Hps203 Chap 8 Concepts

Front 1

why do our definition of basic concepts like 'dog' often fail?

Back 1

because individual dogs may vary and not necessarily fit our definition - our understanding of concepts is not really based solely on mental definitions but can be quite abstract (e.g. game, Ludwig Wittgenstein)

Front 2

definition for categorie

Back 2

certain attributes within a set of boundaries

Front 3

explain the meaning of the term 'family resemblance'

Back 3

members of a category have family resemblance or at least some 'characteristic features'.
many concepts have the same character, with many features shared among the instances of the concept, but no features are shared by all of the instances.

Front 4

describe prototype theory

Back 4

a prototype is the IDEAL or AVERAGE center of a concept

(Norton: the claim that mental categories are represented by means of a single “best example,” or prototype, identifying the “center” of the category. In this view, decisions about category membership, and inferences about the category, are made with reference to this best example, often an average of the examples of that category that you have actually encountered. )

Front 5

what is meant by graded membership

Back 5

membership of a category is not merely a matter of yes or no but rather a more or less in comparison to the category's prototype (some dogs are 'doggier' then others)

Front 6

how can the prototype notion be tested?

Back 6

by sentence verification task and/or production task (also in combination)

Front 7

sentence verification task

Back 7

measures response time for sentences like "a penguin is a bird" versus "a robin is a bird" (robin is faster because it's closer to the prototype of bird)

Front 8

production task

Back 8

asks people to list as many members of a category as possible (e.g. how many bird), first answers will be closest to prototype

Front 9

tasks converge
(sentence verification & production)

Back 9

i.e. first answer in production task will also have quickest response time in sentence verification task i.e. doubly asserts prototype theory

Front 10

rating task

Back 10

ranking members of the species according to how 'biddy or doggy' they are
( e.g. 1. robin to 15. penguin)

Front 11

what are basic level categories

Back 11

a 'natural', most informative level of categorisation, neither too specific nor too general (e.g. chair, rather than upholstered armchair or item of furniture)

Front 12

what are exemplars

Back 12

similarly to prototypes they represent a standard for a certain category, however, not the ideal but just ANY YOU CAN THINK OF

Front 13

explain how we rely on both exemplars and prototypes in our thinking about categories

Back 13

they allow us to decide what category a new 'object' belongs to, prototypes allow this to happen on the basis of a quick summary but exemplars allow us to 'fine tune' our thinking (gift for a 4-year-old who likes sport but recently broke his arm)

Front 14

describe a situation in which categorisation and resemblance do NOT go 'hand in hand'

Back 14

the painted, sweetened, run over lemon
i.e. it is still a lemon due to the fact that it grew on a lemon tree, no matter how we abuse it

Front 15

what else do we use to make category judgements over and above typicality and how do we know?

Back 15

is is natural or human made
is it authentic
example: what makes a doctor (not the having parents who are doctors)? what makes money counterfeit (not issued by the government)? what makes a prayer counterfeit (not the correct wording)?

Front 16

define heuristic strategies

Back 16

a mental short cut, a way of quietly if not always 100% correctly form judgement or make decisions

Norton: A strategy that is reasonably efficient and works most of the time. In using a heuristic, you are in effect choosing to accept some risk of error in order to gain efficiency.

Front 17

describe the explanatory theories that people hold and why we need them for categorisation

Back 17

(not sure about this)
ET allow us to see behind the definitions or prototypes/exemplars of a category, they are what allow us to still see the abused lemon as a lemon

Front 18

how do explanatory theories affect our learning and the inferences that we make?

Back 18

(not sure about this)
theories allow you to apply knowledge of something you already know to a new concept/situation (e.g. hammer theme applied to 'something metal and easy to grasp').

Front 19

why is categorisation important
(txt book, p 307)

Back 19

it allows you to apply your general knowledge to new cases you encounter and conversely, to draw broad conclusions from your experiences.

Front 20

what are some of the different profiles that people have for difference concepts?

Back 20

people may think about different concepts in different ways:
1. natural kind vs. artefacts (man made)
2. concepts with sharp boundaries vs fuzzy (woman vs stamp collector)
3. sensory vs muscular knowledge about the thing

Front 21

how is knowledge stored?

Back 21

knowledge is stored in long-term memory in form of a network, network connections or links are a constituent of the knowledge

Front 22

how is knowledge retrieved?

Back 22

via activation spreading, from node to node

Front 23

describe Collins & Quillian (1969) experiment and what it tell us about the knowledge network

Back 23

sentence verification task (e.g. a robin is a bird, cat have hearts) measuring participants response times
result: longer response times for concepts with more distal relationships e.g. cat = animals = have hearts

Front 24

what are propositional networks?

Back 24

theory by Anderson:
similar to standard knowledge network theory, nodes are connected by associative links, more frequent use makes for stronger connection, activation spreading applies. DIFFERENCE: knowledge is represented as propositions rather then mere links
(I'm actually a little fuzzy on this)

Front 25

what are the differences between a network with local representations and a connectionist network?

Back 25

In local representation one concept causes one node to fire (this applies in Anderson's propositional or standard knowledge networks). this then spreads via activation spreading.

In connectionists network one concept causes multiple nodes to fire (called distributed representations). Spreading and activation of connected concepts happens via parallel distributed processing (PDP) which again involves multiple nodes per concept.

Front 26

what is parallel distributed processing?

Back 26

in connectionists network one concept causes multiple nodes to fire (called distributed representations). spreading and activation of connected concepts happens via parallel distributed processing (PDP) which again involves multiple nodes per concept, they fire at the same time (in parallel) and often single nodes are connected to multiple clusters for different concepts

Front 27

how does learning occur in a connectionist network?

Back 27

via different connection weights caused by learning algorithms: e.g.
Hebbian learning (cells that fire together wire together) (Q is usually followed by U) frequent connections makes 'connection weight' heavier i.e. stronger
contrary incorrect connection trigger error signals which work their way backwards through 'back propagation' and decrease the connection weight (you got it wrong, i'm not listening to you anymore)