Chapter 8 Flashcards

1
Q

you need ______ to have knowledge and you need knowledge to function

A

concepts

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is the definition possibility of understanding concepts

A

You have a dictionary like description of concepts.

- PROBLEM: even simple terms resist being defined and exceptions are common

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Wittgenstein proposed “family resemblance” to explain categories. Explain.

A

There are no defining features that every member has

  • but there are features that are common among most depending on the subgroup (characteristic features)
  • it is a matter of degree (the more characteristics the more likely it is a member)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is prototype theory?

A
  • the centre of a category = prototype
  • It is the average of all the members of the category you’ve seen
  • all judgements about members are made in reference to this ideal
  • different people have different prototypes
  • captures the family resemblance idea
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Graded membership?

A
  • category membership is a more or less decision

- objects closer to the centre prototype are “better” members than those far away

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What were the correlations between the sentence verification task, the production task and the rating task when investigating prototype theory.

A
  • Rxn times were faster for sentences with prototypical members
  • The members with fastest rxn time were also the first to be mentioned in the production task (naming birds)
  • The same members were rated as more prototypical (birdier)
    = conclusion: people perform these tasks by comparing to their prototype
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What happened when Rosch made participants generate sentences about a category?

A

if subject said “I saw 2 birds in a tree”
- experimenter replaces category name with either protoypical or non-prototypical member
- New sentences are then shown to new subjects and asked to rate silliness
- The non prototypical members generated the most silly sentences
CONCLUSION: participants are forming these sentences with the prototype in mind ( ie. certain members are privileged)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is basic-level categorization?

A

A level of categorization hypothesized as the “natural” and most informative level, neither too specific nor too general.

  • these categories are privledged (used most often)
  • Usually represented by one word (ie.chair)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is exemplar theory

A
  • similar to prototype theory in that it involves comparison to a standard
  • however the standard is an example of that category that you have seen in the past (differs b/w people)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

How would the exemplar theory explain the evidence that supports prototype theory (typicality effects)

A
  • verification of category being faster for typical members is because those examples are common and well primed in your memory
  • The same logic can explain the production task: you name the members that are most common in your memory
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

So which theory is correct, prototype or exemplar? explain why.

A

BOTH - each has an advantage

  • THere are instances in which the quick summary of prototypes is most useful
  • However exemplars provide the specific info that is lost
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Can the preferred type of knowledge can change for a person? can they be combine

A
  • yes you may have extensive knowledge for horses (exemplar) and general knowledge about cars (prototype)
  • your knowledge within a subject cna change: when you’re learning you think of specific examples (exemplar) then you switch to prototype and when you become an expert once again you can think of specifics
  • You also have the option to combine both
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

WHat experiment demonstrated that typicality and category membership can be independent of eachother. (even-ness)

A

Armstrong and gleitman asked participants to rate even numbers according to their even-ness. Despite all clearly belonging to the category “even” subjects were just as consistent at rating typicality as they were for more typical categories like ‘dog’

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Why did children agree that a toaster could become a coffee pot, but a skunk couldn’t become a racoon?

A

Because in the animal’s case is is not merely a function of having the relevant features - it has to do with the broader understanding of that category (genetics/inheritance)

  • It is these “deep features” that matter in this case
  • They reason differently for artifacts vs. natural kinds (animates)
  • same is true with counterfeit bills, and squished lemons
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

what do prototype and exemplar theory both have in common at their base

A

judgements of resemblance (either to a prototype or a remembered instance)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Judgements of resemblance is based on what shared features actually matter. How do we decide which are important and which to discard?

A

It depends on your general belief about the concept and it varies from category to category

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What are categorization heuristics?

A

strategies for categorization that give up accuracy for efficiency

  • Using typicality (resemblance to prototype/exemplar) will lead to correct categorization most of the time
  • The categorization heuristic emphasizes superficial characteristics.
    ex. resemblance
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What are explanatory theories? How do they influence your understanding of concepts?

A

= Implicit “theories” formed on beliefs and knowledge about a category and other related concepts - more holistic approach that affects:

  • Membership judgements
  • Possible new members
  • How fast we learn new concepts (theory gives a category coherence which makes it easier to learn)
  • serve the same function as a scientists theory though les precise
  • can be inaccurate
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

If participants are willing to infer that a new fact about a robin is also true of ducks but not vice versa what are they being guided by?

A

Typicality - subjects are more willing to extrapolate facts about typical members to the whole category than the other way around

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

If participants are willing to infer that a new fact about enzymes in a gazelle is also true of a lion but not vice versa what are they being guided by?

A

broader beliefs/knowledge about cause and effect
- in this case knowledge about the food chain: A lion eats a gazelle and is then likely to take on some of it’s properties but not the other way around

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

What is the difference between artifacts and natural kinds

A

Natural kinds: sharp boundaries and characteristics that are unchanging/predictable as they are essential in order to survive. (judged based on biology/chemical composition))

Artifacts: Man made objects. Fuzzy boundaries, characteristics can change and membership can be temporary

> people reason differently for each based on their cause and effect beliefs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

What neurological evidence supports the argument that separate brain systems are responsible for different types of categorical knowledge

A
  1. fMRI scans show that different brain areas are active when thinking about animate vs. inanimate things
  2. Brain damaged people can lose the ability to name only certain categories of things. Can be broad (can’t name nonliving things) or specific (can’t name fruits)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

What’s anomia?

A

inability to name common objects

24
Q

What brain area lesions lead to deficits in naming persons, animals and tools

A

Persons: temporal pole
Persons and Animals: infero temporal region
Tools: Lateral occipital region

25
Q

When thinking about various concepts what have brain scans revealed? what do these results mean?

A

activation in SENSORY and MOTOR areas
- means conceptual knowledge is intertwined with how the object looks, sounds or feels (sensory info) and how one might interact with it (motor/muscular knowledge)

26
Q

How do we travel through memory in our knowledge network to access information about concepts? what is the sructure of this system?

A
  • associative links don’t just tie knowledge together they represent knowledge (knowledge is stored in memory networks)
  • spreading activation travels from one node to the next, it is faster for closely related ideas and slower for distant ones
  • properties common among a category will only be linked with the subordinate category and not all it’s members (Ie has a heart will be linked to animals but not to cat, dog, squirrel ect.)
  • to know that a squirrel has a heart you travel from squirrel to animal to has a heart
27
Q

What did Collins and Quillian’s sentence verification task demonstrate about the knowledge network?

A

When shown sentences (either stating a concepts property or category) participants must judge if their true or false
- When two nodes are connected directly (travels from robin to bird) rxn time is faster for “is a robin a bird” than when the question requires an indirect, two step connection “A robin can fly” (travels from robin to bird to fly)

28
Q

What two results cause Collins and Quillian’s view to be incomplete

A

no explanation in their structure as to why:

  • it is faster to say a robin is a bird than an ostrich is a bird
  • people are faster to say a peacock has feathers than a sparrow has feathers
29
Q

However what does the overall trend of results in Collins and Quillian’s experiment prove

A

proves that associative links play a pivotal role in knowledge representation

30
Q

What are propositions?

A

The smallest units of knowledge that can be either true or false

31
Q

Explain anderson’s model of propositional Networks

A
  • Prepositions are represented as ellipses (circles?).
  • Associations connect these ellipses to the proposition’s constituents.
  • Associations are labeled to specify the constituent’s role within that proposition.
  • incorporates time and location nodes
32
Q

In anderson’s model individual ideas are represented with local representations, what is local representation?

A

information is encoded in some small number of identifiable nodes.
each node represents one idea , so that when that node is activated you’re thinking about that idea and vice versa

33
Q

In contrast to local what are distributed representations?

A

An idea is represented by a pattern of activation across the network and a single node may be part of the pattern of activation for multiple ideas
- also uses distributed processes so that one activation pattern can activate another - all occurring simultaneously and in parallel

34
Q

connectionist models are said to involve __________ processing

A

parallel distributed processing

35
Q

knowledge refers to a _____ rather than a ______

A

potential rather than a state

- because the connections are in place - and activation will flow if you’re thinking about it

36
Q

What are connection weights?

A

The strength of individual connections among nodes

-The greater the connection weight, the more efficiently activation will flow from one node to the other.

37
Q

How do PDP (parallel distributed processing) models detect patterns/ how do they ‘learn’?

A

learning involves an adjustment in connection weights such that a pattern of activation representing one idea can easily flow and activate the pattern representing the new connected idea

38
Q

PDP learning is made possible through powerful computing schemes called ______? How do they work?

A

“learning algorithms”

  • adjust connection weights through mechanisms governed by local considerations ( ie. do neighbouring nodes activate eachother, if yes connection is strengthened, if no it’s weakened)
39
Q

How does learning rely on feedback? What is an error signal? what is back propagation?

A
  • Feedback given to a network to indicate that the network’s response was not the desired one.
  • the magnitude of the error signal is proportional to the difference between the response produced and the response that should have been produced.
  • the signal causes the connection between these nodes to decrease
  • The signal can then be used to adjust the network or system (often via back propagation) so that the error will be smaller in the future.
  • this signal is transmitted backwards through the network called Back propagation
40
Q

summarize main points of chapter.

A
  • people have prototypes and exemplars
  • also have set of beliefs reflecting their understanding of cause and effect relationships
  • these beliefs are woven into a network that stores this info in memory
41
Q

What are the 5 functions of concepts?

A
  1. clasification
  2. understanding
  3. prediction
  4. reasoning
  5. communication
42
Q

category vs. concept

A

concept: a mental representation
category: set of entities or examples described by the concepts

43
Q

connectionism

A

an approach to theorizing about the mind that relies on parallel distributed processing among elements that provide a distributed representation of the information being considered

44
Q

parallel distributed processing

A

A system of handling information in which many steps happen at once (i.e., in parallel) and in which various aspects of the problem or task are represented only in a distributed fashion.

45
Q

define learning algorithms ? give exemple.

A
  • The learning algorithms involve changing connections weights between nodes and levels of the model according to whether the network is producing desired or not desired responses.
  • Over time, the networks “learns” to produce the desired responses because some connections are strengthened and others are weakened. There are lots of different kinds of learning algorithms (e.g. back-propogation).
46
Q

What did the Bruner cards tell us about how we form categories?

A
  • We are abstracting - looking at the attributes that are recurring and rejecting the ones that aren’t
47
Q

What is the critical attribute?

A

An attribute that is required in order for something to quantify as an instance of a concept

48
Q

what is the classic view of categorization?

A

concepts have defining features or attributes

49
Q

in what tv show are contestants trying to choose the most prototypical answer

A

family feud

50
Q

what is the vertical hierarchy of categories?

how can knowledge be generalized across levels? can these vary across people?

A

superordinate level: vehicle, animal, bird
basic level: car, dog, robin
subordinate level: toyota echo, dachshund
–> You apply facts from subordinate to basic (lab can’t have choclate = dogs can’t)
–> But not from basic to superordinate (lab can’t have chocolate = all animals can’t have choclate)
Yes, my basic vehicle may be a car but yours may be a motorcycle

51
Q

______ theory: how much does this example resemble the prototype for category X?

A

prototype

52
Q

______ theory: how much does this example resemble all other examples that I know belong to cateogry X?

A

exemplar

53
Q

What are two key factors effecting membership judgements in exemplar theory?

A
  • frequency is important - if you see one dog a lot (your own dog) you have many exemplars of it
  • recency: if you have recently seen a certain dog this memory comes to mind more easily
54
Q

WHat happened in the test involving medical diagnosis of dermological conditions

A

at practice given 30, at test given 60
-at test some were old and some were new
- of the new some were similar to practice examples and others (though the same category) were disimilar
results: doctors better at categorizing old examples and 10-20% better accuracy for similar vs. disimilar examples
(even after 2 week delay)

55
Q

What advantages does exemplar have over prototype

A
  • represents variability in categories
  • we dont base judgements on average alone, we retain info about particular cases
  • preserves info about correlated features
  • easier to change category description with new info
  • can create new diverse categories easily (ie. things you will regret tomorrow - there is not really a prototype but many exemplars)