Categorisation Flashcards
Definitions of concepts and categorisation
concepts:
> mentally possessed ideas and notions
categorisation:
> set of entities that are grouped together
Why is categorisation so important for human?
if categorisation doesn’t exist, humans will be overwhelmed by different information
what are the benefits of categorisation
flexibility
> every item does not belong to a single group (the categorisation depends on the goal)
composition
> humans can form new concepts from existing concepts
identification
> humans sometimes rely on categorisation to identify unfamiliar or ambiguous objects
generalisation
> categorisation helps us generalise objects
> If the two objects are plotted far apart on the MDS (i.e., they are unsimilar) = lower probability of being generalised)
organisation
> helps humans to reduce the complexity of input of information
> there are 10,000 colours that human can perceive, but only 10 names of colour
what is the order of categories that are categorised based on physical similarities
natural categories > man-made artefacts > ad hoc categories > abstract metaphors and schemas
what is an ad hoc category
categories that are set according to the goal of the items
what did Eleanor Rosch propose?
categories are not determined by rules and principles but by structures
what are the experiments and the findings
Experiment on naming attributes
non-biological:
> superordinate category: less common attributes
> basic category: massive jump of common attributes
> subordinate category: slight increase
biological:
> no big differences (high starting point)
> However, results are consistent if the superordinate = basic
(findings: most information contain in basic level)
Experiment on shape similarity
> outline of figures in the basic level = quickest and most accurate to be identified
Experiment on naming
> People tend to name the basic level name of the object
Experiment on L1 learners
Non-biological items
> items in the basic category = learnt the quickest
Biological
> superordinate category items = quickest
Describe the classical view of categorisation
> categories are set by the definition of the items of same necessary and sufficient conditions
every item are equal
Explain the “set-card experiment” and the two categorisation strategies
research
> show stimulus to participants and have them to come up with the categorisation criteria
> each time = feedback
results
> came up with two strategies that participants used
> scanning strategy
> every pick = testing the hypotheses of the criteria in their mind
successive scanning (1 hypothesis at a time) and simultaneous scanning (all hypotheses tested at a time)
> focus strategy
every pick = testing whether the attribute fits the criteria
conservative focus (changing one attribute at a time) vs focus gambling (changing various attributes at a time)
What are the supporting evidence (LLC) and challenging views on the classical view
supporting
> strategies mentioned seemed to be valid
> more rules = more complex to learn the criteria > take more time and less accurate (correlation between complexity and accuracy tested = Learning Logical Concept)
challenging (all in natural items)
> some items seem to be more typical than others
> in the hierarchical structure of categorisation, some levels seem to be more basic
> categorisation definitions are not valid for natural items
Explain typicality, the relationship between typicality and learning, and correlation between typicality and the number of shared attributes
Typicality
> the degree to how much an object to perceived into their category
> The higher the typicality is, the quicker it is categorised
relationship between typicality and learning
> the higher typicality of the the learnt item is, the easier it is to transfer the knowledge to other items
correlation between typicality and the number of shared attributes
> the more typical the item is, the more shared attributes there are with other items in the category
explain how does typicality affects generalisation
> usually performed with a category induction task
induction: generalising particular to general
- category induction
e.g., spinach = height growth > vegetable = height growth (than pumpkin) - category induction: conclusion example
e.g., spinach + lettuce = height growth > cabbage = height growth (than pumpkin = height growth in the latter statement) - category induction: variability size
e.g., spinach + lettuce = height growth > vegetable to height growth (than food = height growth) - category induction: example variability
e.g., spinach + pumpkin = height growth > vegetable = height growth (than spinach + lettuce)
Explain the prototype model
> by generalising most prominent features of the items in the items in the group > comes up with a prototype model > determines whether the item should be categorised according to the level of similarity to the prototype model
Explain the experiments that support the prototype theory and also the Prototype Enhancement Effect (PEE)
Dot experiment:
> prototype into high/low similarity models
> learn five low similarity models
> show multiple stimuli (including prototype) and ask whether should categorise
> ~80% can
The 5-4 experiment:
> prototype category A: 1111; category B: 0000
> stimuli provided in random fashion > identify which category > feedback
> show 7 transfer items including the prototype of A
> prototype A = highest rate of identified into A; prototype B = lowest rate of identified into B
Prototype Enhancement Effect:
> Prototype of a category has the highest categorisation rate even the stimulus is not shown)
What is the criticism of prototype model and what is the alternative theory that can explain the issue
cannot explain non-linear categories
exemplar theory
> category decision is based on retrieval of previous items and the degree of similarity of the new item to previous items in the category
> can also explain the prototype enhancement effect
> can predict rule-like performance
