Please enable JavaScript.
Coggle requires JavaScript to display documents.
Concepts - Coggle Diagram
Concepts
Prototypes
-
Family resemblance - to categorise
- Shared attributes, common features
- But all features not shared with all members
- Ideal: everyone shares some, but not all, with ideal
Prototype
- 'Best' example, average, 'centre' of category
Sentence verification task
- Closer to prototype -> recognise faster
- T or F: "Robin is bird" > "Penguin is bird"
Production task
- Think of prototype -> name prototypical first
- Name as many birds as possible -> birds faster in SV task named earlier
Ratings
- Use resemblance to prototype to judge
- How 'birdy' -> more birdy if closer to prototype
-
Basic-level categories
- Natural, informative - not too general or specific
- Learn earliest, use most, often 1 word
Categorisation
Categorisation may be independent of typicality
- e.g. in category, but not typical (teen = Greta)
- e.g. typical, but not in category (counterfeit $ = $)
Category judgements
- Use deep properties
- Based on beliefs about category
- -> tells you what is relevant
- If objects shares important/essential features with prototype/exemplar => in category
Beliefs
- Affect learning: can reason using cause-and-effect, can help learn concepts quicker
- Allow inferences: from typical to whole category, not atypical to category; tells how concepts related
Profiles for concepts
- Artefact / man-made: can be changed
- Natural: can't be changed (relatively stable)
Plus consider:
- Goal, e.g. lose weight
- Relational, e.g. predator vs prey
- Event, e.g. date, shopping
Man-made vs natural
- Use diff brain areas: non-living = functional, living = visual
- Non-living = less stable/homogenous, living = stable/homogenous
- Reason differently: man-made can change
Boundaries
- Formal: e.g. triangle, uncle
- Often indistinct: e.g. bachelor, dog
Network Types
Propositions
- Smallest unit of knowledge
- Knowledge stored as propositions
- i.e. agent RELATION object
- e.g. dog CHASE cat
Local representation
- 1 idea = 1 node
- e.g. propositional network
Distributed representation
- 1 idea = many nodes
- Ideas: pattern of activation across network
- Learning: pattern of nodes, activates other patterns of nodes
Parallel Distributed Processing (PDP)
- Many steps at once, knowledge distributed
- Good for detecting patterns & generalising
Exemplars
-
Similar to prototypes: used in same way
- Does object resemble exemplar/prototype in memory?
- Use to 'tune' concepts to match circumstances
- Mix of knowledge is person-specific - depends on knowledge
Different to prototypes
- Exemplar based on knowledge about specific members; Prototypes based on general/typical info about category
- Exemplar used when members distinct;
Prototypes used when categories overlap
Knowledge Network
Knowledge
- Stored: in network, as associative links
- Retrieved: spreading activation, via links
Structure of network
- Non-redundancy: knowledge stored at highest level
e.g. Collins & Quillian: sentence verification
- Quicker if closely related ideas
- e.g. canary 'is canary' > 'is bird' > 'is animal'
- e.g. canary can sing > canary has skin
- BUT exceptions
- Faster if closer to prototype (penguin vs robin for 'is bird')
- Faster if feature prominent (peacock vs sparrow has 'feathers')