Lecture 3 - Semantic Memory

Question 1

semantic memory

Answer 1

as the store of knowledge about the world - contains concepts

Question 2

concept

Answer 2

mental representations and fundamental units of thought/knowledge e.g. concept of an animal

Question 3

how are concepts organised in the memory system

Answer 3

first idea of Hierarchical network model

Question 4

Hierarchical network model

Answer 4

• Collins & Quallian 1969
• semantic memory organised into a series of hierarchical networks
• major concepts as nodes
• properties/features are associated with each concept
• cognitive economy

Question 5

what is cognitive economy in the hierarchical network

Answer 5

properties are stored higher up, and not repeated lower down, to minimize redundancy

Question 6

support for Hierarchical model

Answer 6

• Collins & Quallian
• sentence verification task (say whether sentence is true of false)
• allow us to quantify reaction time
• properties (e.g. sing) stored as the same level as the target concept (e.g. canary) are accessed more quickly (short reaction time)
• those stored higher up need more time to be processed (e.g. canary can fly has longer verification time)

Question 7

reaction time in sentence verification tasks

Answer 7

• unless information is directly linked to a concept in semantic memory, we infer the answer from properties of high nodes
• verification time increases for broader concepts e.g. a bird flying
• more inferences slows (increases) verification time

Question 8

problems with hierarchical model

Answer 8

• problem of familiarity in sentence verification task:
  how often do you hear the sentence ‘a canary has skin’? when controlled reduced hierarchical distance effect
• how typical is the verification concept: verification is faster for more representative member categories, independent of hierarchical distance e.g. penguin vs canary as a bird (canary more typical of category)

Question 9

model that developed from Hierarchical model

Answer 9

spreading activation model

Question 10

spreading activation model

Answer 10

Collins & Loftus 1975
- semantic memory is organized by semantic relatedness/distance
- lengths of links indicates the degree of semantic relatedness
- activity at one node causes activation of other nodes via links
- spreading activation decreases as it gets further away from the original point of activation

e.g. penguin is a bird has slow activation whereas canary is a bird is strong activation
- due to previous experiences

Question 11

supporting evidence for spreading activation model - study

Answer 11

semantic priming tasks
McNamara 1992
- when presenting one stimulus that is more semantically related makes subsequent processing more efficient/faster (short-link)
e.g. red + rose vs red + flower

• semantic links and distance determine the strength and speed of activation spread from one concept to the other

Question 12

supporting evidence for spreading activation model - paradigm

Answer 12

• Deese-Roediger-McDermoot (DRM) Paradigm
• p’s selected words they did not study as they were related to the subject
• suggest activation spreads from all words presented/studied to words related

Question 13

spreading activation model evaluation

Answer 13

• more flexible than hierarchical model (links of different strength and distance)
• can account for more empirical findings
• but reduces specificity of model’s predictions
• and more difficult to test

Question 14

limitations of spreading activation model

Answer 14

• the notion that each concept is represented by a single node is oversimplified what about abstract concepts e.g. justice
• each concept has a fixed mental representation whereas the context in which we encounter them changes the way we process them, so do different people have different representations?

Question 15

the role of context in semantic memory - theory

Answer 15

situated simulation theory

Question 16

situated simulation theory

Answer 16

• concepts are processed in different settings
• their processing is influenced by the current context/setting
• concepts have perceptual/motor/action related properties

e.g. activated aspects of ‘bicycle’ concept reflect current goals … e.g. for a child would have training wheels as a part of the perceptual properties that are activated

Barsalou 2009

Question 17

evidence for situated stimulation theory - does access to concepts involve motor systems?

Answer 17

Hauk et al 2004
- Brain areas activated by action words overlap with activation produced by corresponding movement e.g. leg words (AW) and foot movements

• action words activate motor areas related to the concept - shows action related elements are presented with a concept- therefore facilitate accessing the meaning of the concept

Question 18

evidence for situated stimulation theory - does involvement of motor system facilitate access to concepts

Answer 18

Miler et al. 2018
- presented words either relating to hands or feet (e.g. sprint) and asked p’s to respond with hands or feet
- when consistency between meaning and response then reaction time faster (sprint + feet)
- when inconsistent reaction time slower (sprint with hands)

• understanding of action verbs requires activation of the motor areas used to carry out the names action

Question 19

situated simulation theory evaluation

Answer 19

processing of concepts depends on the situation and perceptual + motor processes in a given task

Question 20

limitations of SST model

Answer 20

• are the concepts variable across situations e.g. concept has a core but there are context dependent elements
• are these action related -
  properties secondary i.e. after the concept meaning has been accessed, then action-related are activated

Question 21

Concepts in the brain - semantic memory approaches

Answer 21

• grandmother cell hypothesis
• feature-based approach

Question 22

grandmother cell hypothesis

Answer 22

• semantic memories are represented in the brain as whole objects
• each concept with its own node or neuron
  e.g. special neuron for your grandmother
• types of nodes are grouped together e.g. all living things
• not a lot of evidence to support this

Question 23

feature-based approach

Answer 23

• different kinds of information about a given object are stored in separate brain regions
  e.g. visual information is stored in one part of the brain, while auditory linked to the same object is stored in another (but networks between areas)
• more likely

Question 24

neurocognitive model

Answer 24

Hub-and-spoke model
(Patterson et al 2007)

Question 25

Hub-and-spoke model

Answer 25

a hybrid model of semantic memory
- Hub: modality independent conceptual representations (meaning related information)
- Spokes: modality-specific brain areas. Sensory and motor processing

Question 26

where is the hub + use

Answer 26

• anterior temporal lobe
• holds general meaning of concepts
• amodal semantics

Question 27

support for Hub-and-Spoke model
tDCS study

Answer 27

transcranial direct current stimulation
- asked two questions
1) tool function questions e.g. ‘are scissors for cutting’
2) tool manipulated questions ‘are pliers gripped by the handles’

• tDCS applied to IPL (inferior parietal lobe) (relating to actions/use of objects) enhanced tool manipulated task
• tDCS applied to ATL (anterior temporal lobe)(for information about concepts)(hub) enhanced performance on both conditions
• consistent with the idea that hub contains amodal information of concepts

Question 28

neuropsychological evidence for Hub-and-spoke model

Answer 28

• semantic dementia problems accessing meaning related information about object e.g. naming objects, sorting objects into categories(damage in ATL - represents hub in model)
• strokes (category-specific deficits) have greater difficulty identifying living vs nonliving objects in a specific category. have compromised cortical systems (spokes) in areas relating to living things (cannot name/ recognize pictures in these categories)(e.g. damage to motor cortex = more difficulty naming/categorizing animate objects than inanimate objects)
• support to evidence of core elements in ATL (hub) and more category specific elements in cortical areas(spokes)

Question 29

evaluation of Hub-and-spoke model

Answer 29

• evidence that concepts are organised in hub (core) + spokes (modality-specific concepts/meanings)

Question 30

limitations of Hub and Spoke model

Answer 30

• role of ATL may be more complex
• does familiarity affect organization in the hub?
• how many spokes?
• how is information integrated between spokes and the hub (communication can be examined via different methodologies e.g. neuroimaging)

Question 31

summary of semantic memory:

Answer 31

• information within semantic memory is organized in various ways
• hierarchical model suggests concepts are organized in hierarchical way with nodes and features along hierarchy
• spreading activation model stresses semantic relatedness and distance between concepts
• according to Barsalou concepts incorporate perceptual and motor features and their processing is affected by the context in which they are encountered
• Hub-and-spoke model provides neurobiological way that concepts are organized incorporating modality-general at the hub and modality specific spoke features

Question 32

Collins and Quillian (1969) devised which model of semantic memory?

Answer 32

c.
Hierarchical network model

Question 33

The Deese-Roediger-McDermott paradigm is typically thought of as providing support for which of the following models?

Answer 33

SPREADING ACTIVATION

Question 34

Collins and Loftus (1975) devised which theory of semantic memory?

Answer 34

SPREADING ACTIVATION

Question 35

There is growing evidence that semantic information is organized in which format within the brain?

Answer 35

DISTRIBUTED