Semantic Representations, Lecture 8

Question 1

Do non-humans animals have language, and what types?

Answer 1

Yes, researched by Washoe, Koko and Kanzi.
Washoe learned around 200 manual signs (sign language)
Koko learned around 1,000 signs, and was able to combine in complex ways (sign language)
Kanzi learned around 200 symbol “lexigrams”

Question 2

What did Searle take away from the Chinese Room Problem?

Answer 2

That syntax doesn’t suffice for semantics - you have to understand meanings of thought contents.

Question 3

What is semantic memory?

Answer 3

Conceptual (general) knowledge of the world - meanings of words and objects, as well as factual knowledge

Question 4

What are the neural bases of semantic memory?

Answer 4

Temporal lobes

Question 5

What role does semantic memory have in human cognition?

Answer 5

Lies at the interface of language, memory and perception

Question 6

What is the N400?

Answer 6

Component of time-locked EEG signals known as ERPs. It is a negative-going deflection that peaks around 400 milliseconds post stimulus onset.

Question 7

How did Kutas & Hillyard use the N400 to measure semantics?

Answer 7

Ppts read sentences word-by-word. Two conditions: Semantically appropriate ending; Semantically inappropriate ending. ERPs were time-locked to the appearance of the last word.

Question 8

What were the results/conclusions from the Kutas & Hillyard, 1980?

Answer 8

Ppts anticipated the last word of the sentence. Incorrect endings - mismatch between expectation and the actually presented word. Mismatch elicited a larger N400 component.

Question 9

How did Parise and Csibra demonstrate that we can use the N400 to investigate the development of semantic representations?

Answer 9

Investigated 9 month old infants - detect the mismatch between an object appearing from behind an occlude and a preceding label. The label primes the object that should appear. Infants understood the meaning of some words.

Question 10

What kind of N400 do semantically related words elicit?

Answer 10

A small N400 in comparison to semantically unrelated words

Question 11

How was N400 modulated?

Answer 11

By the semantic connection between the expected and presented word.

Question 12

What is a concept and what does it represent?

Answer 12

It unites things on the basis of similarity of characteristics. It represents classes of things, events or ideas.

Question 13

When you activate a concept, what happens?

Answer 13

It will activate associated concepts

Question 14

How are concepts organised?

Answer 14

In a network.

Question 15

What are features?

Answer 15

Properties of or facts about a concept. Features are also linked together via a network

Question 16

What do all models propose in terms of concepts and features?

Answer 16

That concepts are comprised of a constellation of constituent features

Question 17

What are amodal representations?

Answer 17

Features are represented as abstract knowledge. They are not tied to sensory or motor information, and are independent of input or output modality.

Question 18

Why was semantic memory traditionally assumed to be amodal?

Answer 18

It can be accessed from multiple kinds of sensory input

Question 19

What is the Hierarchical Model?

Answer 19

Concepts are organised in a hierarchy. Features are associated with the concept at the appropriate levels.

Question 20

What type of model is the hierarchical model and why?

Answer 20

Amodal, as concepts are represented as abstract knowledge.

Question 21

Who proposed the Hierarchical Model?

Answer 21

Collins & Quinlan (1969)

Question 22

What is the evidence from neuroscience for Hierarchical Model?

Answer 22

fMRI study of naming and categorisation (Rogers et al., 2006). Superordinate and subordinate information have different neural substrates.

Question 23

What part of the brain does subordinate-level processing activate?

Answer 23

Anterior temporal pole

Question 24

What part of the brain does superordinate-level processing activate?

Answer 24

Posterior temporal lobes

Question 25

What is the symbol grounding problem?

Answer 25

Each word needs to be defined by other words. When using a lexicon, it is impossible to understand and learn new words/concepts without understanding some words/concepts in advance.

Question 26

What models were proposed as a solution to the symbol grounding problem?

Answer 26

Grounded models of semantic memory

Question 27

What are grounded concepts/features?

Answer 27

Concepts not defined in terms of each other, but in terms of our experiences and interactions with the world. Concepts of “green” and “kick” are linked to sensory and motor experiences rather than abstract/amodal representations

Question 28

How are concept representations distributed?

Answer 28

Over several distinct brain areas

Question 29

What is the Fully-Grounded Model of Allport?

Answer 29

Different features of a concept is represented in different information channels (modalities). These are the same channels the features were acquired through.

Question 30

According to the fully-grounded model of Allport, what is connected?

Answer 30

Representations of features belonging to the same concept.

Question 31

What is the evidence for grounded cognition?

Answer 31

Priming. In this view, retrieving information from the semantic memory is similar to the process of mental imagery.

Question 32

What is another piece of evidence for grounded cognition?

Answer 32

Action words. Processing words active body-based neural representations

Question 33

What part of the brain does naming pictures of animals and tools elicits category-related activity?

Answer 33

Ventral occipito-temporal cortex.

Question 34

How are object representations spread?

Answer 34

They are widespread and overlapping (grounded cognition)

Question 35

What do category-related activations reflect?

Answer 35

The retrieval of information about category-specific features and attributes.

Question 36

Where do category-related patterns for images, animals and tools activate?

Answer 36

Occipital and posterior temporal lobes.