. Formed for a
purpose
2. Can be influenced by
context
I. Types of Categories
D. Ad Hoc Categories
1. Formed for a
purpose
2. Can be influenced by
context
I. Types of Categories
D. Ad Hoc Categories
1. Formed for a
purpose
2. Can be influenced by
context
feet
dust
bird
atom
hands
animal
cell
children
I. Types of Categories
D. Ad Hoc Categories
1. Formed for a
purpose
2. Can be influenced by
context
3. “Small things”
4. Other examples:
Things you take on
vacation
feet
dust
bird
atom
hands
animal
cell
children
I. Types of Categories
F. Levels of Categorization
1. Categories are both horizontally and
vertically organized
• Animals – plants
• Animals
– Birds
 Raptors
I. Types of Categories
F. Levels of Categorization
1. Categories are both horizontally and
vertically organized
2. Categories are also hierarchical with super
ordinate and subordinate categories
Animals
Birds Cats
Raptors Songbirds
Eagle
Hawk Canary Sparrow
Golden
Bald
Lions Tigers
II. Approaches to Concept
Representation
A. Classic approaches
B. Prototype theories
C. Exemplar theories
II. Approaches to Concept
Representation
A. Classic approaches
1. Similar to Feature theories
2. Classification is based on certain features or
characteristics
3. Features are both necessary and sufficient
for categorization
4. Problem – not everything can be categorized
this way: Games
• Typicality effects
• Fuzzy boundaries
II. Approaches to Concept
Representation
B. Prototype theories
1. As we learn, we “abstract out” a prototype
2. A prototype is the most typical, or
representative idea of a category
II. Approaches to Concept
Representation
C. Exemplar theories
1. Category membership is conducted by
comparison to stored examples (exemplars)
2. More typical members will be similar to
many exemplars
3. Categorization is entirely based on
comparison to stored examples
III. Applications of
Concepts and Categories
A. Stereotyping
1. We automatically categorize people based on
visual features (sex, age, race, weight)
2. Social-category level beliefs have the power to
shape impressions of individuals
3. Some are automatic, others are based on
either acquired information or assumptions
about category membership (e.g. either
someone discloses they are gay or assumed to
be gay).
III. Applications of
Concepts and Categories
B. Perceptions of minorities
1. Ethnic minorities
2. Sexual minorities
• Merritt et al (2013) gay actors perceived as
less masculine
3. Knowledge may influence our behavior
in these cases
III. Applications of
Concepts and Categories
C. Stereotype threat
1. Stereotype threat is the self threat
experienced by members of a negatively
stereotyped groups that they will be
judged or behave in ways that confirm
the stereotype
2. Often this precipitates the undesired
behavior
3. Our knowledge of our own group
membership can influence our behavior
III. Applications of
Concepts and Categories
D. How might categories affect
behavior?
1. Social categories may be formed via
prototypes or exemplars
2. Examples
• Gender
• Masculinity and femininity
• Sexual orientation
• Occupations
IV. Other ways our
knowledge influences
behavior
A. Schemas and scripts
1. Guide our actions
2. Can they lead us into trouble
• Sexual schemas
• Sexual scripts
B. Selective perception & Selective
memory

Semantic Memory_2014(1)-3.ppt

  • 1.
  • 2.
    I. Introduction to SemanticMemory A. What is semantic memory? 1. Permanent memory store of our world knowledge 2. Different from episodic memory – no representation of when or where we learned the information 3. Examples • What is the capitol of Colorado? • How many legs does a horse have? • What color is a canary?
  • 3.
    I. Introduction to SemanticMemory B. Key questions: 1. How is information stored? 2. What is the nature of the representation? 3. How is information learned? 4. How is information retrieved?
  • 4.
    I. Introduction to SemanticMemory C. New terminology 1. Concepts • Mental representations • Often “the fundamental unit of thought” • “An idea that includes all that is characteristically associated with it” (Medin, 1989) 2. Proposition • The relationship between concepts • E.G. – A canary is yellow – A canary is a bird – A bird has wings – A bird is an animal
  • 5.
    II. The Collins& Quillian Hierarchical Model A. Key Properties 1. Network: An interrelated set of concepts / body of knowledge. 2. Node: A point or location in the network representing a single concept. 3. Pathways: associations between concepts (propositions) that are directional.
  • 6.
    II. The Collins& Quillian Model A. Key Properties 3. Pathways: associations between concepts (propositions) that are directional. • “ISA” pathways denote category membership – “Canary is a bird” • Property pathways describe properties of concepts “Bird has feathers”
  • 7.
    Sample Portion ofthe Semantic Network
  • 8.
    II. The Collins& Quillian Model B. Spreading Activation 1. The mental activity of accessing and retrieving information from the network. 2. Takes passive concepts (those not currently in working memory) and activates them (puts them in working memory). 3. Activation then spreads to related nodes (e.g., activation to the doctor node would also spread to the nurse node).
  • 9.
    II. The Collins& Quillian Model C. Key Features of the Model 1. Hierarchical – concepts are arranged in a hierarchy 2. This allows for cognitive economy – removes any redundancy 3. Takes time for activation to spread across inactive nodes 4. Activation spreads quickly across recently activated nodes
  • 10.
    Sample Portion ofthe Semantic Network
  • 11.
    III. Smith’s Feature ComparisonModel A. General Structure 1. Information about concepts are represented as feature lists 2. These lists include both defining features and characteristic features • Defining features are essential to the meaning of a concept • Characteristic features are common, but not essential to the meaning of a concept
  • 12.
    III. Smith’s Feature ComparisonModel 2. These lists include both defining features and characteristic features • Defining features are essential to the meaning of a concept • Characteristic features are common, but not essential to the meaning of a concept Robin Physical object Living Animate Feathered Red breasted
  • 13.
    III. Smith’s Feature ComparisonModel 2. These lists include both defining features and characteristic features • Defining features are essential to the meaning of a concept • Characteristic features are common, but not essential to the meaning of a concept Robin Eats worms Roosts in trees
  • 14.
    III. Smith’s Feature ComparisonModel 2. These lists include both defining features and characteristic features • Defining features are essential to the meaning of a concept • Characteristic features are common, but not essential to the meaning of a concept Leos Born between July 23 and August 22
  • 15.
    III. Smith’s Feature ComparisonModel 2. These lists include both defining features and characteristic features • Defining features are essential to the meaning of a concept • Characteristic features are common, but not essential to the meaning of a concept Leos Loyal Self-assured Charming Generous Opinionated Overbearing Proud
  • 16.
    III. Smith’s Feature ComparisonModel A. General Structure 1. Information about concepts are represented as feature lists 2. These lists include both defining features and characteristic features 3. Features are stored starting with most defining followed by most characteristic
  • 17.
  • 18.
    III. Smith’s Feature ComparisonModel B. Feature Comparison 1. Models are often tested using sentence verification tasks – e.g. a ROBIN is a BIRD 2. The model begins with Stage 1 – global feature comparison • ‘Fast yes’ responses occur when there is a large number of shared features • ‘Fast no’ responses occur when there are few shared features
  • 19.
    A ROBIN isa BIRD…
  • 20.
    III. Smith’s Feature ComparisonModel B. Feature Comparison 2. The model begins with Stage 1 – global feature comparison • ‘Fast yes’ responses occur when there is a large number of shared features • ‘Fast no’ responses occur when there are few shared features • Intermediate comparisons (some shared, some not shared) move to a Stage 2 Comparison of Defining Features
  • 21.
  • 22.
    III. Smith’s Feature ComparisonModel B. Feature Comparison 2. The model begins with Stage 1 – global feature comparison • Intermediate comparisons (some shared, some not shared) move to a Stage 2 Comparison of Defining Features • When the defining features match, a ‘slow yes’ response is given • When there is a feature mismatch, a ‘slow no’ response is given
  • 23.
  • 24.
    IV. Direct Comparisonsof Models – Central Themes A. The Principle of Cognitive Economy 1. A primary assumption of the Hierarchical Model is cognitive economy – only non-redundant facts are stored 2. The members of a category inherit the properties of the category itself – the principle of inheritance
  • 25.
    IV. Direct Comparisonsof Models – Central Themes B. Property Statements 1. Problem with the Feature Comparison model, it is unable to account for property statements 2. Can handle “Robin has wings” but… 3. The model assumed a feature list that corresponded to the properties – i.e. “THINGS WITH WINGS” 4. Hierarchical models handle the same as any other statement
  • 26.
    IV. Direct Comparisonsof Models – Central Themes C. Typicality Effects 1. Typicality refers to the degree to which items are viewed as typical, central members of a category 2. Battig & Montague (1969) category membership norms 3. Less frequent category members have lower feature overlap than common members • Robin is a bird – common • Chicken is a bird – uncommon
  • 27.
    IV. Direct Comparisonsof Models – Central Themes C. Typicality Effects 3. Less frequent category members have lower feature overlap than common members • Robin is a bird – common • Chicken is a bird – uncommon 4. Typical members can be judged more rapidly than atypical members
  • 28.
    IV. Direct Comparisonsof Models – Central Themes C. Typicality Effects 3. Less frequent category members have lower feature overlap than common members • Robin is a bird – common • Chicken is a bird – uncommon 4. Typical members can be judged more rapidly than atypical members
  • 29.
    IV. Direct Comparisonsof Models – Central Themes C. Typicality Effects 4. Typical members can be judged more rapidly than atypical members – typicality effect
  • 30.
    IV. Direct Comparisonsof Models – Central Themes D. Semantic Relatedness 1. A direct prediction of feature comparison relates to how similar concepts are 2. Rips (1975) – demonstration
  • 31.
    IV. Direct Comparisonsof Models – Central Themes D. Semantic Relatedness 1. A direct prediction of feature comparison relates to how similar concepts are 2. Multi-dimensional Scaling Solutions
  • 32.
    IV. Direct Comparisonsof Models – Central Themes D. Semantic Relatedness 1. A direct prediction of feature comparison relates to how similar concepts are 2. Multi-dimensional Scaling Solutions
  • 33.
    IV. Direct Comparisonsof Models – Central Themes D. Semantic Relatedness 1. A direct prediction of feature comparison relates to how similar concepts are 2. Multi-dimensional Scaling Solutions 3. Modern semantic network models
  • 34.
    IV. Direct Comparisonsof Models – Central Themes D. Semantic Relatedness 4. Modern semantic network models • Redundant information • Distance represents how similar concepts are Learning and Motivation,
  • 35.
  • 36.
    I. Types ofCategories A. Natural Categories 1. Occur naturally in the world 2. Essentially define themselves 3. Naturally occurring concepts are labeled after discovery
  • 37.
    I. Types ofCategories A. Natural Categories 1. Occur naturally in the world 2. Essentially define themselves 3. Naturally occurring concepts are labeled after discovery
  • 38.
    I. Types ofCategories B. Artifact Categories 1. Objects or conventions designed by humans to serve particular functions 2. Category membership is primarily determined by function / intended use
  • 39.
    I. Types ofCategories B. Artifact Categories 1. Objects or conventions designed by humans to serve particular functions 2. Category membership is primarily determined by function / intended use 3. Should herding dogs be an artifact category?
  • 40.
    I. Types ofCategories C. Nominal Categories 1. Linguistic conventions that involve the arbitrary assignment of a label to things that fit a particular set of conditions 2. Often defined as a matter of convenience
  • 41.
    I. Types ofCategories D. Ad Hoc Categories 1. Formed for a purpose 2. Can be influenced by context
  • 42.
    I. Types ofCategories D. Ad Hoc Categories 1. Formed for a purpose 2. Can be influenced by context
  • 43.
    I. Types ofCategories D. Ad Hoc Categories 1. Formed for a purpose 2. Can be influenced by context feet dust bird atom hands animal cell children
  • 44.
    I. Types ofCategories D. Ad Hoc Categories 1. Formed for a purpose 2. Can be influenced by context 3. “Small things” 4. Other examples: Things you take on vacation feet dust bird atom hands animal cell children
  • 45.
    I. Types ofCategories F. Levels of Categorization 1. Categories are both horizontally and vertically organized • Animals – plants • Animals – Birds  Raptors
  • 46.
    I. Types ofCategories F. Levels of Categorization 1. Categories are both horizontally and vertically organized 2. Categories are also hierarchical with super ordinate and subordinate categories
  • 47.
    Animals Birds Cats Raptors Songbirds Eagle HawkCanary Sparrow Golden Bald Lions Tigers
  • 48.
    II. Approaches toConcept Representation A. Classic approaches B. Prototype theories C. Exemplar theories
  • 49.
    II. Approaches toConcept Representation A. Classic approaches 1. Similar to Feature theories 2. Classification is based on certain features or characteristics 3. Features are both necessary and sufficient for categorization 4. Problem – not everything can be categorized this way: Games • Typicality effects • Fuzzy boundaries
  • 50.
    II. Approaches toConcept Representation B. Prototype theories 1. As we learn, we “abstract out” a prototype 2. A prototype is the most typical, or representative idea of a category
  • 51.
    II. Approaches toConcept Representation C. Exemplar theories 1. Category membership is conducted by comparison to stored examples (exemplars) 2. More typical members will be similar to many exemplars 3. Categorization is entirely based on comparison to stored examples
  • 52.
    III. Applications of Conceptsand Categories A. Stereotyping 1. We automatically categorize people based on visual features (sex, age, race, weight) 2. Social-category level beliefs have the power to shape impressions of individuals 3. Some are automatic, others are based on either acquired information or assumptions about category membership (e.g. either someone discloses they are gay or assumed to be gay).
  • 53.
    III. Applications of Conceptsand Categories B. Perceptions of minorities 1. Ethnic minorities 2. Sexual minorities • Merritt et al (2013) gay actors perceived as less masculine 3. Knowledge may influence our behavior in these cases
  • 54.
    III. Applications of Conceptsand Categories C. Stereotype threat 1. Stereotype threat is the self threat experienced by members of a negatively stereotyped groups that they will be judged or behave in ways that confirm the stereotype 2. Often this precipitates the undesired behavior 3. Our knowledge of our own group membership can influence our behavior
  • 55.
    III. Applications of Conceptsand Categories D. How might categories affect behavior? 1. Social categories may be formed via prototypes or exemplars 2. Examples • Gender • Masculinity and femininity • Sexual orientation • Occupations
  • 56.
    IV. Other waysour knowledge influences behavior A. Schemas and scripts 1. Guide our actions 2. Can they lead us into trouble • Sexual schemas • Sexual scripts B. Selective perception & Selective memory