Memory:Prototype, Rosch and Colleagues, Experiments of Stephen Read

<< Memory:Benefits of Categorization, Levels of Categories

Memory:Schema Theory, A European Solution, Generalization hierarchies >>

Cognitive Psychology PSY 504

Lesson 35

Memory

Representation of knowledge (continued)

Levels of Categories:

i) Superordinate Category: "It is a large category at the top of the hierarchy." e.g. furniture,

tools, vehicles etc.

ii) Basic Level Category: "It is an intermediate category in the middle of the hierarchy." e.g.

table, chair, bed, sofa etc.

iii) Subordinate Category: "It is a small category at the bottom of the hierarchy." e.g. lamp,

screw-driver, truck etc.

Prototype

Experiment

An experiment was conducted by Rosch & Lackoff in 1970's on prototypes. They collected a lot of

evidence for prototypes at the basic level or the subordinate level but not at superordinate level.

Shown picture of a living room chair were:

A piece of furniture (superordinate level)

A chair (basic level)

A living room chair (subordinate level)

Subjects were divided into three groups and were shown a living room chair. One group was

asked you will be shown a piece of furniture. And the second group was asked you will be shown

a chair or not and the third group was asked you will be shown a living room chair or not. They

were assigned different tasks:

Group 1: If you see any furniture press yes.

Group 2: If you see a chair press yes.

Group 3: If you see a living room chair press yes.

Results:

The results showed that subjects were fastest at the basic level category i.e. chair. Chair faster

than furniture or living room.

Subjects identify objects at basic level than make an inference regarding superordinate level

(chair is a piece of furniture), or classify them at subordinate level by looking for distinguishing

features. People look first the category than they decide it is what kind of or level of category.

Conclusion:

It seems that subjects identified objects at the basic level.

What about carpenters?

Experts in an area may be very quick to make subordinate classification.

Experts might be very quick to make subordinate classifications in their area of expertise.

Rosch and Colleagues:

Roch and Colleagues tried to prove that the experts might be very quick to make subordinate

classifications in their area of expertise.

Rosch and colleagues (1976) tested Dog experts and bird experts.

Dog Experts and Bird Experts were taken as subjects.

103

Cognitive Psychology PSY 504

Subjects were shown the color pictures of dogs and sparrows.

They gave these instructions to the Subjects;

i) You will be shown an animal, if you see an animal press the button.

ii) You will be shown either a dog or an animal, if you see a dog press the button.

iii) You will be shown a beagle and a sparrow.

Replicated earlier results when dog experts identified birds and bird experts identified dogs. But in

their area of expertise, experts were as fast at the subordinate level as at basic level.

Results:

Dog experts were fastest at the bird condition rather then in the sparrow condition and slow at the

animal level. Bird experts were fastest at the animal condition rather then in the beagle condition.

Experts were as fast at the subordinate level as at the subordinate level.

The distinction between one level and the other level is a matter of experience rather than the

imitation of mind.

Experiments of Stephen Read:

Stephen conducted an experiment by using these faces.

At the top row the faces were named category one and the second row was named the category

2. The last row was called novel faces. After showing the first row and second row subjects were

shown a novel face and were asked if this face belonged to category one or category two.

Categorization rules

They developed some rules prior to the experiment. These rules were;

1. Nearest neighbor rule

Matches an item to the test item is called nearest neighbor rule. Subjects compare the novel face

with both categories and included this face to the above given categories.

2. Average distance rule

Matches test item on the basis of average similarity is called average distance rule. On the

average this face is matched with one category or second category.

3. Prototype rule

Matches prototype with test item is called prototype rule. By averaging all faces subjects make a

prototype and then match novel face with this prototype.

104

Cognitive Psychology PSY 504

4. Feature frequency rule

Selects on Most feature matches is called feature frequency rule. The answers were included in

one category according to the frequency.

Results:

58% subjects said that they averaged the features.

Over 50% used prototype.

Over 20% use the feature frequency rule.

Subjects were shown 25 novel faces.

Conclusion:

Prototype strategy was the most frequently used strategy.

Feature frequency rule was the second best.

Average similarity and near neighbor rule were least frequently used.

There might have been other ways of looking at the same problem.

There is also evidence for exemplar and feature frequency rule.

Different rules are applicable in different situations. We all have a tendency to apply these four

rules. People selectively use these rules in different situations.

Findings

Prototype most frequent

Feature frequency second best

Some people did use exemplar

Exemplars

Exemplar is not our tendency to match an object with every single item, with every single object in

the category. We only match it with a typical item in the category.

Experiments

Nosofsky (1991) replicated his experiment on read experiment. And he found many students use

exemplars in classification. He found that many subjects were using exemplar not the prototype.

JD Smith and JP Minda (1998) conducted experiment and they came up with a new idea that is

prototype for early learning. When children see goat first time and he says bowh bowh child

thinks it is dog. So in early learning the children compare the goat with dog. Probably we use

something like prototype in our early learning. We may need more integration rather

argumentation.

105

Table of Contents: