L18 - Beyond Similarity Flashcards
What are the limitations regarding using similarity ratings to analyse how we create and use categories?
it is not always clear what the underlying similarity ratings used in the model
is it based on biological similarity, behavioural similarity?
What did Heit and Rubenstein (1994) look at in their experiment?
What type of similarity do we use for our categories
What did Heit and Rubenstein (1994) find in their experiment on what type of similarity we use in induction?
When the predicate was biological, induction was stronger for biologically similar animals
When the predicate was behavioural (i.e., foraging patterns) the induction was strongest for the behaviourally similar animals (e.g., tuna and whales).
What do the results of Heit and Rubenstein (1994) suggest that similarity is dependant on?
What implications may this have for the previous research that has been done?
Similarity is context dependent
this dependency may not be captured in empirical data such as pairwise similarity ratings or featural overlap
Are there any situations when similarity does not appear to play a strong role in induction?
Yes
Lin and Murphy (2001) found that in some situations induction was stronger for thematically related categories than for similar categories
For the category induction paradigm, what is the problem with using blank predicates?
The predicates are not entirely ‘blank’.
We may be relying on deeper knowledge to make our decision, not just the similarity structure of the category
What is issue with replacing ‘blank predicates’ with phrases such as “cats have property X”?
It is not similar to how we see categories in real life
Real-world properties often contain meaningfully structured information.
• Given that we want to understand inductive reasoning in
the real-world, it places researchers in a difficult position
regarding internal and external validity.
What effect might explain the reversal of the assumption of premise monotonicity in the Heusen et al experiments?
Higher-Order Knowledge Effects
Knowing there is a difference between Brahms (classical music) and ACDC (rock music) changes the likelihood of extending from the premise to the conclusion
What knowledge effect did Voorspoels et al. find that influences whether negative evidence can lead to premise monotonicity being violated?
Whether you believe the evidence is ‘helpful’ or ‘random’ can influence the liklihood of you generalising from one situation to another.
- If they believe it is coming from a ‘helpful’ or ‘knowledgeable’ source monotonicity is violated*
- –If they believe the information is simply being randomly sampled from the environment monotonicity holds.*
If we recieve negative evidence from a source that seems random, what effect will this have on the premise monotonicity?
No effect
If we recieve negative evidence from a source that seems helpful, what effect will this have on premise monotonicity?
premise monotonicity is violated
higher-order knowledge effect
Explain the Rips, 1984 coin/pizza example and how it reveals higher-order knowledge effects
if you present 2 circles of different sizes (a, b), then a third in between - 50% of people will say it belongs to category a or b
If you say that the small one is a coin and the large one is a pizza, majority will say pizza, as a coin can’t be that big
This happens because we know that you can’t get coins that big but you can get pizzas that big
What did Lin & Murphy’s (1997) “Tuk” experiment show about higher-order knowledge effects?
That if participants were told an object was used for hunting or it was instead for gardening would depend if we would categorize similar items in the Tuk category.
Our category generalization depends on our higher-order knowledge regarding hunting or gardening and how useful the tool would be.
What conclusions about similarity stem from the Lin & Murphy (1997) “Tuk” experiment?
categorization decisions are not simply based upon similarity
if it was only based on similarity, there should be no difference in how we categorize if we are shown similar objects
What is the issue of using lab-based category learning experiments that use categories that are fairly arbitrary (colour, size, shape etc.)
The features associated with real-world categories tend to have meaningful associations
It appears that our knowledge of the way these features interact influences our ability to learn category structures (Murphy & Allopenna, 1984)
What relationship does meaningful integration have with category membership?
It is much easier to learn categories which have a meaningful integrated relationship than categories that are arbitrary
Things to know for the exam
contrast category effect - he references his paper would be good to understand
