|
 |  Learning: Principles and Applications, 4/e Stephen B Klein,
Mississippi State University
Complex Learning Tasks
Chapter Outline
Chapter Outline
CONCEPT LEARNING
A concept is a symbol that represents a class of objects
or events with common characteristics. Concepts allow for
the grouping of objects or events that then permits a
common response to each example of the concept. Thus,
concept learning greatly facilitates effective interactions
with our environment. Concept learning is similar to
generalization and discrimination in that we must first
learn to discriminate between stimuli and then generalize
that discrimination to other stimuli.
The Structure of a Concept
Attributes and Rules
Concepts have two properties: attributes
and rules. An attribute is any feature of
an object or event that varies from one
instance to another. A rule defines which
objects or events are examples of a
particular concept.
Types of Rules
An affirmative rule specifies that a
particular attribute defines a concept. A
negative rule indicates that any object or
event having a certain attribute is not an
example of the concept. The conjunctive
rule states that the simultaneous presence
of two or more attributes defines the
concept. A disjunctive rule allows an
object or event to be a member of a concept
if either one or more than one attribute is
present.
The Prototype of a Concept
The prototype of a concept is the object
that has the greatest number of attributes
in common with other members of the concept
and, therefore, is the most typical member
of a conceptual category. In other words,
the prototype has a strong degree of family
resemblance with other examples of the
concept.
Boundaries of the Concept
Certain properties of objects are used to
distinguish the difference between objects
that belong to two different concepts.
These defining properties are based on a
set of rules that determine the boundary of
the concept. Sometimes, it is difficult to
establish the boundary conditions for
concepts. When the boundary of a concept is
unclear, it is often difficult to determine
whether an object or event is a member of
that concept.
Studying Concept Learning in Humans
Studying Concept Learning in Animals
Concept learning involves the identification of
properties that define a concept. Research by
Herrnstein and colleagues (1976) indicates that
pigeons can learn concrete concepts. Additional
research has shown that pigeons can learn two and
three dimensional concepts as well as distinguish
between paintings by Picasso and Monet. In addition
to pigeons, dolphins have also been shown to
evidence concept learning using three dimensional
objects. D'Amato and associates (1984, 1985) found
that primates learn the abstract concepts of "same"
and "different" in studies using a procedure called
matching to sample.
Theories of Concept Learning
Two main theories have been proposed to explain
concept learning. One views concept learning as an
associative process; the other posits that concept
learning is a cognitive process.
Associative Theory
Hull developed an associative account of
concept learning in which he proposed that
learning a concept is a form of
discrimination learning. This approach
assumes that subjects associate the correct
feature with a concept name as a result of
reinforcement. Incorrect or irrelevant
features do not become associated due to
lack of reinforcement. The Smoke (1933)
study and Hull's (1920) classic study using
Chinese characters both support this view.
Prototype Theory
A more modern associated view is
exemplified by prototype theory.
According to this view, a person
compares a new stimulus with a
prototype stored in memory. If
there is sufficient commonality,
the person recognizes that stimulus
as a member of a particular
concept.
Exemplar Theory
An alternative associative view,
known as exemplar theory suggests
that a person associates specific
stimuli or exemplars with a
particular concept. Rather than
comparing a new stimulus with a
prototype, a person compares the
new stimulus with exemplars of the
concept. Evidence supports both
prototype and exemplar theories
since people can assign a new
stimulus to a concept based on
either exemplars or a prototype.
Cognitive Process in Concept Learning
According to the influential view of
Bruner and associates (1956), a concept is
learned by testing hypotheses about the
correct solution. Research by Levine (1966)
indicates that humans prominently use
hypotheses in concept learning tasks.
Levine also found that subjects generally
do not use a specific hypothesis again once
it has been shown incorrect and that the
subjects were able to test more than one
hypothesis at a time.
PROBLEM SOLVING
The Missionaries-and-Cannibals problem illustrates the
difficulty in problem solving when the solution seems to
lead away from the goal. Many real-life problems do not
have a direct path to the solution and are therefore
difficult to solve.
The Nature of the Problem
A problem exists when a person is motivated to
reach a goal and finds that the path to the goal is
blocked by some obstacle. To solve the problem and
reach the goal a person must overcome the obstacle.
To remove the obstacle a person might engage in a
trial-and-error process as proposed by Thorndike or
mentally explore the problem before attempting a
solution as Kohler would believe. Thorndike felt
that problem solving was a trial-and-error process
while Kohler felt that problem solving involved
insight, a preliminary mental exploration of the
problem until a solution is discovered.
Defining the Problem
Defining a problem entails identifying both the
starting point or initial state and the endpoint or
goal state of the problem. In addition, operations
that solve the problem must be identified as well
as those restrictions that limit problem solving.
Well-Defined Versus Ill-Defined Problems
A well-defined problem is one in which the
initial state and goal state can be clearly
delineated. An ill-defined problem exists
when there is no clear starting and/or
ending point.
Solving Ill-Defined Problems
By clearly defining the initial state and
goal state, ill-defined problems can be
analyzed. Creating a set of manageable
sub-problems provides the structure for
converting an ill-defined problem into a
well-defined problem.
A Strategy for Solving Problems
Two major strategies, algorithms and heuristics
are used to solve problems.
Algorithms
An algorithm is a precise set of rules
that specify how to solve problems of a
particular type. An algorithm may be time
consuming but it guarantees a solution if
applied properly.
Heuristics
Heuristics represent a "best guess"
problem-solving strategy. Some heuristics
involve a systematic evaluation of the
problem while others represent cognitive
short-cuts for problem solving. Heuristics
can provide a solution in less time than
algorithms but do not guarantee a solution
and errors can occur.
Working Backward Heuristic
The working backward heuristic is
a strategy that finds a solution by
literally using the end point to
suggest connections to the starting
point. This strategy is commonly
used in mathematical proofs and
other systems of formal logic.
Means-End Analysis
The means-end analysis is a
procedure that breaks down a
problem into a series of
sub-problems. Solution of the
sub-problems eventually leads to a
solution to the entire problem.
Representativeness Strategy
Representativeness is a heuristic
that makes judgments only on the
obvious characteristics of the
problem. The problem is solved
based on the initial facts
presented and will not require
additional information.
Availability Strategy
The availability heuristic uses
information that can be easily
remembered or observed. As a
result, solutions are based on the
most recent information that can be
brought to mind.
Execution of the Strategy
Well-defined problems more easily lead to a
selection of a strategy than ill-defined ones.
Failure to make a precise identification of the
initial and goal states, the operations that can
solve the problem, and restrictions on the solution
can result in complications in executing a
problem-solving strategy.
The Problem Solved
The final stage of problem solving requires an
evaluation of the accuracy of the solution. When we
have chosen the appropriate solution we can
overcome the obstacles and reach our goals. If the
solution is not appropriate, this feedback will let
us know that we need to find another way to solve
the problem.
The Consequences of Past Experience
Functional Fixedness
Future problem-solving activities may be
hampered by the inability to see a novel
use for a familiar object or functional
fixedness. Functional fixedness occurs when
we have a tendency to view an object only
according to its most common usage and fail
to see that objects can have more than one
function. Functional fixedness is an
example of negative transfer.
Set
Negative transfer also occurs when we find
ourselves habitually using the same
strategy under all circumstances. The
tendency to use the same problem-solving
strategy in future situations is referred
to as set. Set leads to negative transfer
when the solution to a problem requires a
novel approach. Set does not always lead to
a negative result. If the habitual approach
effectively solves the problem, positive
transfer occurs and problem solving
proceeds smoothly.
The Nature of Expertise
An expert is an individual who has much experience
and knowledge about a particular area. An expert
possesses more abstract concepts and uses more
general concepts than a novice. Experts are also
able to perceive the similarity between new and old
problems as well as perceive the entire problem
better than the novice. However, experts are more
likely to suffer from functional fixedness relative
to a novice.
LANGUAGE
The Nature of Language
Language serves three functions: communication,
thinking, and memory processes. Psycholinguistics
is a sub-discipline of psychology that studies
language from its lowest to highest levels.
Psycholinguistics is interested in phonemes (speech
sounds), morphemes (combinations of phonemes),
grammar (rules for combining morphemes), and
semantics (the meaning of language).
The Structure of Language
Phonemes
A phoneme is the simplest functional
speech sound in a language. English is
comprised of 45 basic phonemes while some
languages have as few as 15 and as many as
85. Phonemes also differ from one language
to the next.
Morphemes
A morpheme is the smallest meaningful unit
of language. Morphemes are the simplest
combinations of phonemes that can be formed
and still have meaning. Not all phonemes
are words (free morphemes), some are bound
morphemes such as -ow which when added to
the word pill forms the word pillow. By
itself -ow has no meaning but when bound to
another morpheme it takes on meaning.
Sentences
Words can be combined into a phrase which
is a group of two or more words that
expresses a thought. A sentence consists of
two or more phrases and conveys a complete
message.
Syntax: The Rules of Language
Phrases cannot be randomly combined to form a
sentence. Syntax is the system of rules for
combining various units of speech. With syntax, an
infinite number of language patterns can be
produced making language a generative process. Each
language has its own unique syntax.
Phonology
Languages do not use every possible
combination of phonemes. The rules that
dictate how phonemes are combined into
morphemes defines phonology.
Grammar
Just as phonemes follow rules to become
morphemes, morphemes follow the rules of
grammar in combining to create meaningful
phrases, clauses, and sentences. Linguists
have used phrase-structure grammar to
analyze the constituents of a sentence.
Semantics: The Meaning of Language
Grammar is not the same as semantics which
is an analysis of the meaning of a
sentence. Chomsky (1965) distinguished
between the surface structure and the deep
structure of a sentence. The surface
structure refers to the physical
arrangement of words in a sentence;
however, the deep structure conveys the
actual meaning of the sentence. The initial
step in comprehension of a sentence is to
divide it into clauses which are thoughts
or propositions. Once a sentence is divided
into clauses its meaning can be determined.
The Acquisition of Language
There are two views of how humans acquire
language: a learning view proposed by Skinner based
on operant conditioning, and a psycholinguistic
position initially described by Chomsky who argued
that language learning is innate.
A Reinforcement View
Skinner analyzed language acquisition
through the principles of reinforcement and
operant conditioning. Therefore, children
are "shaped" through the use of
reinforcement to learn how to use words.
Successful efforts are reinforced while
unsuccessful efforts are not.
Psycholinguists have been extremely
critical of Skinner's approach based on
three major difficulties inherent to a
reinforcement position. First, teachers of
language such as parents do not
consistently apply reinforcement. Second,
language is generative or creative in that
children produce novel yet grammatically
correct sentences without reinforcement.
And finally, language acquisition is
relatively invariant regardless of the
pattern of reinforcement.
A Psycholinguistic View
Chomsky was impressed with the ease that
children throughout the world have in
learning the complexities of language. He
described a pattern of language acquisition
that is independent of culture. On the
basis of these observations, Chomsky
proposed that children are born with an
innate capability for language learning.
Chomsky proposed that children have an
innate language acquisition device (LAD)
that "knows" the universal aspects of
language and allows them to learn the
syntax of language with minimal effort.
While only minimal exposure to language is
necessary, there appears to be an early,
sensitive period during which exposure to
language is maximally effective.
Application: Teach Chimpanzees Language
The Gardners (1971) were able to teach the chimp,
Washoe, American Sign Language (ASL). Rumbaugh and
his colleagues successfully taught chimpanzees
language using a symbolic language and a computer.
Two of Rumbaugh's chimpanzees were able to
communicate with each other by means of symbols. On
the other hand, Terrace taught his chimpanzee, Nim
Chimpsky ASL but failed to note any spontaneous use
of language on the part of Nim. Others have found
that their chimpanzees learned the meanings of many
symbols but were unable to create meaningful
sentences. However, the species of chimpanzee may
be an important variable with respect to language
acquisition; the bonobo or pygmy chimpanzee, which
has a highly developed social structure similar to
humans, is accomplished in acquiring the use of
symbolic language.
|
|
|
2002 McGraw-Hill Higher Education