Uploaded by seahfaith

Categories, Knowledge, and Semantic Organisation

advertisement
For Module Use Only: Do Not Share or Post Online
PL3103: Cognitive Psychology
Categories,
Knowledge, and
Semantic Organisation
Lecture 7 ▪ Week 8 ▪ 7 March 2022
For Module Use Only: Do Not Share or Post Online
How do people organise knowledge about the
world around them?
2
For Module Use Only: Do Not Share or Post Online
Today’s Agenda
1. Categories
and
Concepts
 Classical view
of categories
 Natural
categories
PL3103 Cognitive Psychology
2. Categorisation
Models





Prototype
Exemplar
Goal-derived
Social, other
Strategies
3. Semantic
Memory,
Tasks, Models
 Semantic
memory
 Tasks
 Models
 Video/demo:
Spreading
activation
4. Schema
Theory and
Scripts
 Schemas
 Scripts
 Video clip:
Schemas and
stereotypes
3
For Module Use Only: Do Not Share or Post Online
Categories, Knowledge, and Semantic Organisation
Categories and
Concepts
4
For Module Use Only: Do Not Share or Post Online
Categories and Concepts
 Categories: objects or events that are grouped together
because of relatedness.
• Enables us to interact with our environment without being
overwhelmed by its complexity.
 Concepts: ideas that represent a class of objects, events, or their
properties.
• Alternatively: mental grouping that is used to remember and
understand what things are, what they mean, and what categories
or groups they belong to.
PL3103 Cognitive Psychology
5
For Module Use Only: Do Not Share or Post Online
Categories and Concepts
 Knowing the category provides a lot of information.
PL3103 Cognitive Psychology
6
For Module Use Only: Do Not Share or Post Online
Categories and Concepts
 Categories are helpful and necessary for multiple reasons
(Bruner et al., 1956):
1.
2.
Reduce complexity of the environment.
Ability to identify objects in the world.
3. Reduces the need for constant learning.
4. Allows us to decide what constitutes an appropriate action.
5.
Enables us to order related classes of objects and events.
PL3103 Cognitive Psychology
7
For Module Use Only: Do Not Share or Post Online
Imagine having to interpret this scene without
any knowledge of categories…
8
For Module Use Only: Do Not Share or Post Online
Classical View of Categories
 Classical view of categorisation posits that all categories have
defining features (i.e., attributes), and that these features are:
• Singly necessary (a member of the category must have each
feature)
• Jointly sufficient (something with these features must be a
member of the category)
 Logical rules and defining features used to define categories.
PL3103 Cognitive Psychology
9
For Module Use Only: Do Not Share or Post Online
Category membership can be controversial…
10
For Module Use Only: Do Not Share or Post Online
Classical View of Categories
 Although the classical view of categorisation captures the
intuition that concepts have “essences,” it has been criticised.
 Disagreement over defining features.
• Example: what are the features of a “game”?
 Some categories, particularly those found in everyday
circumstances, are based on continuous dimensions rather than
defining features.
PL3103 Cognitive Psychology
11
For Module Use Only: Do Not Share or Post Online
Classical View of Categories
 The classical view of categories does not readily explain
typicality effects (Rosch, 1973).
• Example: some birds rated as more “birdy;” as in robin vs. penguin.
• If the classical view was true, there should be no gradation of
membership within the same category (since they all have the
necessary features)
• Typicality affects how people respond to objects; e.g., reaction
times.
PL3103 Cognitive Psychology
12
For Module Use Only: Do Not Share or Post Online
Natural Categories
Rosch et al. (1976)



Real world categories have
hierarchical organisation.
Some category members are more
representative (typical) than others.
Categories are organised into three
nested levels from largest to
smallest.
PL3103 Cognitive Psychology
13
For Module Use Only: Do Not Share or Post Online
Natural Categories
Superordinate
Rosch et al. (1976)
level



Real
Superordinate
world categories
level categories
have
are
hierarchical
the largest. organisation.
Some
Difficulty
category
is thatmembers
membersare
share
more
few
representative
attributes.
(typical) than others.
 Examples:
furniture,
musical
Categories
are
organised
into three
instruments,
vehicles
nested
levels from
largest to
smallest.
PL3103 Cognitive Psychology
14
For Module Use Only: Do Not Share or Post Online
Natural Categories
Superordinate
Basic levellevel




Superordinate
level categories
Basic
level categories
are most are
the largest.
important.
Difficulty
that members share few
First
to beislearned.
attributes.
Most
often used to name objects.
 Examples: furniture, musical
Provides most new information
instruments, vehicles
compared to other levels.


Basic level objects are most
differentiated from one another.
Examples: chair, table, lamp.
PL3103 Cognitive Psychology
15
For Module Use Only: Do Not Share or Post Online
Natural Categories
Subordinate
Basic level
level





Basic
Subordinate
level categories
level categories
are most
are
important.
most well known by experts.
 Experts
can turn “subordinate”
First
to be learned.
categories into basic levels, e.g., a
Most often used to name objects.
dog expert shown a picture may say
Provides
most
newjust
information
“cocker
spaniel”
as quickly as
compared
to other
levels.
other people
would
say “dog”
 Basic level
objects
are most may
Difficulty
in that
members
differentiated from one another.
share attributes with members of
 Examples: chair, table, lamp.
other subordinate categories.
PL3103 Cognitive Psychology
16
For Module Use Only: Do Not Share or Post Online
Categories, Knowledge, and Semantic Organisation
Categorisation
Models
17
For Module Use Only: Do Not Share or Post Online
Prototype Models
Posner & Keele (1968)
•
Categories are organised around a
prototype or “best example” which is an
average of the most typical features.
•
Members have a family resemblance.
•
There is no single feature in common,
but several members share common
features.
•
Similarity between a specific item and
the prototype determines how likely it is
a member of the category.
PL3103 Cognitive Psychology
18
For Module Use Only: Do Not Share or Post Online
Prototype Models
 Strengths:
• Typicality and unclear cases are predicted
• Items with the most attributes in common with the prototype
are judged to be more typical.
• Fits nicely with the notion of a Basic level
• Prototypes should be at the level where category members
have most features in common.
• A prototype of a category should be at the level where it is
most distinct from the prototypes of other categories.
PL3103 Cognitive Psychology
19
For Module Use Only: Do Not Share or Post Online
Prototype Models
 Criticisms:
• Does not preserve enough information about individual members
• People are sensitive to frequency, variability, correlations of
features among members.
• Does not explain context effects:
• Example: is a harmonica a typical musical instrument around a
campfire? In an orchestra?
PL3103 Cognitive Psychology
20
For Module Use Only: Do Not Share or Post Online
Exemplar Models
Medin & Schaffer (1978)
•
Concepts are represented by the
individual instances or exemplars.
•
While each exemplar may not be
retrievable, an item is compared to most
of them and overall similarity is found.
PL3103 Cognitive Psychology
21
For Module Use Only: Do Not Share or Post Online
Prototype Models
 Strengths:
• Does not lose information.
• Sensitive to context.
 Criticisms:
• Does not explain how categories are formed in the first place.
PL3103 Cognitive Psychology
22
For Module Use Only: Do Not Share or Post Online
Goal-Derived Categories
 Both prototype and exemplar models of categorisation are
similarity-based; items classified based on how much they
resemble a prototype or each other.
 Goal derived categories exemplify ill-defined categories
wherein individual items are very dissimilar, with no family
resemblance, but share the same “essence” for a particular
situation or goal.
• Example: things to bring on a picnic.
PL3103 Cognitive Psychology
23
For Module Use Only: Do Not Share or Post Online
Social and Other Categories
 People can be categorised into groups based on religious or
social causes, among other attributes.
• Clinical diagnoses.
 Stereotypes are a form of social categorisation in which withingroup similarity is exaggerated.
• May result in misleading impressions and assumptions.
PL3103 Cognitive Psychology
24
For Module Use Only: Do Not Share or Post Online
Categorisation Strategies
 How does one identify whether a presented item is a member of
a given category?
•
Nearest neighbor rule: select category containing an item that is most
similar to the new item.
•
Average distance rule: compare new item with all members in a category
in order to determine the average similarity between the new item and all
members of the category.
•
Feature frequency rule: select the category with the most feature
matches with the classified item.
•
Prototype rule: select the category with the prototype that is most similar
to the new item (appears to be most popular strategy).
PL3103 Cognitive Psychology
25
For Module Use Only: Do Not Share or Post Online
Categories, Knowledge, and Semantic Organisation
Semantic Memory
and Semantic
Memory Models
26
For Module Use Only: Do Not Share or Post Online
Semantic Memory
 Type of LTM involving the capacity to recall words, concepts, or
numbers.
• Organised knowledge of the world.
• Fairly constant knowledge structure (stable).
• Includes knowledge of words, and also things that cannot be
readily expressed in words.
PL3103 Cognitive Psychology
27
For Module Use Only: Do Not Share or Post Online
Semantic Memory
 Examples:
• I know that the meaning of the word “semantic” is closer to the
meaning of the word “vocabulary” than it is to the word
“disarmament.”
• I remember that the chemical formula for water is H20.
• I know that velvet is soft.
• I know that a penguin is a bird, but a robin is a better example of
the category “bird.”
PL3103 Cognitive Psychology
28
For Module Use Only: Do Not Share or Post Online
Semantic Memory Tasks
Hierarchical organization
 Information that is presented in an
organised way is better remembered
(e.g., Bower, 1970).
Sentence verification
 Given a sentence, indicate whether
it is true or false.
 Reaction time is measured.
True or False:
“A poodle is a dog.”
PL3103 Cognitive Psychology
29
For Module Use Only: Do Not Share or Post Online
Semantic Memory Tasks
Category size effect
 Sentences containing pairs drawn
from smaller categories are verified
faster than sentences containing
pairs drawn from larger categories.
Category
Sentence
size verification
effect violated
 In
Given
some
a sentence,
cases, sentences
indicatewith
whether
pairs
drawn
it is true
from
or false.
larger categories are
than
sentences with
 verified
Reactionfaster
time is
measured.
pairs drawn from smaller categories.
True or False:
“A robin is a bird.” (faster)
“An “A
apepoodle
is an animal.”
(faster)
is a dog.”
vs.
vs.
“A robin is an animal.” (slower)
“An ape is a primate.” (slower)
PL3103 Cognitive Psychology
30
For Module Use Only: Do Not Share or Post Online
Semantic Memory Tasks
Relatedness effects
 Sentences containing pairs that are
highly unrelated (i.e., anomalous)
semantically are verified faster.
“A car is a bird.” (faster)
vs.
“A bat is a bird.” (slower)
PL3103 Cognitive Psychology
Typicality
effects
Category
size effect
violated
 Sentences
referring
to more
typical
In some cases,
sentences
with
pairs
members
of larger
a category
are verified
drawn
from
categories
are
faster than
sentences
referringwith
to
verified
faster
than sentences
less typical
of acategories.
category.
pairs
drawnmembers
from smaller
“Aape
robin
bird.” (faster)
“An
is is
anaanimal.”
(faster)
vs.
“A chicken
a bird.” (slower)
“An
ape is aisprimate.”
(slower)
31
For Module Use Only: Do Not Share or Post Online
Semantic Memory Models
 Hierarchical network (Collins &
Qullian, 1969)
•
Category information is stored
directly in memory by means of
associations.
•
Features are stored at highest
level at which they are true for all
concepts/instances below that
level.
•
Verification time as a function of
hierarchy of levels.
PL3103 Cognitive Psychology
32
For Module Use Only: Do Not Share or Post Online
Semantic Memory Models
 Feature comparison (Smith,
Shoben, & Rips, 1974)
•
Concepts are stored as lists of
features (no property inheritance).
•
Classifications made by:
1. Compare all features of two
concepts to determine
similarity
2. If needed, examine defining
features to determine
category membership.
PL3103 Cognitive Psychology
33
For Module Use Only: Do Not Share or Post Online
Semantic Memory Models
 Spreading activation (Collins &
Lofthus, 1975)
•
Concepts are interconnected in a
network.
•
Length of links between concepts
is a function of relatedness.
•
Activation spreads over network,
but loses strength with distance.
•
Could explain false recognition of
words in the Deese-RoedigerMcDermott (1995) paradigm.
PL3103 Cognitive Psychology
34
For Module Use Only: Do Not Share or Post Online
Semantic Networks
National Geographic (2014; 3:17)
PL3103 Cognitive Psychology
35
For Module Use Only: Do Not Share or Post Online
Semantic Networks
 The available evidence does seem to suggest that people
organise knowledge around hierarchies and association
networks.
• Example: given “bread,” one is faster at recognising “butter” as a
word than “pigeon” or some semantically unrelated word.
 However, the nature of such networks, and cognitive
mechanisms involved, remains a matter of debate.
• Yet other theories have been proposed (e.g., ACT model of
Anderson, 1976).
PL3103 Cognitive Psychology
36
For Module Use Only: Do Not Share or Post Online
Categories, Knowledge, and Semantic Organisation
Schemas and
Scripts
37
For Module Use Only: Do Not Share or Post Online
Schema and Scripts
 Schema: an active organisation of past experiences in which the
mind abstracts a general cognitive structure to represent many
particular instances of those experiences (Bartlett, 1932).
• Framework for organising clusters of knowledge
• Allows us to make sense of new information – match to existing
schemas
• Allows us to make intelligent guesses in the absence of specific
information
PL3103 Cognitive Psychology
38
For Module Use Only: Do Not Share or Post Online
Schema and Scripts
 Scripts are schemas of routine events (Schank & Abelson, 1977)
 Examples:
• Knowledge of important activities for a type of event.
• Knowledge of sequence of activities for a type of event.
 Violations of schemas and scripts can trigger surprise, confusion,
curiosity, and other emotional responses.
PL3103 Cognitive Psychology
39
For Module Use Only: Do Not Share or Post Online
Schema and Scripts
UCLA (2016; 2:31)
PL3103 Cognitive Psychology
40
For Module Use Only: Do Not Share or Post Online
Categories and Semantic Organisation
 Concepts and categories allow us to create equivalence classes
for different objects.
 The classical view of categorisation is based on logical rules and
defining features.
 The classical view ignores the characteristics of natural
categories, which are based on continuous dimensions, rather
than defining features.
PL3103 Cognitive Psychology
41
For Module Use Only: Do Not Share or Post Online
Categories and Semantic Organisation
 Prototype and exemplar models are instances of similarity-based
approaches to categorisation.
• Prototype models focus on matching an instance with the
prototype or “best example” of that category.
• Exemplar models focus on matching an instance with the
prototype or “best example” of that category.
 Other types of categorisation rely on a more functional
underlying principle or goal.
 Different categorisation strategies have been proposed.
PL3103 Cognitive Psychology
42
For Module Use Only: Do Not Share or Post Online
Categories and Semantic Organisation
 Semantic memory refers to memory for meanings; general
knowledge.
 The hierarchical organisation and sentence verification
techniques have been used to explore semantic memory.
 The available evidence does seem to suggest that people
organise knowledge around hierarchies and association
networks.
PL3103 Cognitive Psychology
43
For Module Use Only: Do Not Share or Post Online
Categories and Semantic Organisation
 Models of semantic memory have been proposed based on
hierarchical structure, feature lists, and spreading activation
networks.
 Schema theory provides a framework for organising knowledge
clusters at a more macro level.
 Scripts store knowledge about the structure of routine events.
PL3103 Cognitive Psychology
44
For Module Use Only: Do Not Share or Post Online
Categories and Semantic Organisation
 What do you think is a good schema to adopt when thinking
about cognitive psychology?
PL3103 Cognitive Psychology
45
For Module Use Only: Do Not Share or Post Online
PL3103 Week 8
Lecture Readings
 Textbook chapters 10-11 (for next week)
On LumiNUS
 Quiz 4 (Chs 8-9) released on Friday, due Monday
Tutorial Readings
 Squire & Knowlton, 1995 (for next week)
PL3103 Cognitive Psychology
46
For Module Use Only: Do Not Share or Post Online
Next Week…
Special Guest Lecture
Language and Text
Comprehension
PL3103 Cognitive Psychology
47
Download