wk10 - Concepts & Categories Flashcards
What is an equivalence class in relation to categorisation?
- Categorisation: the ability to form equivalence classes of discriminable entities.
- Categories based on hierarchy - animals - dogs - labradors.
- Not bound by perceptual features - i.e. “pets” - ad hoc categories..
What are the benefits of categorisation in regards to:
- survival,
- identification,
- Reduction of the complexity of the environment & the organisation of knowledge.
- generalisation
: Provides a basis of deciding what constitutes appropriate action.
- e.g. E. coli encounters a stream of molecules that it categorizes as a nutrient, it suppresses tumbling and swims upstream to the nutrient source.
Means of identification: We feel as though we can recognize a pattern when we can classify it into a familiar category such as letter A.
Reduces Complexity & Organises Info:
- world colour survey - researchers studied 110 different languages, none of which had a written component - 6-10 colours identified.
Generalization: We do not have to be taught about novel objects if we can classify them; we can use our knowledge of items in the category to respond to the novel object - generalisation.
- Bird Task: The probability of generalising a category drops off quickly the further away we get from the first referent (the robin).
How do category induction tasks work? And what is induction?
Induction is generalising form the particular to the general
- given a set of examples, what is the general conclusion that one could draw.
A typical task:
- The premise statements above the line are taken to be true, and the task is to judge the degree to which the conclusion statement, below the line, follows from the premises.
- Essentially this is a judgment of argument strength.
What is the typicality of instances effect for category induction?
Typical instances are more strongly related to the category & so allow for greater generalisation.
- e.g. Robins have higher potassium = all birds have high potassium vs Penguins have higher potassium = all birds have high potassium. More likely to assume robins are typicality.
What is the typicality of generalisations effect for category induction?
Generalisation is greater to more typical category members.
- e.g. Robins 5HT as NT + Bluejays 5HT as NT = Sparrows do it..
- Vs. Robins 5HT as NT + Bluejays 5HT as NT = Geese do it..
What is the effect of conclusion category size?
Generalisation is greater to more specific categories.
- e.g. Bluejays require K for liver function + Falcons require K for liver function = All birds do.
- Vs. Bluejays require K for liver function + Falcons require K for liver function = All animals do.
What is the effect of premise example variability on generalisation of categories?
Generalisation is greater when the examples are more variable.
- Hippos & Hamsters have higher sodium than humans = all mammals higher than humans.
- Vs.
- Hippos & Rhinos have higher sodium than humans = all mammals higher than humans.
How did Eimas & Quinn show that early development of rudimentary categorisation in infants? (hint: Horse Category habituation task)
- Habituated infants to images of different horses (horsey-ness category) compared to novel stimuli including - cats, zebra, & giraffe.
- 3-4-month-olds fixated on the new categories more often than new horse image
- Cats > 60.1%
- Zebras > 62.1%
- Giraffes > 57.2%
- i.e. Increased looking time for novel categories suggests categorisation of horses.
How is adult categorisation different from children’s?
- Adult categories are abstract, based on unobservable attributes, relational concepts, & rules.
- Children’s are based on perceptual grouping.
In Shepard, Hovland & Jenkins (1961) category learning difficulty task, what feature of a category causes the greatest learning difficulty?
In their Category Type Learning paradigm (Type 1-6) which category types were easy to learn and which were hard? Why?
Category learning difficulty depends on how many different dimensions are used to define the category.
e.g. Type I categories are defined by only one dimension - colour - therefore they’re easy to learn.
Whereas Type 3, 4, 5 categories are defined by Rules + exceptions - i.e. All black except small white triangle. Although learning eventually takes place.
Type 6 was defined by individual dimensions, therefore no rules can be used to simplify learning & were the hardest to learn.
How did Smith (1989) show the development of one-dimensional grouping over the lifespan? What is the general pattern of object grouping from young childhood to adulthood? (similarity > one-dimensional).
Children tend to sort on the basis of overall similarity across all dimensions.
Adults sort stimuli on the basis of a single individual dimension of those stimuli.
Shown with 3 tested conditions:
- (D) Discriminable condition: overall degree of similarity increased by making colour & size difference smaller.
- (S) Standard condition: two objs same size diff colours, one bigger and diff colour.
- (E) Extreme Condition: overall degree of similarity decreased.
- For the S task 2-year olds tend to put all items together in one group. When 3 are very different - group by similarity or in separate groups.
- 4 & 5-year-olds group by similarity when objects but tend to 1-D group when objects are far apart.
- Adults full 1-D grouping.
- Appears 1-D grouping takes time to develop.
How does eye-tracking data reflect attention for the different types of categories (Type I - V categories)?
- Number of dimensions fixated in Type I categories approaches 1 after just a few trials.
- Number of dimensions fixated in Type VI categories remains at 3 for trials up to 27+.
What are causal models and how do they help with categorisation?
Knowing the cause underlying the category provides an additional “deeper” dimension or feature that can be used to understand the category.
In relation to the Comparative Mirror hypothesis for photorealism painting, what argument can be made that this approach provides a sound model which translate to Causal Theory of Categorisation?
- CM hypothesis produces data which looks like data we want to explain - paintings match.
- Explains several well-known facts about Vermeer’s painting - photorealism, no tracing.
- Allows generalisation & prediction - curvature of lens cause curve in painting/use of lens in experiment produced curve.
- Causal Theory of Categorisation
- Existence Proof that use of comparative mirror could be have used to produce painting.
- Suggests photorealism was result of optics.
