chapter 9: concepts, categories & knowledge

concepts

-concept: a mental representation that groups or categorizes objects, events, or relations around common themes
-e.g., concept for "dog" = four-legged furry animal that wags tail and barks
-categories: include all possible examples of a particular concept
-e.g., parties:
-housewarming party
-pool party
-Halloween party
-family party

categorization

-we constantly form new categories
-e.g., playlists for music (by artist, by genre, by decade, by mood, etc.)
-has a biological basis, as demonstrated by:
a) fMRI studies
b) people with brain lesions

-evolved mechanism - confers a survival advantage
-e.g., making categories for prey vs. predator; edible vs. poisonous foods; etc.

the three ways categories are formed

1)based on gross or fine appearance
-gross = all dogs
-fine = poodles, labs, collies, etc.

2)based on functional equivalence
-e.g., pens, pencils, crayons
-things you can write with

3)categories that address particular situations
-e.g., things you would try to take out of house if on fire

1. defining attribute model

-concepts characterized by a list of features necessary to determine if an object is in a category
-problem: we make all kinds of exceptions
-e.g., some dogs have 3 legs instead of 4
-e.g., some mammals live underwater

2. prototype model

-membership in category determined by comparing object to best example for that category
-prototypical: bird
-non-prototypical: turkey
-graded membership: the idea that some objects in a category are closer to prototype than others
-ones closer to prototype share family resemblance
-some birds (cardinal, robin) are "birdier" than others (penguin)

prototype model: Smith et al. (1974)

-sentence verification task: press button to indicate true or false
-typicality effect: sentences with prototypical items tend to be verified faster
-e.g., "apple is a fruit" faster than "avocado is a fruit"

prototype model: Rosch et al. (1975)

-Primed with name of colour​

-Ps saw two circles; either same colour (good examples or poor examples) or different​

-Decision: same or different​

-Results: Ps faster to say “same” when colours were good examples (RT = 610 ms) than poor examples (RT = 780 ms)

-Conclusion: prototypical category members affected more by priming

basic-level categories

- we tend to use basic level categories when identifying objects
-experts more likely to use a specific level
-and some situations require it (e.g., ordering a coffee when there are many different kinds available)

example:
global - fruit
basic - pear
specific - Anjou

3. exemplar model (Medin & Schaffer)

- any concept has no single best representation;
-make category judgment by comparing new instances with stored memories for other instances
-e.g., you see a dinosaur-like creature you've never seen before, and compare it in your mind with examples of other dinosaurs you've seen
-if you find a good match, decide that creature is also a dinosaur

what are the processes we rely on?

1. trigger of a memory
2. judgment of resemblance
3. conclusion based on resemblance

exemplars are in what brain area?

prefrontal cortex

prototypes are in what brain area?

visual cortex

category-specific deficit (anomia)

-an inability to recognize / name objects in a particular category (but ability not impaired for other categories)

problems

1. some objects show no resemblance (to prototype or exemplar) or are atypical but we still identify as part of category
-e.g., Justin Trudeau is atypical in that the typical Canadian is not the Prime Minister, yet we still easily classify him as being a Canadian

2.other things show high resemblance but are fake
-e.g., a person wearing a highly realistic costume / disguise
-e.g., a fake watch or purse

beyond resemblance

-judgments guided by your sense of what's essential for a category
-the importance of a feature / attribute varies from one category to another, according to your beliefs about what matters
-e.g., Being P.M. / not being P.M. is not an essential feature of being Canadian
-depend on knowledge / understanding
-e.g., of biological inheritance
-e.g., do a dog and a cat belong in the same category?
-furry four-legged mammals, house-pets: yes
-same species? no.

concepts as theories

-concepts are like schemas
-they allow people to form generalizations (i.e. apply general knowledge to new cases you encounter)
-e.g., new dog will behave similarly to previously encountered dogs

inferences based on theories

-generalization more likely from typical cases
-e.g., argument 1:
-Premise: Robins have a high concentration of potassium in their bones​
-Conclusion: All birds have a high concentration of potassium in their bones

Argument 2:​
-Premise: Penguins have a high concentration of potassium in their bones​
-Conclusion: All birds have a high concentration of potassium in their bones

-conclusion to argument 1 is more valid

knowledge networks

-knowledge is represented via a vast network of connections and associations between all of the information you know

Collin & Quillian's Hierarchical model

-it’s a hierarchical model, because it’s arranged so that more specific concepts like canary and terrier are at the bottom, and more general concepts are at higher levels (birds, cats, dogs) ​

-by moving up the hierarchy from bottom to top, following the lines, you can determine all the properties of an object​

-e.g., a Cheshire is a type of cat, so it has claws and purrs, and it’s an animal, so it has a heart, eats food, and breathes

-Criticisms​

-Doesn’t account for typicality effect!

definition of cognitive economy

-(storing shared properties just once at higher level) makes network more efficient​

propositional networks

-Abstract knowledge represented via time and location nodes​

knowledge networks

propositional networks
-localist representations - each node is equivalent to one concept

connectionist networks
-distributed processing-information involves a pattern of activation
-parallel processing of information occurs at the same time
-changes in the connection weights or strength of connections
-A better proxy of how brain operates than is basic hierarchical model ​
-Partial damage to network doesn’t completely disrupt operation of network