Concepts and Categories Flashcards
Rosch said..
categorisation is a reasoned system determined by perceptual abilities and cognitive economy
Eysenck and Keene 2010
cognitive economy is not enough: trade off between economy and information –reach middle ground
name the 5 types
rule based, prototype, exemplar, cluster, multiple systems
Explain rule-based systems
Led by simple logical rules; if first rule doesn’t work the reject and move to next rule- based on hypothesis testing.
Rules are a condensed version of every item in lexicon.
example of a rule-based system
RULEX (nosofsky)
where do rule-based systems live
in PFC because depend on WM and exec function to maintain hyp testing
Evidence for rule based?
Able to verbalise rules and lesion studies show damage to system, good for socially defined categories (love)
Criticism for rule-based
not every category can be covered by one rule eg where would ‘game’ go? (Witgenstein)
Some members are not alike but should fit in same category
Explain Prototype models (Rosch 78)
average of all members=protoype, relies on family resemblance; bases on most simple level, unless necessary to specify more
Evidence for prototype models
- Naming at a basic level is better than subordinate levels (jolicoeur 84)
- using prototype is good for economy
Cai and nipola 84:
free recall task –prototypical members of category are quicker to remember
what do computational models show about prototype models? (rosch 76b)
natural language is matched on computer
- categories wiht more typicality were quicker
- no info stored on frequency but were sensitive to it
criticism of the prototype model
separable categories, context, semantic effect and expertise not accounted for (johnson and mervis 97)
Rosch 99: ‘prototype’ not objective concept more about many not just ‘one’ vary interpersonal level
Explain exemplar based models (komatsu 92)
Stores every item of training (good info bad economy), and novel word is compared to every member in history. New item is compare by mathematical pairwise comparisons to work out best fit.
evidence for exemplar
- sensitive to correlations, variability and overlap and accounts for typicality.
- works mathematically.
Anderson 10: new words will affect processing current items as all shift towards it
Storms 00: explain more than other models
criticism for exemplar
people show enhanced recogntion for items which violate key category rules but realistically this would’t change it so doesn’t work (sacamoto and love 04)
Reduced theoretical utility: no constraints on processing so can adapt to fit results
where do exemplar models live?
in the triathum and mid brain dopamine areas: for learning and strengthening associations
explain cluster models.
intuitive that similar items cluster together in memory.
regular: prototype, irregular: exemplar basis
evidence for cluster models:
computer simulations with accurate vocab sizes repeat results of humans (liu and croft 04)
What is Anderson’s Rational Model?
eg of a cluster; probability of novel item being from cluster is high - gets assimilated BUT if maths say probability is low then creates own cluster.
What is SUSTAIN?
eg of a cluster model; computer simulation, can be inclusive of selective attention– uses a lot of perceptual cues, model can learn which parts are most defining.
Able to direct attention to the optimal parts (active process)
general criticism
medial temporal lobes though of as controlling whole thing but actually studies are not convincing. HPC thought to bind whole thing together
Explain multiple systems model
no true model- a process supported by a variety of different systems
e.g. one system for selective attention, one for info strategies based on stimulus