Semantic memory Flashcards
semantic memory
- facts and general information
- acquired over multiple learning episodes
- doesn’t include information about the context and event in which the knowledge was acquired
-Theories of categorization
- categorization is when you want to know what something is
- Classical theory
- exemplar model
- prototype model
Classical theory of categorization
- rule based
- hierarchical
- does it fit the rules
category
combination of defining features
problems with the classical theory of categorization
- difficult to come up with defining, unique features
- sometimes critical/ defining features don’t exist
- doesn’t account for goodness of fit (wouldn’t expect variation in goodness of fit in classical categorization)
goodness of fit
-rate how good of an example this is for a category
Exemplar model
- store every example of category
- semantic memory becomes a by product of stored episodic memories of every encounter with an item in the category
- compare probe to every exemplar and assess similarity
- goodness-of-fit is proportional to the average similarity to all exemplars of category
- definition of category is the average of all exemplars
prototype model of categorization
- extract and store prototype (central tendency) of category separately from exemplar
- compare probe to prototype
- goodness-of-fit: how similar is the probe to the prototype
- definition of category = prototype
- doesn’t account for variability or for items that don’t fit well into prototype
summary of different theories of categorization
-classic theory:
- doesn’t account for goodness of fit
-exemplar:
-never store prototype
-compare new items to exemplars
- only requires one memory system (episodic)
prototype:
-store prototype separately from exemplars
-requires two memory systems (exemplar and prototype)
Exemplar vs prototype: Posner random dot pattern task
-show people dots that slightly vary and learn to cateogrize as A or B
-never see prototype or unstudied exemplars
-test; show studied exemplars, unstudied, and prototype
Result:
-classification of prototype is more stable over time
- initially studied exemplars more accurate
-after delay, prototype categorized more accurately than studied exemplars
-prototype extracted during learning delay and decays more slowly than memory for individual exemplars
-suggests prototype memory is separate from exemplar memory
-exemplars are classified better than unstudied
take away from posner random dot task
- we use exemplar and prototype
complementary learning systems theory
- hippocampus and cortex as two interacting systems
- (because remembering individual episodes and extracting central tendencies are incompatible goals)
role of hippocampus in complementary learning system
- fast-learning, episodic (exemplar representations)
- rapidly binds together information from neocortex to remember episodes
role of cortex in complementary learning system
- slow learning
- semantic
- central tendencies, prototypes
- neo-cortex slowly binds information together over many experiences (not hippocampal reinstatment)
- eventally central tendencies are represented without aid of hippocampus= semantic memories
Multiple trace versus standard consolidation
- hippocampus is always needed for true episodic memory
- memories spared in retrograde amnesia are autobiographical semantic memories
- episodic vs semantic isn’t events vs facts, it is reliving vs knowing (specific events vs generalizations)
