Semantic memory Flashcards
What is a ‘concept’?
A mental representation of a category/class of objects
Which theories demonstrate how adults represent concepts?
- Classical Defining Feature theory
- Feature Comparison model
- Prototype model
According to Classical Defining Feature theory, what is a ‘concept’ defined by?
Represented in memory & defined by a set of features
What are defining features of a CHAIR? Why are they important? Can a chair have additional features?
Defining features = inanimate, used for sitting on, has a seat
Chair must have these defining features - they are central to the representation of our understanding of the object
A chair can have additional features (e.g. made of wood)
Features are ‘primitives’. What does this mean?
They can’t be broken down into smaller units
Equally important
All examples of a concept are equally representative of that concept. True/false?
True - all instances of a concept are equally representative of that concept
How are concepts organised in memory?
Concepts are organised in a hierarchy of superordinate & subordinate categories
What type of categories (superordinate/subordinate) inherit defining features?
Defining features of the SUPERORDINATE are inherited by the SUBORDINATE
What were Collins & Quillian’s (1969) key claims?
- knowledge is organised hierarchically in a semantic network
- the network has NODES (concepts) with CONNECTIONS (relationships) between each one
- there are 2 types of connection = ISA (is a member of) & FEATURE (is a defining feature of)
What happens when a node in the semantic network is activated?
Activation spreads through connections to other related nodes
What is the key argument of Classical Defining Feature theory? What does it mean?
Cognitive economy
E.g. since most birds ‘can fly’, this feature is attached to the general category of ‘birds’ (highest possible level) rather than to every individual bird node
Minimises processing effort & resources
What happens in memory when we make a judgement between 2 concepts?
When we make a judgement between 2 concepts, they become activated in semantic memory & spread out among the connections
At some point they will intersect, which will suggest a possible relationship
This will trigger a decision stage to verify the relation
What does spreading activation in the network allow us to do?
Make predictions - the time it takes to access info in the network depends on the distance that is ‘travelled’
We can test our predictions to assess the validity of this account to semantic memory
What types of sentence verification tasks can we use to test Collins & Quillian’s (1969) predictions?
- Instance category task
2. Instance feature task
What is involved in an instance category task (type of sentence verification task)?
Pps are asked
“Which leads to the quickest answer?”
a) Is a canary a canary?
b) Is a canary a bird?
c) Is a canary an animal?
How many levels in the hierarchy are involved in each question of the instance category task?
a) Is a canary a canary?
b) Is a canary a bird?
c) Is a canary an animal?
a) Is a canary a canary? - 0 levels in the hierarchy
b) Is a canary a bird? - 1 level
c) Is a canary an animal? - 2 levels
To which question do pps respond fastest to in an instance category task?
Pps say “yes” faster to a) than b), & faster to b) than c)
a) Is a canary a canary?
b) Is a canary a bird?
c) Is a canary an animal?
What is involved in an instance feature task (type of sentence verification task)?
a) Is a canary a canary?
b) Is a canary a bird?
c) Is a canary an animal?
Pps are asked
“Which leads to the quickest answer?”
a) Does a canary breathe?
b) Is a canary yellow?
c) Can a canary fly?
How many levels in the hierarchy are involved in each question of the instance feature task?
a) Does a canary breathe?
b) Is a canary yellow?
c) Can a canary fly?
a) Does a canary breathe? - 0 levels
b) Is a canary yellow? - 1 level
c) Can a canary fly? - 2 levels
To which question do pps respond fastest to in an instance category task?
a) Does a canary breathe?
b) Is a canary yellow?
c) Can a canary fly?
Pps say “yes” faster to a) than b), & faster to b) than c)
a) Does a canary breathe?
b) Is a canary yellow?
c) Can a canary fly?
Spreading activation explains what type of priming?
Semantic priming
What is semantic priming?
We respond faster to an item when we have already seen an item that is semantically related to it
Pps were shown word pairs & asked if both words were real.
Types of pairs they were shown = related real words vs. unrelated real words vs. nonsense words
Who did this study & what did they find?
Meyer & Schvaneveldt (1971)
Found that pps responded faster when the words were related real words
Collin & Quillian (1969) said that Classical Defining Feature theory have a number of exceptions. Give an example about ostriches being birds.
Ostriches can’t fly - they can have a separate feature that is linked to a particular object/etc.
What is a problem with Classical Defining Feature theory, according to Conrad (1972)?
Not all features are equal - some are more closely associated with a concept than others
e.g. robins have a red breast vs. robins lay blue eggs
What is a problem with Classical Defining Feature theory, according to Rosch (1973)?
Not all instances are equal - some are more typical than others
e.g. a robin is a more typical bird than an ostrich; an ostrich isn’t captured by the defining feature of a bird (flying)
What is the typicality effect?
We respond faster to typical instances of a concept than atypical
Smith et al. (1974) found that pps were faster to verify OSTRICH-BIRD or OSTRICH-ANIMAL?
Pps were feaster to verify OSTRICH-ANIMAL
When thinking about the hierarchy, OSTRICH should be closer to associate with BIRD than ANIMAL
Name some limitations of Classical Defining Feature theory.
X not all features are equal
X not all instances are equal
X it can be hard to find a set of features that defines membership & excludes non-members
X differences between categories should be clear-cut (i.e have the defining feature or not) but they aren’t
In Classical Defining Feature theory, are categorical boundaries clearly defined or fuzzy?
Boundaries between categories are fuzzy
People agree with others & consistent across sessions about typical & unrelated instances
People don’t agree about atypical instances
Who proposed the Feature Comparison model?
Smith et al. (1974)
What are the 2 types of features you can get in the Feature Comparison model
Defining features = define a concept (e.g. ‘bird’ - has a beak, feathers)
Characteristic features = some instances have them but not all; not essential (e.g. ‘bird’ - flying)
What sort of mechanism is used in the Feature Comparison model?
A 1-/2-stage processing mechanism
What is the processing mechanism in the Feature Comparison model for?
The 1-/2-stage processing mechanism is for judging the similarity between an instance & a category
What is involved in Stage 1 processing of the Feature Comparison model?
Rapid, general comparison of the defining & characteristic features of both concepts
ROBIN-BIRD > high similarity > YES (rapid response)
CHAIR-BIRD > low similarity > NO (rapid response)
BAT-BIRD > intermediate similarity > ?? (slower response, difference responses)
What is involved in Stage 2 processing of the Feature Comparison model?
Slower, more careful comparison of defining features ONLY (ignore characteristic features)
BAT - has teeth + fur
BIRD - has beak + feathers
= no match - a bat is not a bird
What are some pros of the Feature Comparison model?
- accounts for the typicality effect (faster category judgement for typical instances than atypical instances)
- copes with issues of organisation
How is the Feature Comparison model more effective than Classical Defining Feature theory?
Accounts for the typicality effect
How does the Feature Comparison model cope with issues of organisation?
We are faster to verify OSTRICH-ANIMAL than OSTRICH-BIRD
OSTRICH-BIRD - hierarchical levels = 1; stages of processing = 2
OSTRICH-ANIMAL - hierarchical levels = 2; stages of processing = 1
OSTRICH is more similar to ANIMAL (has skin, alive, breathes) than BIRD (can fly) so we make that judgement quickly
What are the limitations of the Feature Comparison model?
X can’t explain the fact that noun order changes response times in verification tasks
X doesn’t explain what constitutes a defining feature
X doesn’t explain how features are related to each other (e.g. wings + flying) or how we represent knowledge about relationships
Which researcher/s found evidence for the influence of noun order on response times in verification tasks?
Loftus (1973) found pps were faster to verify ROBIN-BIRD than BIRD-ROBIN
Word order shouldn’t change performance as judgements are based on the comparison of features
What is the Prototype Model of Semantic Memory about?
Category judgements aren’t based on similarity between an item & a set of features, but rather we form an internal central prototype which we can represent in our LT semantic memory
New instances are compared to the prototype & judged to be a member of the category if its features match those of the prototype
What are the pros of the Prototype Model of Semantic Memory?
- allows for the typicality effect
- allows for fuzzy categorical boundaries
- allows for family resemblance
How does the Prototype Model of Semantic Memory allow for the typicality effect?
Instances of a category vary closely to the central prototype
Typical instances are classified faster than atypical instances
e.g. ROBIN-BIRD is faster than OSTRICH-BIRD - robin is closer to our prototype of a bird than an ostrich so we make this judgement faster
How does the Prototype Model of Semantic Memory allow for fuzzy categorical boundaries?
Labov (1973) asked pps whether 4 pictures depicted a cup or bowl. Pps agreed that 3 pictures were cups/bowls but couldn’t agree on 1 picture.
People can have different prototypes of a concept depending on their previous experiences
People usually have little agreement about the point at which an intermediate item resembles one prototype more than another
How does the Prototype Model of Semantic Memory allow for family resemblance?
Category membership depends on similarity (to a prototype) rather than definition – it isn’t based on the presence of defining features
No defining feature is shared by all instances of a concept but all instances have some features in common with the prototype
What are some limitations of the Prototype Model?
X doesn’t explain ad hoc categories
X people know about relationships between attributes (e.g. wings + flying) which isn’t captured by the representation of a single prototype for individual concepts - model doesn’t tell us about relationships between attributes
X category membership judgements aren’t just based on similarity
Why doesn’t the Prototype Model explain ad hoc categories?
We can create new categories ‘on the fly’/ad hoc (e.g. things to sell at garage sale)
We are unlikely to have pre-existing prototypes for these categories but can still categorise things in this way
Which researcher/s found evidence that category membership judgements are not just based on similarity?
Rips (1989) gave pps 2 categories - 25-cent coin (fixed diameter) & pizza (variable diameter)
Pps imagined an object with a diameter between a coin & pizza
2 groups:
- “which category is the object most similar to?” (similarity judgement) –> pps were most likely to say ‘coin’
- “which category does the object most likely belong to?” (category judgement) –> pps were most likely to say ‘pizza’
–> we can think that something is similar to another thing but categorise it as something else
The Prototype Model doesn’t explain the dissociation between similarity & category membership
What are the difference/s between Prototype theories & Exemplar theories?
Prototype theory: concepts are represented by an abstract prototype; specific instances aren’t stored in semantic memory
Exemplar theory: concepts are represented by a collection of specific instances that we have previously encountered; we store these instances in semantic memory; category membership is determined by comparing a new instance to stored exemplars
Semantic categorisation may involve different approaches (varies for each situation)
What are the Prototype & Exemplar approaches best for?
Prototype approach: best for categories with many members
Exemplar approach: best for smaller categories/ones we have little experience of
Semantic memory isn’t just about simple categories, it also contains complex knowledge about events, actions & people.
What are these complex knowledge units called?
Schemas & scripts
What is a schema?
An organised body of knowledge (info, beliefs) about a topic/concept
May be inaccurate
What are person schemas?
General info & beliefs about a person’s traits & characteristics
What is a self-schema?
General info about our own traits, attitudes, abilities, goals, etc.
What research is related to self-schemas?
Kendzierski (1980) found that we are more likely to remember self-relevant info than info about others
Markus (1977) found that we are more likely to attend to info that fits with our self-schema than schema-incongruent info
What are scripts?
A type of schema
Describes knowledge that is common to a frequently-occurring event
Scripts only contain generic info. True/false?
True - doesn’t include all of the memories we have about a particular event, only the generic info (info about what happens normally)
What did Bower et al. (1979) find? (scripts)
Bower et al. (1979) found a high level of agreement between people when they were asked to list their actions (e.g. going to a restaurant)
Why do we have scripts & schemas?
- help us understand events & form expectations
- help us make sense of & draw inferences about info/events
- provide a short-hand in communication – don’t need to detail an event’s individual episodes
- helps retrieval/reconstruction of info about an event
What ‘thing’, based on schemas, can affect what info we recall?
Our expectations
Pps waited in an office, & then wrote down what objects were in the office
They recalled more schema-consistent objects than inconsistent objects, & falsely recalled schema-consistent objects that weren’t actually in the office
Who did this study?
Brewer & Treyens (1981)
Failures of what type of memory is commonplace among healthy people?
Semantic memory
Brown & McNeill (1966) studied the tip-of-the-tongue phenomenon. What is it?
When we are momentarily unable to recall info that we know is stored in our LT semantic memory
Strong ‘feeling of knowing’ – can recall some aspects (e.g. 1st letter) but not the whole word