Lecture 3 - Concepts and Categories

Question 1

Define Concept and Category.

Answer 1

A concept can be defined as a mentally possessed idea or notion, whereas a category refers to a set of entities that are grouped together.
E.g. democracy as a concept and then countries that are considered democratic - this is a concept and then categorisation respectively.

Question 2

Define categorisation.

Answer 2

Categorisation is the ability to form equivalence classes of discriminable entities.

Question 3

What are some of the challenges to the Classical View of Categorisation?

Question 4

What are the three theories of how we learn categories discussed in the lecture?

Question 5

What are the limitations of the Prototype Theory?

Answer 5

Prototype theories cannot explain how we learn non-linearly separable categories.

Question 6

What is the alternative to the Prototype Model that we discussed that can explain how we learn non-linearly separable categories?

Answer 6

Exemplar Theory.

Question 7

Why do we form categories, and why is it so important for how we manage the world we perceive?

Answer 7

Being able to categorise things is an evolutionarily beneficial ability that we have developed. It allows us to perceive the complex world in more simplified and manageable way. It also allows us to transfer knowledge about one member of a category to another - e.g. knowing that some snakes are poisonous means that we tend to be cautious around most snakes in case they too are poisonous.

Question 8

Why does the meaning of a concept change based on the user of the concept?

Answer 8

The meaning of a concept changes depending on who it is using the concept. This is because the goals someone has for using that concept change. E.g. Someone might say that stress on the body is what can lead to chronic health conditions, whereas another might use stress as a pivotal driver of change. The concept of stress is taking on different meaning based on the goal or aim of the user for using that concept.

Question 9

What happens to generalisation as similarity decreases?

Answer 9

Our willingness to generalise between entities decreases as the similarity between entities decreases. So, for example if we know that robins have a certain behavioural characteristic we would likely being ok saying that a blue wren also has this behavioural feature as these two birds are quite similar, whereas we would probably be less likely to say that flamingoes have this behavioural pattern, due to them being quite different to robins.

Question 10

What did the World Color Survey find about how we categorise colours?

Answer 10

They found that even though we are able to discriminate around 10,000 colours, 100 different languages had around 10-11 words for the main colours.

Question 11

In the study done by Eleanor Rosch in 1978 she wanted to understand the structure of categorisation. She looked at the structure of categorisation as having a hierarchy of three main levels - Superordinate, Basic, Subordinate. What did she find about the number of words people used in the three levels?

Answer 11

As you go from superordinate to basic to subordinate the number of words that people used in these categories increased - this seems pretty intuitive.

Question 12

Eleanor Rosch published a paper in 1978 with a number of experiments illustrating the amount of information we get from the superordinate level of categorisation.
True or false?

Answer 12

False. The experiments were illustrating the large amount of information and importance of the basic level of categorisation.

Question 13

When does the Classical View of Categorisation date back to?

Answer 13

Ancient Greece.

Question 14

What is the Classical View of Categorisation?

Answer 14

Members of a category belong to the category based on the fact that they meet the definition of a member of that category. All members are equal members of that category as the all meet the same definition.

Question 15

Bruner, Goodnow & Austin (1956) did a study to understand the Classical View of Categorisation. They did experiments looking at how participants learned category membership based on arbitrary rules.
They found that there were two types of strategies people used to learn the rules for membership to an arbitrary category.
What were these two types of strategies and what was the difference between the two?

Answer 15

The two types of strategies were Scanning strategies and Focusing strategies.
Scanning Strategies were when participants tried out different hypotheses for what the category membership might be each time they were presented with new members/non-members of the category, i.e. they were “scanning” members to see whether they fit the hypothesis.
Focusing Strategies on the other hand involved focusing on the features/attributes that change or do not change to determine what defined category membership - participants that did this would vary one feature at a time in each choice based on what features changed/stayed the same for the previous item and whether that was considered a member or not.

Question 16

What is the Classical View Theory of Categorisation?

Answer 16

The Classical View Theory of Categorisation posits that people use rule-based hypotheses to figure and apply categorisation membership. There is a definition that of membership and any entity that fits that definition belongs, just as much as anything else, in that category.

Question 17

What are some Challenges to the Classical View of Categorisation?

Answer 17

The Classical View of Categorisation sees all members as equally belonging to the category, but people have challenged this view as it appears that we also categorise based on if members partially fit the definition and there are more “typical” members or prototypes of the category.

Question 18

What did Rosch find in regards to how often a member of a category had one of the attributes listed by participants when they were asked to list attributes of the category members?

Answer 18

Rosch (1978) found that the most most frequent attributes listed more a category tended to only apply to one member of the category. This proposed a challenge the The Classical View Theory of Categorisation. Furthermore, she found that on average only one attribute listed out of all of them belonged to all members of the category.

Question 19

Rosch (1978) also got participants to rate members of the category as typical or not for that category. e.g. a chair is a more “typical” member of furniture, whereas fish tank (also in the category of furniture), but is “not typical”.
How does this finding challenge the Classical View Theory of Categorisation?

Answer 19

The Classical View sees members of a category as equal members and as all fitting a definition that permits the entity to that category. Rosch’s findings suggest otherwise. She finds that there are typical and atypical members and that members rarely share many of the attributes that people use to define the category.

Question 20

What is the importance or implication of the typicality of members of a category?

Answer 20

Knowing the typicality of a category is important. It has been shown that people are able to identify entities as members of a group if they are typical members of that group (very intuitive, but still interesting). Also, knowing the typicality of a category means that we know something about what is important for membership to this category.

Question 21

In regards to categorisation studies, what does induction refer to?

Answer 21

It refers to generalising information from the particular to the general. An example being, if we know all horses get tricket’s diseases and we know all cats get tricket’s disease are we willing to say all mammals get tricket’s disease?

Question 22

Osheron et al. (1990) did a number of experiments to highlight the importance of typicality of categories to how we learn and generalise across members of a category.
What were some of their main findings?

Answer 22

They found that willingness to generalise information across members of a category is affected by the typicality of the members the information pertains to and the typicality of the members the information is potentially being generalised to.

Question 23

What is the Prototype Model of Categorization?

Answer 23

The Prototype Model of Categorization assumes that categorization decisions are made based on how similar an object is to the prototype/ideal member of that category.

Question 24

What was the experiment we looked at in the lecture that wanted to explore the Prototype Theory of Categorization? What were the main findings?
Hint: the arbitrary artificially generated dot constellations were the category members used.

Answer 24

We looked at an experiment that used artificially generated dot constellations that were had a prototype that the rest of the members were generated from. The generated members were of low or high similarity to the prototype. The participants were trained on the category by showing them members of the category that were not the prototype, but the low or high similarity to the prototype. After they had been trained they were shown the prototype or other low/high similarity members of the category. Even though they had never seen the prototype before they were more likely to say that the prototype was a member of the category than the low/high similarity members even when they had seen these members during training.
The researchers concluded that this was evidence in support of the Prototype Model of Categorization. Their theory being that participants, during training, would extract an average from the members they saw and this average would resemble (at least more than other non-prototype members) the prototype hence they would be more willing to say it was a member of the category.

Question 25

What is the Prototype Enhancement Effect?

Answer 25

The Prototype Enhancement Effect is the phenomenon seen in categorization tasks where participants are more likely to assign the prototype member of a category to that category even when they have never seen the prototype, compared to other members of the category even when they have been shown/told that is a member of the category.

Question 26

What was the 5-4 Task of Categorization discussed in the lecture and what was the main finding?
Did the findings provide support for the Prototype Model of Categorization?

Answer 26

The 5-4 Task used 5 category members that varied in 4 different features. The findings were very similar to the dot constellation experiments - participants were more likely to assign the prototype/ideal member to the category that other less similar members even if they had been shown and told those members were part of the category.

Question 27

What is one of the main challenges/criticisms of the Prototype Model of Categorization?

Answer 27

The Prototype Model cannot explain how we learn non-linearly separable categories.

Question 28

What is the Prototype Enhancement Effect seen in categorization tasks? And what two models of categorization explain this effect?

Answer 28

The Prototype Enhancement Effect is the tendency of people to assign the prototype/ideal member of the category to the category even though they had never seen the prototype before and even compared to other members that they had seen before and had been told that they were members of the category.
The two theories of categorization that explain this effect are The Prototype Model and The Exemplar Model of Categorization.

Question 29

What Model of Categorization is able to explain and predict how we learn non-linearly separable categories?

Answer 29

The Exemplar Model.

Question 30

What is the basic idea behind the Exemplar Model of Categorization?

Answer 30

The Exemplar Model says that the way we categorize entities is by comparing novel entities to the examples of members of a category we have stored in our memory. This differs to the prototype model which says that we generate a prototype of a category based on averaging out the data we have of members of the category that we are aware of.

Question 31

e What does the Prototype Model predict about non-linearly separable categories as discussed in the lecture)?

Answer 31

The Prototype Model predicts that when learning non-linearly separable categories the protoypes we generate will be similar or the same for both categories. However, we are able to learn non-linearly separable categories and therefore we need a theory that can explain this - hence the Exemplar Model.

Question 32

In Eleanor Rosch’s experiment on category hierarchies based on the idea that categories consist of superordinate, basic, and subordinate levels, what did she find in regards to the number of attributes people could list for these levels.

Answer 32

Superordinate were significantly less than basic and subordinate.
Basic and subordinate seemed to have similar levels.
Rosch proposed that the most meaningful information is found at the basic level.

Question 33

What are some of the characteristics of basic level categories discussed in lecture?

Answer 33

1) Basic level categories tell us most of the information about an entity compared to the super- or subordinate features.
2) Basic level categories are identified more quickly than superordinate
3) Are retrieved more often as a category name
4) Are learned first in language acquisition.

Question 34

Do Rule-Based Theories of Categorisation arise from Classical View of Categorisation?

Answer 34

Yes.

Question 35

In Brunner et al. (1956), what were the two types of strategies of categorisation found?

Answer 35

Scanning and Focusing strategies of categorization.

Question 36

What is the PROTOTYPE ENHANCEMENT EFFECT?

Answer 36

Prototype Enhancement Effect refers to the phenomenon that occurs when people are more likely to assign prototype of a category to a category over other members of the category even if they have seen them in training and have not seen the prototype.

Question 37

What can EXEMPLAR models of categorisation explain that PROTOTYPE models of categorisation cannot explain?

Answer 37

The learning non-linearly separable categories.