INTRODUCTION
Concepts are important language used in thinking. In fact, most of the nouns in our vocabulary are names of concepts; the only exceptions are proper nouns names of specific things or persons.-In the previous unit we learned about knowledge representation, models of semantic memory. In fact they all include the process of concept formation. In this article we will learn about various processes in concept formation, including hypothesis testing, analytic and non analytic strategies et cetera.
CONCEPTS : CATEGORIZING THE WORLD
If someone asked you what was in your kitchen cabinet, you might answer with a detailed list of every item. More likely, though, you would respond by using some broader categories, such as "food' and "dishes'.
The use of such categories reflect the operation of concepts. Concepts are categorization of objects, events, or people that share common properties. By employing concepts, we are able to organize the complex phenomena into simpler, and therefore more easily usable, cognitive categories. Examples are "man", "red", "triangle", "motivation", "atom", "anger", and the word concept itself.
Concept allow us to classify newly encountered objects on the basis of our past experience. For example, we are able to tell that a small rectangular box with buttons I sitting an a chair near a television is probably a remote control-even if we have never encountered that specific brand before.
Concepts would enable us to think about and understand more readily the complex word in which we live.
PROCESS IN CONCEPT FORMATION
Clark Hull (1920), one of the most influential psychologists of this century initiated an experimental analysis of concept formation. Through his experiments Hull viewed that concept formation involved the discrimination of common. features.
But later research has shown that people form concepts by actively testing hypothesis
and using sophisticated, logical strategies. In this section we examine the process of testing hypothesis and using strategies.
HYPOTHESIS TESTING
As psychologists have shown human beings are not merely passive recipients of information about the meanings of concepts. We all engage in the active process of 1 hypothesis testing. In concept formation, hypothesis testing is a process in which we attempt to search out the meanings or salient features of concepts by testing our assumptions.
People actively test-hypothesis about defining features (discussed in previous article) of a concept. All possible features constitute the pool of available hypothesis. The learner samples one or more of the features and then tries to classify examples of the concept based on the current hypothesis.
Even young children may use hypothesis testing to speed the learning of concepts such as shapes and colors. When you show a child a square and say "This is a square", the child may try to understand what accounts for the squareness of the figure. Is it the figure's boxiness, its size, its color or the fact that it is closed? After being shown several figures that are identified as squares and several that are not, the child appears to develop a concept of squareness that involves more basic concepts such as box, straight line, and (lines) just as long as the others. Years may pass before the child's definition becomes sophisticated enough to involve concepts such as a closed geometric figure that has four sides equal in length and right angles.
Levine (1966, 1975), through his experiments_designed blank trials procedure in order to externalize subject's hypothesis. The results of these studies have shown that adults engage in hypothesis testing on over 95 percent of the trials. Additionally, adults tend to adopt a win-stay, lose-shift strategy-they stick with a particular hypothesis only as long as its works i.e. if feedback provided to the learner is positive, then the current hypothesis is retained. If feedback is negative, then the hypothesis is rejected and a new hypothesis is sampled. When a hypothesis was confirmed on a feedback trial, that hypothesis was retained throughout the following block of no-feedback trials 95 percent of the time. But when a hypothesis was not confirmed on a feedback trial that hypothesis was not used in the subsequent no-feedback trials.
Incremental or all-or-none Learning?
As learning occurs gradually, with steady improvement overtime, perhaps, concept learning too, is a gradual, incremental process.
An alternate possibility, however, is that people learn concepts in a discontinuous, all of none manner. Indeed, early models of hypothesis testing had emphasized that concept formation is an all-or-none process. In this view, individual subjects do not show gradual improvement while learning a concept. Rather, the learning process is discrete; subjects perform at a chance level on one trial and more immediately to a high, stable level of performance on the next trial.
Testing Multiple Hypothesis
Using the observations from several experiments, Levine ( 1966) estimated that individuals can test multiple hypothesis at a time. Thus each of the feedback trials can eliminate half of the viable hypothesis. It was also proved that subjects tend to remember particular hypothesis that has. been tested previously and individuals-eliminate more hypothesis on trials in which correct responses occurred than on trials in which incorrect responses occurred.
The above observations demonstrate that human cognition is flexible and our flexibility is highlighted by the diversity of strategies we use in learning concepts.
ANALYTIC AND NON-ANALYTIC STRATEGIES
People learn concepts by actively planning a method or strategy that guides their performance. In this section we discuss some of the different kinds of strategies that people use in learning concepts.
There are two general types of strategies: Analytical and non analytic. Analytic strategies involve the abstraction of attributes, rules, or prototypes. Rules are logical statements that specify how attributes are combined. For instance affirmative rule specifies a particular attribute that by itself defines the category. In the experiments on prototypes, the subjects seemed to have abstracted a prototype (Robin for example) or best representative of the category (for instance animal).
In contrast, non-analytic strategies involve remembering particular positive instances, which arc then used to decide whether stimuli belong to the category. The chief characteristic of non-analytic strategies is that they involve the us~ of particular instances, not attribute~ or global properties abstracted from these instances. The usefulness of analytic and non-analytic strategies depends upon the type of task, and subjects often use a blend of analytic and non-analytic strategies.
Types Of Analytic Strategies
(i) Conservative Focusing : One analytic strategy inferred by Bruner et al. is called 'conservative focusing'. This strategy involves identifying a positive instance, formulating a hypothesis that includes all of the attributes contained in that instance, and systematically testing the relevance of each attribute.
(ii) Focus Gambling : A related strategy called focus gambling, which involves changing two or more attributes at a time. The focus gambling strategy can lead to either very rapid or very slow solutions, but which occurrence depends largely on chance
(iii) Scanning : A third type of analytic strategy is called 'Scanning'. Under this the subjects test hypothesis that do not include all of the potentially relevant attributes. In the successive scanning the subject tests one hypothesis at a time. This method is relatively inefficient since there are many individual hypotheses to test and the subjects who use the strategy do not use all of the information about which attributes are relevant and irrelevant.
(iv) Simultaneous scanning: A related strategy is simultaneous scanning in which I the subject tests numerous attributes at once. The difference between this one and focus gambling is that, as in other scanning strategies, the subject's hypothesis does not include all of the relevant attributes. The main problem with this strategy is that it places heavy demands on memory for the subject must remember which attributes have been tested and which were correct or incorrect.
In the Bruner et al. study, many adults used the conservative focusing strategy and often solved the problem faster than subjects who used other strategies. The chief advantage of conservative focusing is that it reduces the demands on memory. Conservative focusing is not, however, an ideal strategy for fanning concepts, especially for complex concepts.
Non analytic Strategies
Non-analytic strategies do not involve; the abstraction of attributes or prototypes from the positive instances of the concept. Subjects who use a non-analytic strategy remember specific positive instances; then they compare these instance to new stimuli in deciding whether the new stimuli belong to the category. Experimental evidence concerning the use of non-analytic strategies has come from studies by Brooks (1978).
In many situations, adults, use a mixture of analytic and non-analytic strategies. Robert and Allen ( 1978), for example observed that subjects who have participated in a paired-associate procedure do show some knowledge of the rules used to produce the sequences. These same subjects also used knowledge concerning particular stimuli. So adults use both types of strategy, sometimes in combination. Which strategy, if either, predominates depends upon numerous situational factors, which we consider next.
Factors Favoring The Use Of Analytic Strategies
(i) Memorial factors can lead people to use analytic strategies. By abstracting features of the positive instances, the burden on memory is reduced.
(ii) The use of analytic strategies is also facilitated by stimulus and procedural factors. When the stimuli have highly salient or noticeable attributes in common then people tend to abstract the salient attribute. People tend to use analytic strategies when the rules that define a category are salient.
(iii) Prior experience may also lead us to use analytic strategies. Through formal education, we learn to give definitions for many categories. In learning to define categories such as noun, dinosaur, and ocean, we were taught to state the attributes that define the category.
(iv) We often use analytic strategies in every day affairs, because abstracting common properties helps us to notice and to make use of the regularities in our environment. For example by abstracting, we learn that different types of birds have common properties, that birds and mammals have common properties and so on.
Factors Favoring the use of non analytic strategies
(i) Non-analytic strategies are favored when the relevant attributes are not obvious or when we have seen only one or two instance~ of a complex category. These conditions are likely to occur when one is just beginning to learn a new category, as when a child who has seen only one dolphin tries to classify novel animals as dolphins or non-dolphins. It then makes sense to classify novel stimuli by comparing them to remembered positive instances.
(ii) Non -analytic strategies are also likely to be used in classifying atypical category members.
(iii) People may tend to use a non-analytic strategy when they remember particular instances very well and when a speedy classification must be made. Through incidental learning, people remember particular stimuli their cars, their friends, their pets and so on very well. These remembered items can be used to classify novel stimuli. These comments indicate that different situations favor the use of analytical and non-analytic strategies. Yet we should also recognize that the two types of strategy are complementary and can both be used in a particular situation. In classifying a novel animal, we might begin by evaluating the creature's attributes. But if we were uncertain that our classification was correct we might take a non-analytic route.
It is through our flexibility in using strategies that we satisfy the demands of the diverse categorization tasks that occur in everyday life.
SUMMARY
Conceptual categories are formed through diverse processes. People often use analytic strategies in which they abstract attributes or prototypes from the members of the category. One could learn the category dog,by abstracting out attributes such as has hair, has four legs, barks, and has sharp teeth. Adults often isolate relevant attributes through a process of hypothesis testing in which they vary one or a group of attributes in order to determine which attributes are essential. People also form concepts through the use of non-analytic strategies which do not involve the abstraction of attributes. The user of a non-analytic strategy memorizes instances of a category and compare those instances to novel stimuli in order to decide whether the novel stimuli belong to the category. Adults probably use analytic and non-analytic strategies in different situations, and the two types of strategy can be viewed as complementary.