Knowlegde Flashcards
Concept
Mental representation of some class of things or ideas Eg (skunks, liberals)
Category
A group of items or ideas that include examples of a certain concept.
items are grouped together and distinguished from other items
Eg food- fruit- apple- granny smith
Categorisation
Mental process where we decide what category a particular object belongs to
Why is categorisation useful
1 - communication: rapid communication using shared category knowledge eg don’t eat green potatoes they’re poisonous
2- inference: category items share properties = generalisation eg, humans and gorillas have carbunculoid feet so infer chimps do too
3- decision making: use past experience with category items to inform decisions with members of that group eg, had a bad experience with cheese so I avoid all dairy
Definitional categorisation
Consult a mental dictionary of concepts and apply definition to stimuli eg- dogs or wolf
Humans don’t type use as its problematic eg- how do we define a chair
may for abstract knowledge eg- define a triangle
Semantic dementia
Fronto-temporal dementia - bilateral temporal pole atrophy (wasting away of body part)
Was mistaken as language disorder
But aphasia (language loss) is 2nd to loss of knowledge
speech is fluent but lacks semantic content
Family resemblances
Wittgenstein - categorisation judgements are based on family resemblance between objects within a category
Categorisation depends on similarity judgements
Prototype model of categorisation
Assumes that people decide whether an item belongs in a category by comparison to a typical member (the prototype)
Assumes prototype is a single member of the category
Rosch’s work on prototypes
Prototypes are more likely an average of the typical category member
eg ‘Average’ faces are rated more attractive than the individual face - hypothesised because it resembles prototype closer
Typicality effect
We are faster to categorise typical members than atypical ones.
‘Typical’ is defined in terms of proximity to the category prototype
Category priming effect
Asking people to think about a category primes typical members more strongly than atypical members
Shows that categories help in retrieving knowledge
Exemplar models of categorisation
Calculating how typical a member is by comparing it to exemplars of the category
Exemplar is an actual member you have previously encountered
Easier at judging average of typical member
Also store exceptions to the rule = more flexible
Eg - pasta normally short - can be long and skinny
Why are categories helpful
Studies from typicality effect and category priming say they help us organise and rapidly access our knowledge of world
Hierarchies of categories
Not how brain works
More abstract to more specific categories eg - fruit- apples- Granny Smiths
Rosch - basic level is most informative
Amount of knowledge determines level centrality,
1- global (superordinate) eg furniture -loses lots info
2- basic eg table
3 - specific (subordinate) eg kitchen table - gains little info
Semantic networks
NETWORKS - Collin and Quillian
Nodes (categories/concepts) and connected by links (category membership)
Number of properties for each concept too
Spreading activation hypothesis
Activation of one node -reduces activation in the network that spreads to nearby nodes
Answer quickly if info is closely related
Answer slowly if needing to integrate distant parts of network
Move through property levels answers become slower response
Eg - canary sings, canary flies (birds), canary has skin (animals)
Evidence - lexical decision making task (Meyer and schvaneveldt 1971)
Limitations and strengths of hierarchical category organisation
Strengths; spreading activation and speed of information depends on network proximity
Limitations: cannot explain all human knowledge eg pig animal retrieved quicker than pig mammal
Non-hierarchical semantic network
Network links denote strength not category membership
Assume activation can spread along links
Amount of spread depends on strength of association
Connectionist models
Non-hierachical semantic network
Networks of layers (input, hidden, output)
Activation fed toward through layers depending on connection weights (excitatory or Inhibitory)
Concepts are represented by patterns of activation
Connectionist models and training
Connectionist weights underlie patters emerge via training
Learning algorithm is backpropagation
- debate it brain has backpropagation
Connectionist model advantages
1-fairly robust to damage
2- explains generalisation to new categories
3- achieve super human performance
Neural modularity
Representation of knowledge in the brain is modular.
Neural representations of different types of knowledge are represented in different parts of the brain
Eg - faces in fusiform area and places in parahippocampal
Neural modularity evidence
Transcranial magnetic stimulation (TMS)
Depending en location TMS can disrupt knowledge in specific ways
Semantic dementia is the difficulty
Retrieving knowledge from specific level of category knowledge (global, basic, specific)
Eg - camel no humps but looks like animal
Semantic dementia brain region
Neural decay or damage in temporal pole (region within anterior temporal lobe)
Suggests region is important for retrieval of knowledge at basic and specific level from other area
Hub-and-spoke model
Concepts formed by interactions of modality-specific sources info (spokes) and a central hub that coordinates spokes and has modality - invariant info
Spokes- sounds, verbal descriptors visual features
Semantic hub proposed location in anterior temporal lobe
Hub-and-spoke evidence
TMS to inferior parietal lobe inferfers with naming artifacts
TMS to anterior temporal lobe interferes with naming artifacts and living things
