Week 8 Knowledge Flashcards

1
Q

what kind of system of knowledge do humans use

A

distributed

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

what is a concept

A

mental representation of an object, event or pattern

decreases the amount of info to learn

allows us to make predictions

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

category

A

class of things that share a similarity

concepts are organized into this

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

Definitional Approach (forming concepts/categories)

A

to belong to a category you need necessary/sufficient features

absolute (can make it hard to find specific features)

rigid boundaries

all members are equally good examples

learning involves discovering defining features

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

the two types of probabilistic theories

A

prototype

exemplar

both based on experience

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

prototype theory

A

categories made on an idealized average

the prototype

prototypes have high family resemblance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

exemplar theory

A

idealistic representation

category decisions based on all of the examples stored in memory

more flexible as you generate a prototype for the situation you are in

takes into account atypical cases and explains the typicality effect

easily deals with variable categories/allows overlap

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

if something is more similar to the prototype are you faster or slower to categorize it

A

faster

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

typicality effects

A

we are slower to categorize something that is not as similar (typical) to the prototype of the category

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

difference of how close something is to the prototype is called the __

A

typicality

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

problems with prototypes

A

edge cases: overlap of features might mean something should be in a diff category

we lose individuality of each concept/object

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

characteristics of categories

A

graded membership (some exemplars are better exemplars than others)

family resemblance (members of a category typically share common features)

related concepts (central tendencies, typicality effects)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

the three main levels of categorization

A

superordinate (groups basic levels)

basic (level where members share most of the attributes of the category)

subordinate (more specific than basic)

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

explain how knowledge affects categorization

A

the more we know about a topic the more likely the subordinate becomes your basic level for categorization

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Semantic Hierarchal Theory

A

semantic: general facts and info

information that is related is linked together, when one node is activated it makes it easier/faster to think of the connecting nodes

nodes and links

activation spreads

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

what model explains priming

A

semantic hierarchal theory

priming facilitates the activation of related concepts

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

categories and nodes have ___

A

properties (sing, fly, skin)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

generalizations of semantic hierarchy

A

DRM paradigm: theme word not present although you recall it

lexical decision task (process cow faster with milk than with wall)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

issues with semantic hierarchy

A

cannot explain typicality effects (canary is a bird and ostrich is a bird are same distance away, but have different reaction times)

cognitive economy (closeness of nodes = faster) is not always true
-faster to say a horse is an animal rather than a mammal

20
Q

family resemblance theory was a jumping point from what theory to the next

A

definitional to prototype

21
Q

naming effect

A

people more likely to list things of high-protoypicality before low-prototypicality items

22
Q

how does exemplar theory explain typicality effect

A

as exemplars are based on what we have experienced in the past, reaction times are faster for better category items that we see more of, but our knowledge still holds space for those of lower typicality

23
Q

learning a cateogory often involves a shift from ___ to ____

A

prototype to exemplar thinking

24
Q

explain the gain and loss of information between organizational leevels

A

common features
- global 3
-basic: 9
-specific; 10.3

big loss of info going global but only slight gain going more specific

25
cognitive economy
properties of a category shared by many members of a category are stored higher/more general level nodes in a network saves storage space! fly goes with birds rather than each individual bird
26
predictions of the semantic network
the time to retrieve info about a concept (property) corresponds to distance between the concept and property (the links) spreading activation
27
connectionist approach includes
distributed processing connectionism
28
what is connectionism
an approach using computer model for representing cognitive processes
29
parallel distributed processing (PDP)
a connectionist approach where concepts are represented by activity that is distributed across a network knowledge is represented in the distributed activity of many units weight determined at each connection says how strongly the next unit will be activated
30
what makes up a connectionist network
uses units -input (first layer, activated by environmental stimuli) -hidden units (middle layers of the process) -output units (create the output) based on neural networks
31
How much it is weighted (each piece of info)
- Heavy: included in processing - Light: decrease in activation connections between units, corresponds to what happens at the synapse
32
key property of connectionist networks, how does this make it different from semantic models
a specific category is represented by activity that is distrivuted over many units in the netowrk contrast semantic where specific info is at a particular node, also contrasts as it is not hierarchal
33
activation of units in a connectionist network depend on
the signal that originates in the input units the connect weights throughout the netowrk
34
connection weights determine ___
which units are activated in a pattern of activity distributed for a stimulus
35
examples of relation statements
`is is a can has
36
what do relation statements do
activate representation units as well as hidden units
37
a network must be ___ to achieve results
trained
38
how do we train a network
through a learning process connection weights are adjusted by error signals to ensure input and relation units activate the appropriate output units
39
what is an error signal? what does it do?
diff between produced output signal generated and the output that actually represents the stimulus through this process (back propagation) the error signals that are sent back to the hidden/representation units to provide info about how the weights should be adjusted to affect the correct property units
40
feedback is provided through
back propagation
41
how NN are like the brain
graceful degradation (disrupting one part does not halt performance, causes gradual disruption) generalization (neural networks can generalize from particulars - robin and canary - properties of one can provide info about the other) distributed
42
4 proposals about how concepts are represented in the brain
sensory functional hypothesis multiple factor approach semantic category approach the embodied approach
43
What does the Sensory-Functional Hypothesis propose about how we represent concepts in the brain?
It suggests we use separate systems to distinguish visual sensory attributes (for living things) and functional properties (for artifacts). Based on category-specific impairments in patients, but it doesn't explain all cases.
44
What is the key idea behind the Multiple-Factor Approach?
Concept representation is distributed and determined by multiple properties (e.g., color, motion, action). Overlap exists between categories. Difficulty distinguishing animals may be due to crowding—shared features (legs and eyes), not category deficits.
45
What does the Semantic Category Approach say about concept representation?
Specific neural circuits evolved for important survival categories (e.g., faces, places). These innate circuits support both concrete (food, people) and abstract (thoughts, values) understanding. Brain responses are distributed but show category sensitivity.
46
What is the Embodied Approach to concept representation?
Concepts are understood by reactivating sensory and motor experiences tied to interactions with the object (e.g., hammer = shape + action). Involves mirror neurons and semantic somatotopy—brain areas for action/perception light up when thinking about related concepts.
47
Hub and Spoke Model
semantic dementia (global loss of knowledge) anterior temporal lobe damage damage to one of the specialized areas of the brain (spoke) causes specific deficits damage to ATL (hub) causes general deficits