Learning

Question 1

What is conceptual knowledge?

Answer 1

The knowledge that enables us to recognize objects and events. It also allows us to make inferences about their properties.

Question 2

What is a concept?

Answer 2

A mental representation of a class or individual.

Question 3

What is categorization?

Answer 3

The process by which things are placed in categorical groups.
Categorical groups enable us to understand individual cases not previously encountered. Without them, we’d be constantly mystified/helpless.
They are pointers to knowledge –> give us general information about the item; once it’s categorized, we can look at individual differences.

Question 4

What are the theories answering how we categorize?

Answer 4

• Definitional approach
• Prototype approach
• The exemplar approach

Question 5

Explain the definitional approach.

Answer 5

Says that we determine the category membership based on whether the object meets the category definition. It doesn’t work as not all category members meet the definition. Can’t explain why some category members are more typical than others.

Question 6

What is family resemblance?

Answer 6

The idea inspired the prototype approach, which states that category members resemble one another in various (but not all) ways.

Question 7

Explain the prototype approach.

Answer 7

States that there are typical members of the category that are the average representation of all category members. Prototypical objects are preferentially processed (typicallity effect).

Prototype approach –> no clear boundaries.

Question 8

What did Rosch (1975) show in her study about typicallity?

Answer 8

She asked participants to rate things based on how typical they were. She found that highly typical members strongly resemble each other.

Question 9

What did Smith et al (1974) show in their experiments on prototypical objects?

Answer 9

They verified that prototypical objects are verified faster. E.g., apple is verified faster than pomegranate.
Other experiments also showed that prototypical objects are named first when asked to list things in a category. E.g., sparrows named before penguins.

Question 10

What is a priming effect? and what is the relationship between prototypical objects and priming effects?

Answer 10

Priming effect - when an initial stimulus affects how fast we process the second stimulus.

Prototypical objects show stronger priming effects in faster RT.

Question 11

Explain the exemplar approach.

Answer 11

States that a concept is represented by multiple actual examples, not a single prototype. To categorize a new item, it’s compared to stored examples. It predicts many prototype theory effects.

Question 12

What is an advantage of the prototype approach and what is an advantage of the exemplar approach?

Answer 12

Prototype - may be better for larger categories.
Exemplar - better at handling atypical cases and unusual categories. Also explains typicallity.

We might use both theories.

Question 13

How are categories organized?

Answer 13

They are organized hierarchically from more specific to more general. There are at least 3 categories.
- Subordinate/ Specific
- Basic
- Superordinate/ Global

Question 14

Which hierarchical category do we use the most?

Answer 14

Basic level because we gain a lot of information going from global to basic but not so much going from basic to specific.

There are a lot of individual differences. People with specialized interests transform their subordinate category into their basic category.

Question 15

What are semantic networks?

Answer 15

A way of arranging concepts by meaning. Collins and Quillan propose hierarchical semantic networks.

Question 16

Describe Collins and Quillan’s hierarchical semantic networks.

Answer 16

Nodes - represent categories or concepts

Semantically related nodes are linked by lines/ arrows –> mostly vertical arrows linking with bigger categories, but also some branches (specific attributes of the node).

Question 17

What is cognitive economy?

Answer 17

An important concept of Collins and Quilan’s model.
Shared properties are at higher levels and are inherited by lower levels. There is additional information in smaller categories. A way of making the system more efficient.

Question 18

What is Collins and Quillan’s prediction?

Answer 18

Verifying properties should take longer the more “nodes” you have to go through. It’s faster when it’s specific characteristics.

E.g., canary sing RT < canary fly RT

Question 19

What is spreading activation? and what idea does it inspire?

Answer 19

Thinking about a concept increases the activation of that concept, plus the activation of related concepts. A lot of this happens automatically.

The idea of priming derives from it.

Question 20

Explain the lexical decision task.

Answer 20

Derived from the spreading activation idea.
1) stimulus can be a word or a non-word
2) You need to decide if the target is a word or not
3) Before you are presented with the word you are shown a prime (word related to target)
How related the prime and the target are predicts the RT.

Question 21

What are some criticisms of Collins and Quillan’s semantic networks?

Answer 21

• Can’t explain some typicallity effects.
  E.g., canary is identified as a bird faster than an ostrich, despite the same distance from the bird node.
• Little evidence for cognitive economy/inheritance.
• Some sentence-verification results are problematic.
  E.g., A pig is identified faster as an animal than as a mammal, even though it’s closer to mammal.

Question 22

Explain the connectionist approach.

Answer 22

Parallel distributed processing (doing many steps at the same time).
It’s a computer model that represents cognitive processes. Concepts are represented in the distributed activity of many linked units. –> not in a single node.

It seems analogous to real brain/neurons. It exhibits graceful degradation –> gradual disruption of performance when part of the system is damaged.

Question 23

What are the 3 categories of linked “units”?

Answer 23

Input units: activated by stimulation from the environment
Hidden units: receive input from input units
Output units: receive input from hidden units

Question 24

What are connection weights?

Answer 24

They determine how strongly signals from 1 unit increase or decrease the activity of the next unit.

Question 25

How do connection networks learn?

Answer 25

Knowledge is gained over many training trials.
They don’t have knowledge programmed in. They begin with random response parameters, and then they are trained to produce the correct response over many trials.

The goal is to reduce the error signal (difference between output response and correct response) to zero. This is done through back propagation (transmitting error signal back to the circuit to train it).

Question 26

How are categories represented in the brain? What are the four main theories?

Answer 26

• Sensory-functional hypothesis
• Semantic category approach
• Multiple factors approach
• The embodied approach
• Hub and spoke model

Correct view probably involves many of them

Question 27

Explain the sensory-functional hypothesis.

Answer 27

Concepts are organized based on their sensory or functional properties.
Some brain-damaged individuals have trouble categorizing animals but not artifacts.
Animals –> sensory info
Artifacts –> functional basis
This is inconsistent with the evidence. Found that people with impaired artifact ID don’t have damage in their functional knowledge, but do have damage in their sensory knowledge.

Question 28

Explain semantic category approach.

Answer 28

Specific neural circuits in the brain are used for specific categories. Not functional vs. sensory basis. There is no emphasis on specific areas but on circuits in the brain.

Question 29

Explain the multiple factors (property cluster) approach.

Answer 29

Proposes are represented neurally as conjunctions of properties.

Question 30

Embodied approach

Answer 30

Concepts are represented by activation of sensory and motor circuits that are active when we interact with the object.
Support:
-Mirror neurons support perception/action connection
- Semantic somatotopy: hearing body part words activates similar areas of the brain as moving those body parts.
Problems:
- Impairment in functional movements doesn’t predict trouble recognizing objects.

Question 31

Hub and spoke model

Answer 31

Anterior temporal lobe (ATL) is a hub that integrates information from specialized category areas (spokes) in the brain.
Support:
- ATL damage leads to semantic dementia: inability to identify any object, regardless of category, but other damage leads to specific category impairment.
- Transcranial magnetic stimulation (TMS) of ATL impairs the naming of artifacts and living things.
- TMS of parietal area only affects the naming of artifacts.

Question 32

What is TMS?

Answer 32

Transcranial magnetic stimulation –> uses magnetic pulses to stimulate or impair certain brain areas. Can be used causationally to establish double dissociations.