What goes on in the minds of students when they are comparing and con-trasting sets of exemplars? What kinds of hypotheses occur to them in the early stages and how do they modify and test them? To answer these questions, three factors are important to us as we design a lesson or unit. First, we con-struct concept attainment exercises so that we can study how our students think. Second, the students should be able to not only describe how they attain concepts, but also learn to be more efficient by altering their strategies and learning to use new ones. Third, by changing the way we present information and by modifying the model slightly, we affect how students will process information.
The key to understanding the strategies students use to attain concepts is to analyze how they approach the information available in the exemplars. In particular, do they concentrate on just certain aspects of the information (par-tistic strategies), or do they keep all or most of the information in mind (holistic strategies)? To illustrate, suppose we are teaching concepts for analyzing liter-ary style by comparing passages from novels and short stories. The first set of positive exemplars includes the following passage:
A new country seems to follow a pattern. First come the openers, strong and brave and rather childlike. They can take care of themselves in a wilderness, but they are naive and helpless against men, and perhaps that is why they went out in the first place. When the rough edges are worn off the new land, busi-nessmen and lawyers come in to help with the development— to solve prob-lems of ownership, usually by removing the temptations to themselves. And finally comes culture, which is entertainment, relaxation, transport out of the pain of living. And culture can be on any level, and is. (Steinbeck, 1952, p. 249) The students know that this passage will be grouped with the others to come, on the basis of one or more attributes pertaining to style. Some students will therefore concentrate on just one attribute, say, the use of declarative sen-tences or the juxtaposition of contrasting ideas about the opening of the fron-tier. Others will scan the details of the passage, noting the presence or absence of metaphors, the use of evocative language, the author’s stance of being an observer of the human scene, and so on.
When comparing this passage with another positive one, a partist (some-one who focuses on just (some-one or two aspects of the use of language) will in some sense appear to have an easier task— just looking to see whether the attribute present in the first is also present in the second, and so on. However, if the student’s focus does not work out, he or she must return to the earlier exam-ples and scan them for something else on which to concentrate. A holist, on the other hand, has to keep many attributes in mind and eliminate nondefining elements one at a time. But the holistic strategy places the learner in a good position to identify multiple attribute concepts, and the loss of a single attri-bute is not as disruptive to the overall strategy.
There are two ways that we can obtain information about the way our stu-dents attain concepts. After a concept has been attained, we can ask them to recount their thinking as the exercise proceeded— by describing the ideas they came up with at each step, what attributes they were concentrating on, and what modifications they had to make. (“Tell us what you thought at the begin-ning, why you thought so, and what changes you had to make.”) This can lead to a discussion in which the students can discover one another’s strategies and how they worked out.
Older students can write down their hypotheses, giving us (and them) a record we can analyze later. For example, in a study of the classification of
ChaPter Six concept attainment 139
plants conducted by Baveja, Showers, and Joyce (1985), students worked in pairs to formulate hypotheses as pairs of exemplars (one positive and one neg-ative) were presented to them. They recorded their hypotheses, the changes they made, and the reasons they made them. The holistic students painstak-ingly generated multiple hypotheses and gradually eliminated the untenable ones. The students who selected one or two hypotheses in the early stages needed to review the exemplars constantly and revise their ideas in order to arrive at the multiple- attribute concept that was the goal. By sharing their strategies and reflecting on them, the students were able to try new ones in subsequent lessons and to observe the effect of the changes.
If we provide students with a large number of labeled exemplars (ones identified as positive and negative) to commence a lesson, they are able to scan the field of data and select a few hypotheses on which to operate. If we provide the exemplars pair by pair, however, the students are drawn toward holistic, multiple- attribute strategies.
Many people, on first encountering the concept attainment model, ask about the function of the negative exemplars. They wonder why we should not simply provide the positive ones. Negative exemplars are important because they help the students identify the boundaries of the concept. For example, consider the concept impressionism in painting. Impressionistic styles have much in common with other painting styles. It is important for students to
“see” examples that have no traces of impressionism for them to be absolutely certain about the defining attributes. Likewise, to identify a group of words as a prepositional phrase, we need to be able to tell it from a clause. Only by com-paring exemplars that contain and do not contain certain attributes can we identify the characteristics of the attributes precisely, and over time. The con-cept attainment model is designed to produce long- term learning. Having struggled our way to precise definitions of prime number, element, developing nation, irony, and so on, we should recognize members of their categories posi-tively and surely when we encounter them in the future.
Tennyson and Cocchiarella (1986) have conducted important research into concept learning and developed a number of models that can be used to improve instructional design. In the course of their explorations, they have dealt with a number of questions that can help us understand the concept attainment model. They have compared treatments where students induce attributes and definitions, finding that the students developed clearer concepts and retained them longer when the examination of the exemplars preceded the discussion of attributes and definitions. Tennyson and Cocchiarella also discovered that the first positive exemplars presented should be the clearest possible prototypes, especially with multiple- attribute concepts. In other words, the teacher should not try to “fake out” the students with vague exemplars, but should take care to facilitate concept learning by arranging the data sets so that less- clear exem-plars are dealt with in the phases where the principles are applied.
Tennyson and Cocchiarella (1986) also concluded that students develop procedural knowledge (how to attain concepts) with practice, and that the more procedural knowledge the students possess, the more effectively they attain and can apply conceptual knowledge. Thus, the analysis of thinking to facilitate concept attainment appears to be very important.
The idea of learning concepts and then clarifying attributes and defini-tions runs counter to much current teaching practice. We have learned that some teachers, when first using concept attainment, have an urge to provide definitions and lists of attributes. It is important to remember that the appropriate time for clarification is after the students have abstracted the concepts.
Data are presented to the students in the form of sets of items called exemplars— for instance, a set of poems. These are labeled “positive” if they have characteristics or attributes of the concept to be taught (for example, the sonnet form). The exemplars are labeled “negative” if they do not contain the attributes of the concept (for example, poems that do not have all the attri-butes of a sonnet).
By comparing the positive and negative exemplars, the students develop hypotheses about the nature of the category. They do not, however, share their hypotheses at this point. When most of the students have developed a hypoth-esis, some unlabeled exemplars are presented to them and they indicate whether they can successfully identify positive exemplars. They may be asked to produce some of their own (as by scanning a set of poems and picking out some positive and negative ones).
Then they are asked to share their hypotheses and describe the progression of their ideas during the process. When they have agreed on the hypotheses that appear most likely, they generate labels for them. Then the teacher sup-plies the technical label, if there is one (sonnet, for example).
To consolidate and apply the concept, the students then search for more items of the class (poems, in this case) and find which ones most closely match the concept they have learned.
| T he Model oF TeAChIng
The phases of the concept attainment model are outlined in Table 6.1.
Syntax
Phase one involves presenting data to the learner. Each unit of data is a separate example or nonexample of the concept. The units are presented in pairs. The data may be events, people, objects, stories, pictures, or any other discriminable units. The learners are informed that all the positive examples
ChaPter Six concept attainment 141
have one idea in common; their task is to develop a hypothesis about the nature of the concept. The instances are presented in a prearranged order and are labeled yes or no. Learners are asked to compare and justify the attributes of the different examples. (The teacher or students may want to maintain a record of the attributes.) Finally, learners are asked to name their concepts and state the rules or definitions of the concepts according to their essential attri-butes. (Their hypotheses are not confirmed until the next phase; students may not know the names of some concepts, but the names can be provided when the concepts are confirmed.)
In phase two, the students test their attainment of the concept, first by cor-rectly identifying additional unlabeled examples of the concept and then by generating their own examples. After this, the teacher (and students) confirm or disconfirm their original hypotheses, revising their choice of concepts or attributes as necessary.
In phase three, students begin to analyze the strategies by which they attain concepts. As we have indicated, some learners initially try broad con-structs and gradually narrow the field; others begin with more discrete constructs. The learners can describe their patterns— whether they focused on attributes or concepts, whether they did so one at a time or several at once, and what happened when their hypotheses were not confirmed. Did they change strategies? Gradually, they can compare the effectiveness of different strategies.
Phase One: Presentation of Data and Identification of Concept 1. Teacher presents labeled examples.
2. Students compare attributes in positive and negative examples.
3. Students generate and test hypotheses.
4. Students state a definition according to the essential attributes.
Phase Two: Testing Attainment of the Concept
1. Students identify additional unlabeled examples as yes or no.
2. Teacher confirms hypotheses, names concept, and restates definitions according to essential attributes.
3. Students generate examples.
Phase Three: Analysis of Thinking Strategies 1. Students describe thoughts.
2. Students discuss role of hypotheses and attributes.
3. Students discuss type and number of hypotheses.
Syntax of the concept attainment Model Table