A review of cognitive teaching models

A Review of Cognitive Teaching Models

[ ] Brent Wilson Peggy Cole

Brent Wilson and Peggy Cole are at the University of Colorado at Denver.

The purpose of this article is to review from an instructional-design (ID ) perspective nine teaching programs developed by cognitive psychologists over the last ten years. Among these models, Collins" cognitive apprenticeship model has the most explicit prescriptions for instructional design. In the article, the cognitive apprenticeship model is ana- lyzed, then components of the model are used as an organizing framework for understanding the remaining models. Differences in approach are noted between traditional ID prescriptions and the cognitive teaching models. Surprisingly, no design strategies were found to be common to all the model programs. Key differences among programs included: (1) problem solving versus skill orientation, (2) detailed versus broad cognitive task analysis, (3) learner versus system control, and (4) error-restricted versus error- driven instruction. The article concludes with an argument for the utility of continuing dialogue between cognitive psychologists and instructional designers.

[] The field of instructional design (ID) emerged more than 30 years ago as psychologists and educators searched for effective means of planning and producing instructional systems (Merrill, Kowallis, & Wilson, 1981; Reiser, 1987). Since that time, instructional designers have become more clearly differen- tiated from instructional psychologists working within a cognitivist tradition (Glaser, 1982; Glaser & Bassok, 1989; Resnick, 1981). ID theorists tend to place priority on developing explicit prescriptions and models for designing instruction, while instructional psychologists focus on understanding the learning processes in instructional settings.

Of course, the distinction between designers and psychologists is never clearcut. Over the years, many ID theorists have explored learning processes, just as many psychologists have put considerable energy into the design and implementation of experimental instructional programs. However, because the two fields support different literature and theory bases, communication is often lacking. Cog- nitive psychologists have tended to overlook contributions from the ID literature, and like- wise much design work of psychologists has gone unnoticed by ID theorists.

Collins, Brown, and colleagues (e.g., Collins, 1991; Collins, Brown & Newman, 1989) have developed an instructional model derived from the metaphor of the apprentice working under

ETR&D, Vol. 39, No. 4, pp. 47-64 ISSN 1042-1629 4 7

48 ErR~, Vol, 39, No. 4

the master craftsperson in traditional socie- ties, and from the way people seem to learn in everyday informal environments (Rogoff & Lave, 1984). They have called their model cognitive apprenticeships and have identified a list of features found in " ideal" learning environments. According to the Collins-Brown model, instructional strategies would include modeling, coaching, scaffolding and fading, reflection, and exploration. Additional strategies are offered for representing content, for sequencing, and for maximiz- ing benefits from social interaction.

Of course, many of Collins' recommended strategies resemble strategies found in the ID literature (e.g., Reigeluth, 1983a). Clearly, both fields could benefit from improved communication concerning research findings and lessons learned from practical tryout. With that goal in mind, the purpose of this article is to review programs and strategies developed by instructional psychologists, using the cognitive apprenticeship model as a conceptual framework. Several teaching systems employ- ing cognitive apprenticeship ideals are de- scribed. The resulting review is valuable in two ways: Cognitive psychologists should be able to make a better correspondence between their models and current ID theory, hopefully seeing areas needing improvement, and ID theorists also should be able to see correspondences and differences, which may lead to revision or expansion of our current models.

THE NEED FOR COGNITIVE APPRENTICESHIPS

The cognitive apprenticeship model rests on a somewhat romantic conception of the "ideal" apprenticeship as a method of becoming a master in a complex domain (Brown, Collins, & Duguid, 1989). In contrast to the classroom context, which tends to remove knowledge from its sphere of use, Collins and colleagues recommend establishing settings where worth- while problems can be worked with and solved. The need for a problem-solving orientation to education is apparent from the difficulty schools are having in achieving sub- stantial learning outcomes (Resnick, 1989).

Another way to think about the concept of apprenticeship is Gott's (1988a) notion of the "lost apprenticeship," a growing problem in industrial and military settings. She noted the effects of the increased complexity and auto- mation of production systems. First, the need for high levels of expertise in supervising and using automated work systems is growing, while the need for entry levels of expertise is declining. Workers on the job are expected more and more to be flexible problem solv- ers. Human intervention is often most needed at points of breakdown or malfunction, at which points the expert is called in. Experts, however narrow the domain, do more than apply "canned" job aids or troubleshooting algorithms; rather, they have internalized considerable knowledge which they can use to flex- ibly solve problems in real time (Gott, 1988b).

Gott's second observation relates to training opportunities. Now, at a time when more problem-solving expertise is needed due to the complexity of systems, fewer on-the-job training opportunities exist for entry-level workers. Often there is little or no chance for beginning workers to acclimatize themselves to the job, and workers are expected to per- form like seasoned professionals very quickly. True apprenticeship experiences are becoming relatively rare. Gott calls this dilemma of more complex job requirements with less time on the job to learn "the lost apprenticeship" and argues for the critical need for cognitive apprenticeships and simulation-type training to help workers develop greater problem- solving expertise.

A BRIEF REVIEW OF ID MODELS

It is assumed that readers have some prior knowledge of ID models and theories; however, a short overview is offered here to allow a dear contrast with selected cognitive approaches.

ID models come in two generic varieties: procedural models for systems design (e.g., Andrews & Goodson, 1980) and conceptual models that incorporate specific instructional strategies for teaching defined content (Reige- luth, 1983a, 1987). The procedural models often are represented as flowcharts reflecting

COGNITIVE TEACHING MODELS 49

a series of project phases, progressing from needs and problem analyses to product implementation and maintenance. Procedural ID models depend less on learning theory and more on systems theory and project manage- ment methodologies (Branson & Grow, 1987). Of greater interest for our purposes are the instructional-strategy models. All such models are based on Robert Gagn6's conditions- of-learning paradigm (Gagn6, 1966), which in its time was a significant departure from the Skinnerian operant conditioning paradigm that was dominant among American psychologists. The conditions-of-learning paradigm posits that a graded hierarchy of learning outcomes exists, and for each desired outcome a set of conditions that lead to learning exists. Instruc- tional design is a matter of darifying intended learning outcomes and then matching up appropriate instructional strategies. The designer writes behaviorally specific learning objectives, classifies those objectives according to a taxonomy of learning types, then arranges the instructional conditions to fit the current instructional prescriptions. In this way, designers can design instruction to success- fully teach a rule, a psychomotor skill, an attitude, or piece of verbal information.

A related idea within the conditions-of- learning paradigm claims that sequencing of instruction should be based on a hierarchical progression from simple to complex learning outcomes. Gagn6 developed a technique of constructing learning hierarchies for analyz- ing skills: A skill is rationally decomposed into parts and sub-parts, then instruction is ordered from simple subskills to the complete skill. Elaboration theory uses content structure (concept, procedure, or principle) as the basis for organizing and sequencing instruction (Reige- luth, Merrill, Wilson, & Spiller, 1980). Both methods depend on task analysis to break down the goals of instruction, then on a method of sequencing that proceeds from simple to gradually more complex and complete tasks.

Some of the teaching models being offered by cognitive researchers bear strong resemblance to traditional tD models. For example, Larkin and Chabay (1989) offer design guidelines for the teaching of science in the schools:

1. Develop a detailed description of the processes the learner needs to acquire.

2. Systematically address all knowledge induded in the description of the process.

3. Let most instruction occur through active work on tasks.

4. Give feedback on specific tasks as soon as possible after an error is made.

5. Once is not enough. Let students encounter each knowledge unit several times.

6. Limit demands on students' attention. (pp. 160-163)

By any standard, these design guidelines are very close to the prescriptions found in component display theory, elaboration theory, and Gagn6's instructional-design theory. The strong correspondence can be seen as good news for ID theories: Many current cognitive researchers seem to agree on some fundamen- tals of design that also form the backbone of ID models.

On the other hand, other cognitive teaching models emphasize design elements that traditional ID models historically have under- emphasized, such as learner-initiated inquiry and exploration, social "scaffolding," cooperative-learning methods, and empathic feedback (Lepper & Chabay, 1988). Some cognitive teaching models explicitly differentiate themselves from traditional instructional development. Constructivist theorists, for example, have offered alternatives to the conditions-of- learning paradigm. Based on a neo-Piagetian framework, Case (Case, 1978; Case & Bereiter, 1984) has advocated an approach that focuses more explicitly on individual learners' misconceptions and knowledge structures. Bereiter (1991) develops a connectionist argument that suggests that human performance cannot be explained by explicit rules, but rather by a pattern-matching process. Hence, the "family of instructional theories in which rules, deft- nitions, logical operations, explicit procedures, and the like are treated as central (Reigeluth, 1983). . . has produced an abundance of technology on an illusory psychological foundation" (p. 15).

Other theorists, influenced by Vygotsky's concept of a zone of proximal development (Wertsch, 1985), think of tasks differently. New-

50 ETR&D, Vol. 39, No. 4

man, Griffin, and Cole (1989) characterize traditional design methods:

First, the tasks are ordered from simple or easy to complex or difficult. Second, early tasks make use of skills that are components of later tasks. Third, the learner typically masters each task before moving onto the next. This conception has little to say about teacher-child interaction since its premise is that tasks can be sufficiently broken down into component parts that any single step in the sequence can be achieved with a minimum of instruction. Teacherless computerized classrooms running "skill and drill" programs are coherent with this conception of change. (p. 153)

The authors contrast this task-analysis approach with a more teacher-based approach in which simplicity is achieved by the social negotiation between teacher and learner:

The teacher and child start out doing the task together. At first, the teacher is doing most of the task and the child is playing some minor role. Grad- ually, the child is able to do more and more until finally he can do the task on his own. The teacher's actions in these supportive interactions have often been called "scaffolding," . . . suggesting a temporary support that is removed when no longer necessary. . . . There is a sequence involved. . , but it is a sequence of different divisions of labor. The task in the sense of the whole task as negotiated between the teacher and child--remains the same. (p. 153)

Several of the cognitive teaching models reviewed below embody similar constructivist concepts. It is not yet clear, however, precisely how such concepts would translate into guidelines for designing teaching materials. The cognitive apprenticeship model discussed at length below inc ludes some wel l -worn design elements, such as model ing and fading, and others that are relatively neglected in ID models, such as situated learning, exploration, and the role of tacit knowledge.

FEATURES OF COGNITIVE APPRENTICESHIPS

In this section, the design elements of the cognitive apprent iceship model (based primarily on Collins, 1991) are reviewed and related to traditional ID concepts.

1. Content: Teach tacit, heuristic knowledge as well as textbook knowledge. Collins et al. (1989) refer to four kinds of knowledge that differ somewhat from ID taxonomies:

• Domain knowledge is the conceptual, fac- tual, and procedural knowledge typically found in textbooks and other instructional materials. This knowledge is important, but often is insufficient to enable students to approach and solve problems independently.

• Heuristic strategies are "tricks of the trade" or "rules of thumb" that often help narrow solution paths. Experts usually pick up heuristic knowledge indirectly through repeated problem-solving practice; slower learners usually fail to acquire this subtle knowledge and never develop competence. There is evidence, however, that at least some heuristic knowledge can be made explicit and represented in a teachable form (Chi, Gla- ser, & Farr, 1988).

° Control strategies are required for students to moni tor and regulate their problem- solving activity. Control s trategies have monitoring, diagnostic, and remedial components; this kind of knowledge is often termed metacognition (Paris & Winograd, 1990).

• Learning strategies are strategies for learning; they may be domain , heurist ic , or control s t rategies a imed at learning. In- quiry teaching, to some extent, directly models expert learning strategies (Collins & Stevens, 1983).

ID taxonomies (Gagn6, 1985; Merrill, 1983) pertain primarily to domain or textbook knowledge, al though the distinction is not explicit. Gagn6's "cognitive strategies" fit the last three strategies listed by Collins et al. (1989). Mer- rill 's "f ind a principle" or "find a procedure" seems closest to those three strategies. Both Gagn~ and Merrill are least specific about instruction of this type of learning, although they both acknowledge its importance. They recently have developed a framework for integrating learning goals into more holistic structures (Gagn6 & Merrill, 1990).

2. Situated learning: Teach knowledge and skills in contexts that reflect the way the knowledge will

COGNfflVE TEACHING MODELS 51

be useful in real life. Brown, Collins, and Duguid (1989) argue for placing all instruction within "authentic" contexts that mirror real-life problem-solving situations. Collins (1991) is less forceful, moving away from reaMife requirements and toward problem-solving situations: For teaching math skills, situated learning could encompass settings "ranging from running a bank or shopping in a gro- cery store to inventing new theorems or find- ing new proofs. That is, situated learning can incorporate situations from everyday life to the most theoretical endeavors" (p. 122).

Collins (1991) differentiates a situated-learning approach from common school approaches:

We teach knowledge in an abstract way in schools now, which leads to strategies, such as depending on the fact that everything in a particular chapter uses a single method, or storing information just long enough to retrieve it for a test. Instead of trying to teach abstract knowledge and how to apply it in contexts (as word problems do), we advocate teaching in multiple contexts and then trying to generate across those contexts. In this way, knowledge becomes both specific and general. (p. 123)

Collins cites several benefits for p l a ~ g instruction within problem-solving contexts:

• Learners learn to apply their knowledge under appropriate conditions.

• Problem-solving situations foster invention creativity (see also Perkins, 1990).

• Learners come to see the implications of new knowledge. A common problem inherent in classroom learning is the question of relevance: "How does this relate to my life and goals?" When knowledge is acquired in the context of solving a meaningful problem, the question of relevance is at least partly answered.

• Knowledge is stored in ways that make it accessible when solving problems. People tend to retrieve knowledge more easily when they return to the setting of its acquisition. Knowledge learned while solving problems gets encoded in a way that can be accessed again in problem-solving situations.

Several of the models reviewed below, including some not cited by Collins et al. (1989), place instruction within a problem-solving context and at least some simulation of a real-life setting.

Taken in its extreme form, which would require "authentic," real-life contexts for all learning, the notion of situated learning is somewhat vague and unrealistic. Instruction always involves the dual goals of generalization and differentiation (Gagn6 & Driscoll, 1989). In its more modest form, however, the idea of context-based learning has considerable appeal. Gagn6 and Merrill (1990) have pointed to the need for better integration of learning goals through problem-solving "trans- actions." The notion of situated learning, however it is viewed, challenges the conditions-of-learning paradigm that prescribes the breaking down of tasks to be taught out of context.

3. Modeling and explaining: Show how a process unfolds and tell reasons why it happens that way. Collins (1991) cites two kinds of modeling: modeling of processes observed in the world and modeling of expert performance, including covert cognitive processes. Computers can be used to aid in the modeling of these processes. Collins stresses the importance of integrating both the demonstration and the explanation during instruction. Learners need access to explanations as they observe details of the modeled performance. Computers are particularly good at modeling covert processes that otherwise would be difficult to observe. Collins suggests that truly modeling compe- tent performance, including the false starts, dead ends, and backup strategies, can help learners more quickly adopt the tacit forms of knowledge alluded to above in the section on content. Teachers in this way are seen as "intelligent novices" (Bransford et al., 1988). By seeing both process modeling and accom- panying explanations, students can develop "conditionalized" knowledge, that is, knowledge about when and where knowledge should be used to solve a variety of problems.

ID models presently incorporate modeling and demonstration techniques. Tying explanations to modeled performances is a useful idea, similar to Chi and Bassok's (1989) studies of worked-out examples. Again, the emphasis is on making tacit strategies more explicit by directly modeling mental heuristics.

52 ETI~&.D, Vol, 39, No. 4

4. Coaching: Observe students as they try to complete tasks and provide hints and helps when needed. Intelligent tutoring systems sometimes embody sophist icated coaching systems that model the learner's progress and provide hints and support as practice activities increase in difficulty. Burton and Brown (1979) developed a coach to help learners with "How the West Was Won," a game originally implemented on the PLATO system. Anderson, Boyle, and Yost (1985) developed coaches for geometry and LISP programming. Clancey (1986) and White and Frederiksen (1986) include coaches in their respective programs on medical diagnosis and electric circuits, both discussed later.

Coaching strategies can be implemented in a variety of settings. Bransford and Vye (1989) identify several characteristics of effective coaches:

• Coaches need to monitor learners' performance to prevent their getting too far off base, but leaving enough room to allow for a real sense of exploration and problem solving.

• Coaches help learners reflect on their performance and compare it to others'.

• Coaches use problem-solving exercises to assess learners' knowledge states. Miscon- ceptions and "buggy" strategies can be identified in the context of solving problems; this is particularly true of computer-based learning environments (Larkin & Chabay, 1989).

• Coaches use problem-solving exercises to create the "teachable m o m e n t . " Posner, Strike, Hewson, and Gertzog (1982) present a four-stage model for conceptual change: (1) students become dissatisfied with their misconceptions; (2) they come to a basic understanding of an alternative view; (3) the alternative view must appear plau- sible; and (4) they see the new view's value in a variety of new situations (see also Strike & Posner, in press).

Coaching probably involves the most "instructional work" (cf. Bunderson & Inouye, 1987) of any of the cognitive apprenticeship methods. Short of one-on-one tutoring, coaching is likely to be partial and incomplete. Coop- erative learning and smal l -group learning

methods can provide some coaching support for individual performance. Computers can help tremendously in monitoring learner performance and providing real-time helps, yet presently coaching is only fully implemented in resource-intensive intelligent tutoring systems. Although much work is being done to model the essentials of coaching functions on computer systems, resource-efficient methods for achieving the coaching function are needed. Still at issue is when and how coaching should be incorporated into instruction, particularly as it relates to learner errors and misconceptions. For example, should learners be permit- ted to fail, perhaps even required to learn from failure, as in Clancey's medical tutor?

5. Articulation: Have students think about their actions and give reasons for their decisions and strategies, thus making their tacit knowledge more explicit. Think-aloud protocols are one example of articulation (Hayes & Flower, 1980; Smith & Wedman, 1988). Collins (1991) cites the benefits of added insight and the ability to compare knowledge across contexts. As learners' tacit knowledge is brought to light, that knowledge can be recruited to solve other problems.

John Anderson (1990) has suggested that procedural knowledge---the kind in which people can gain automaticity--is initially encoded as declarative or conceptual knowledge, which later fades as the skill becomes proceduralized. Methods for articulating tacit knowledge help to restore a conscious awareness of those lost strategies, enabling more flexible performance. Traditional ID models suggest practicing problem solving to learn problem solving, but are surprisingly lacking in specific methods to teach learners to think consciously about covert strategies.

6. Reflection: Have students look back over their efforts to complete a task and analyze their own performance. Reflection is like articulation, except it is pointed backwards to past tasks. Analyz- ing past performance efforts can also involve elements of strategic goal-setting and intentional learning (Bereiter & Scardamalia, 1989). Collins and Brown (1988) suggest four kinds or levels of reflection:


• Imitation occurs when a batting coach dem- onstrates a proper swing, contrasting it with your swing.

• Replay occurs when the coach videotapes your swing and plays it back, critiquing and comparing it to the swing of an expert.

• Abstracted replay might occur by tracing an expert's movements of key body parts such as elbows, wrists, hips, and knees, and comparing those movements to your movements.

• Spatial reification would take the tracings of body parts and plot them moving through space.

The latter forms of reflection seem to rely on technologies--video or computer for con- venient instantiation. Collins (1991) uses Anderson et al.'s (1985) Geometry Tutor as an example of reflective instruction; Clancey's (1986) GUIDON medical tutor requires reflective thinking because so much metacognitive responsibility is given to learners. Both are computer-based programs.

Articulation and reflection are both strategies to help bring meaning to activities that might otherwise be more "rote" and procedural. Reigeluth's (1983b) concern with meaningful learning is indicative of the need; however, much of traditional ID practice tends to undervalue the reflective aspects of performance in favor of getting the procedure right. The risk of ignoring reflection is that learners may not learn to discriminate in applying procedures; they may fail to recognize conditions where using their knowledge would be appropriate and may fail to transfer knowledge to different tasks.

7. Exploration: Encourage students to try out different strategies and hypotheses and o b s e ~ their effects. Collins (1991) claims that through exploration, students learn to set achievable goals and to manage the pursuit of those goals. They learn to set and try out hypotheses and to seek knowledge independently. Real-world exploration is always an attractive option; however, constraints of cost, time, and safety sometimes prohibit instruction in realistic settings. Sim- ulations are one way to allow exploration; hypertext structures are another.

As Reigeluth (1983a) notes, discovery learning techniques are less efficient than direct instruction techniques for simple content acquisition. The choice must depend, however, on the goals of instruction: For near-transfer tasks (cf. Salomon & Perkins, 1988), direct instruction occasionally may be warranted; for far- transfer tasks, learners must learn not only the content, but also how to solve unforeseen problems using the content; in such cases, instructional strategies allowing exploration and strategic behavior become essential (McDaniel & Schlager, 1990).

Having thus represented a traditional ID mode of thinking about exploration, we are still left unsatisfied. Following a constructivist way of thinking, there is something intrin- sicaUy valuable about situated problem-solving activity that makes learning more effective than straight procedural practice. This is an area that needs better artiaxlation, induding a ration- ale for appropriate goals for exploratory and inquiry-based instruction, and effective strategies for reaching those goals.

8. Sequence: Present instruction in an ordering from simple to complex, with increasing diversity, and global before local skills.

• Increasing complexity. Collins et al. (1989) point to two methods for helping learners deal with increasing complexity. First, instruction should take steps to control the complexity of assigned tasks. They cite Lave's study of tailoring apprenticeships: apprentices first learn to sew drawers, which have straight lines, few pieces of material, and no special features like zippers or pock- ets. They progress to more complex gar- ments over a period of time. The second method for controlling complexity is through scaffolding, for example, group or teacher support for individual problem solving.

• Increasing diversity refers to the variety in examples and practice contexts.

• Global before local skills refers to helping learners acquire a mental model of the problem space at very early stages of learning. Even though learners are not engaged in full problem solving, through modeling and help-

54 ETR&D, Vol, 39, No, 4

ing on parts of the task (scaffolding) they can understand the goals of the activity and the way various strategies relate to the problem's solution. Once they have a clear "conceptual map" of the activity, they can proceed to developing specific skills.

The sequencing suggestions above bear a strong resemblance to those of elaboration theory (Reigeluth & Stein, 1983) and component display theory (Merrill, 1983). The notion of teaching global skills before local skills is implicit in elaboration theory; simple-to-complex sequencing is the foundation of elaboration theory. The notion of increasing diversity is the near-equivalent to the prescription to use varied examples and practice activities in concePt or nile teaming. The cognitive apprenticeship extends these notions beyond rule learning to problem-solving contexts. At the same time, a number of research findings call into question the simple-to-complex formula. For a more complete discussion of sequencing issues, see Wilson and Cole (1992).

A SURVEY OF COGNITIVE TEACHING MODELS

In this section, several instructional systems developed by cognitive psychologists are reviewed briefly. Some of these systems were cited by Collins and colleagues as exemplify- ing cognitive apprenticeship features. Others were cited by Glaser and Bassok (1989) as incor- porating the best new knowledge coming out of cognitive psychology. Still others were selected because they embody noteworthy design concepts. After each program is de- scribed individually, a comparison is made between the models in an effort to identify common and not-so-common themes for discussion.

Anderson's Intelligent Tutors

John Anderson accounts for cognitive performance through the ACT* model of information processing (see, e.g., Anderson, 1990). According to this model, declarative knowledge is compiled into procedural skill through repeated practice at performing a task. Whereas a novice must keep a procedure's steps in mind

while performing, an expert tums several steps into an integrated routine, similar to the way subroutines work together in a computer program. Once knowledge is compiled from declarative to procedural knowledge, the procedure may be performed with a minimum of allocated conscious attention.

Instruction based on the ACT* model would therefore emphasize guiding the learner through repeated practice opportunities to proceduralize the skills of the curriculum. Anderson and colleagues at Carnegie Mellon University have tested out this instructional approach through a series of intelligent tutoring systems aimed at teaching well-defined procedural skills, including programming in LISP (Anderson, Farrell, & Sauers, 1984), generating proofs for geometric theorems (Anderson et al., 1985), and solving algebraic equations (Lewis, Milson, & Anderson, 1987). Based on a cognitive analysis of the task, a model of the ideal student performance is developed for varying stages of competence, including assorted "buggy" procedures. As instruction proceeds, the system fits the learner's response pattern to its performance model, selects problems to minimize errors and opti- mize learning, and provides feedback and remediation accordingly. The technique of comparing learners' performance with a preexisting performance model is termed model tracing. The program does not make available the model 's production rules to students directly; rather, the production rules trigger various instructional events, most notably intervention and feedback following incorrect performance.

Because new knowledge is best learned in the context of solving relevant problems, instruction is centered around problem-solving practice. Glaser and Bassok (1989) summarize the intelligent tutoring approach:

The CMU group advocates shortening preliminary instructional texts, and restraining from elaborated explanations. Texts should focus on procedural info~ mation, and students should begin actual problem solving as soon as possible. Although textual instruc- tions should be carefully crafted to maximize correct encoding, the inevitable misunderstandings should be corrected [immediately] during problem solving. (p. 638)

COGNITIVE TEACHING MODELS 5~

Clancey's Intelligent Tutoring Environments

Clancey's (1986) program GUIDON and its descendents use heuristic classification methods as the basis for an intelligent tutoring environment for medical diagnosis. The program differs from ACT* model prescriptions in several ways. First, it assumes that the learner has a basic understanding of terms, concepts, and disease processes. Second, it assumes that learning is more efficient if the student deter- mines what he or she needs to know next with- out being explicitly controlled by the system. Third, and most significantly, it is failure driven; that is, primary instruction occurs in the form of feedback to student errors.

GUIDON requires the student to make a diagnosis and then justify it with reasons. When the student's diagnosis "fails," he or she must take steps to correct the reasoning that led to it. Thus, while the student develops expertise in medical diagnosis in a realistic problem- solving context, he or she also learns to detect and correct "buggy" procedures and misconceptions. The program is nearly completely learner-controlled; at any point, the student can choose to browse through the expert taxonomies and tables, examine the expert's reasoning during problem solving, ask questions, or request explanations. Ultimately, however, the student must generate the appropriate links in a solution graph for each case.

Qualitative Mental Models

White and Frederiksen's (1986) program to teach troubleshooting in electrical circuits emphasizes the relationship between qualitative models and causal explanations. White and Frederiksen believe that mastery of qualitative reasoning should precede quantitative reasoning. Their program builds on students' intuitive understandings of the domain, carefully sequencing "real-world" problems that require the student to construct increasingly complex qualitative models of the domain. Although the program encourages students to engage in diverse learning strategies (exploring, requesting explanations, viewing tutorial demonstrations, or problem solving), it tries

to minimize errors. Thus, like the ACT* model, it does not directly address "buggy" algorithms and misconceptions.

Reciprocal Teaching

Brown and Palincsar (1989; Palinscar & Brown, 1989) have developed a cooperative learning system for the teaching of reading, termed reciprocal teaching. The teacher and learners assemble in groups of 2 to 7 and read a paragraph together silently. A person assumes the "teacher" role and formulates a question on the paragraph. This question is addressed by the group, whose members are playing roles of "producer" and "critic" simultaneously. The "teacher" advances a summary and makes a prediction or clarification, if any is needed. The role of teacher then rotates, and the group proceeds to the next paragraph in the text. Brown and colleagues have also developed a method of assessment, called dynamic assessment, based on successively increasing prompts on a realistic reading task. The reciprocal teaching method uses a combination of modeling, coaching, scaffolding, and fading to achieve impressive results, with learners showing dramatic gains in comprehension, retention, and far transfer over sustained periods.

Procedural Facilitations for Writing

Novices typically employ a knowledge-teUing strategy when writing: They think about their topic, then write their thought down, think again, then write the next thought down, and so on uni t they have exhausted their thoughts about the topic. This strategy, of course, is in conflict with a more constructive, planning approach in which writing pieces are com- posed in a more coherent, intentional way.

To encourage students to adopt more sophisticated writing strategies, Scardamalia and Bereiter (1985) have developed a set of writing prompts called procedural facilitations that are designed to reduce working-memory demands and provide a structure for com- pleting writing plans and revisions. Their system includes a set of cue cards for different

56 ~R~D, Vol. 39, No, 4

purposes of writing, structured under five headings:

• new idea (e.g., ' ~ n even better idea i s . . . " ; '~An important point I haven't considered yet i s . . . " )

• improve ("I could make my point clearer • , , " )

• elaborate ( '~n example of t h i s . . . " ; '~4 good point on the other side of the argument i s . . . " )

• goals ("My purpose i s . . . " )

• put t ing it together ("I can tie this together b y . . . ")

Each prompt is written on a notecard and drawn by learners working in small groups. The teacher makes use of two techniques, solo-

ing and co-investigation. Soloing gives learners the opportunity to try out new procedures by themselves, then return to the group for critique and suggestions• Co-investigation is a process of using think-aloud protocols that enables learner and teacher to work together on writing activities. It allows for more direct modeling and immediate direction. Bereiter and Scardamalia (1987) have found up to ten- fold gains in learning indicators, with nearly every learner improving his or her writing through the intervention.

Schoenfeld's Math Teaching

Schoenfeld (1985) studied methods for teaching math to college students. He developed a set of heuristics that were helpful in solving math problems. His method introduces those heuristics, as well as a set of control strategies and a productive personal belief system about math, to students. Like the writing and reading systems, Schoenfeld's system includes explicit modeling of problem-solving strategies, and a series of structured exercises afford- ing learner practice in large and small groups, as well as individually. He employs a tactic he calls "postmortem analysis," retracing the solution of recent problems and abstracting out the generalizable strategies and components. Unlike the writing and reading systems,

Schoenfeld carefully selects and sequences practice cases to move learners into higher levels of skill.

Another interesting technique is the equivalent to "stump the teacher," wherein time at the beginning of each class period is de- voted to learner-generated problems that the teacher is challenged to solve. By witnessing the occasional false starts and dead ends of the teacher's solution, learners acquire a more appropriate belief structure about the nature of expert mathematical problem solving. Schoenfeld's positive research findings support a growing body of math research that suggests the importance of acquiring a conceptual or schema-based representation of mathematical problem solving.

Anchored Instruction

John Bransford and colleagues at Vanderbilt University have developed several instructional products using video settings• Films such as Young Sherlock Holmes and Indiana Jones provide macrocontexts within which problems of various kinds can be addressed. For example, when Indiana Jones quickly replaces a bag of sand in place of the gold idol, the booby trap is tricked into thinking the idol is still there. This scene opens up questions of mass and density: If the idol were solid gold, how big would a sand bag need to be to weigh the same, and could Indy have escaped as he did carrying a solid-gold idol of that size?

Based on a single macrocontext, learners may approach a variety of problems that draw on science, math, language, and history. Bransford and colleagues have applied the name anchored instruction to this approach of grounding instruction in information-rich situations, and have reported favorable findings in field-based and laboratory studies (The Cog- nition and Technology Group at Vanderbilt, 1990). Recently, several videodisc-based macrocontexts called the Jasper Series have been developed as a basis for research and dass~om instruction (Cognition and Technology Group at Vanderbilt, in press; Sherwood and The Cognition and Technology Group at Vander- bilt, 1991; Van Haneghan et al., in press).


Cognitive Flexibility Hypertexts

Spiro and colleagues (Spiro, Coulson, Felto- vich, & Anderson, 1988; Spiro & Jehng, 1990) have developed hypertext programs to address problems typically associated with acquiring knowledge in complex, ill-structured domains. The programs utilize videodiscs that construct multiple "texts" (audio/video mini-cases, about 90 seconds each) of a domain. The multiple representations in KANE (Exploring Thematic Structure in Citizen Kane) induce students to restructure their thinking as they grapple with the thematic issues that are presented through the minicases. The program codes various themes and automatically generates appropri- ately juxtaposed sequences when the learner selects a theme to explore. It relies on the learner to generate the relationships embod- ied in a set of mini-cases.

Spiro's programs are best thought of as "rich" environments that allow sophisticated learners to pursue their learning goals in a flexible way. They do not typically include skill practice in a traditional sense, but instead rely on learner purposes and externally imposed assignments to give meaning to student brows- ing. However, there is an authoring shell built into the system for teachers or students to use. Students, for example, may construct a series of mini-cases into a "visual essay" illustrating a theme not present on the Theme menu.

Like Clancey's intelligent tutoring environments , KANE assumes that s tudents have basic knowledge of the content (in this in- stance, they must have seen the full movie in its natural sequence at least once).

The idea of mini-cases is central to Spiro's approach. Spiro and Jehng (1990) emphasize the differences between actual experiences and mini-cases:

a) the cases are immediately juxtaposed (hours and days do not pass between nonroutine cases);

b) the cases are thematically related (whereas there is no guarantee of instructive relatedness across naturally occurring adjacent cases);

c) the cases are stripped down to structurally significantly features, making it easier to extract dimensions of structural relatedness;

d) the cases are accompanied by expert commen- tary and guidance; and, finally,

e) because the cases are short, they are much easier to remember and more of them can be presented in a short amount of time, facilitat- ing the recognition of relationships across cases. (p. 202; reformatted)

Spiro and Jehng emphasize that this instructional approach is difficult it places great metacognitive demands on learners--but it addresses goals which are often overlooked in instruction precisely because they are difficult.

Instruction designed around mini-cases is similar to Schank's use of stories in instruction (Riesbeck & Schank, 1989; Schank & Jona, 1991). Schank's approach includes videotap- ing and assembling a database of short "stories" or cases from experts, then accessing and sequencing those stories according to the needs of the learner.

Learning Through Design Activities: Computer Tools

Harel and Papert (1990) and Lippert (1988, 1990) have examined the effect of learning a content domain while simultaneously learning other domains, particularly software design.

Instructional Software Design Project

Harel and Papert (1990) created a "total learning environment" in which fourth-graders simultaneously learned fractions and Logo programming. The students" goal was to design and develop a Logo program to teach something about fractions. "The 'experimental treatment ' integrated the experimental children's learning of fractions and Logo with the designing and programming of instructional software" (p. 7) four hours a week dur ing a 15-week semester. The students in the first control group spent the same amount of time learning Logo, but with a different purpose. The second control group programmed only once a week. Each group completed a two- month unit on fractions as part of the regular math curriculum.

The experimental treatment purposely cul- tivated a sense of meaningful activity, colle- giality, and metacognit ive awareness. The

~ ETR&D, Vol, 39, No. 4

students had primary control over the task of designing and developing individual programs to" 'explain something about fractions' to some in tended audience" (p. 3). Initially, Harel guided the students and teacher in a discussion of instructional design and educational software; she also provided models of designs, flowcharts, and screens from her previous pro- jects and shared notes and excerpts from programs developed by "real" designers and engineers. As needed, the teacher and Harel led discussions on issues related to understanding and teaching fractions, as well as on specific Logo programming skills.

On each day that the students worked on the project, they initially spent 5 to 7 minutes writing designs and plans in special notebooks and 45 to 55 minutes working on their computers, and then wrote about the problems they had encountered and changes they had made; occasionally they also included designs for the next session's activity. "The only two requirements were: (1) that they write in their Designer Notebooks before and after each working session; and (2) that they spend a specific amount of time at the computer each day" (p. 4) in order to fit the project into the schedule.

In many respects the students, teacher, and researcher were like a community of scholars jointly in pursuit of knowledge, or a community of craft apprentices. At all times, students had access to "experts" and to continual evaluation; they could compare their work with each other's work and to the work of experts, all representing various stages of expertise.

By the end of the study, not only did the experimental group significantly exceed both control groups in mastery of fractions, but on some items the students scored twice as high as sixth- to eighth-graders. The students were also more persistent than students in both control groups in trying to solve various Logo programming problems, and they developed more metacognitive sophistication: They could find problems, develop plans, discard or revise inef- ficient plans, and control distractions and anxiety.

Harel and Papert hypothesize that no single factor alone contributed to the students" achievements. Instead, they conclude that several factors interrelated in a holistic approach

which may include but is not limited to:

the affective side of cognition and learning; the children's process of personal appropriation of knowledge; the children's use of LogoWriter; the children's constructivist involvement with the deep structure of fractions knowledge; the integrated-learning principle; the learning-by-teaching principle; and the power of design as a learning activity. (p. 30)

Expert Systems

Lippert (1988, 1990) describes the way in which an expert system shell can be used as both an instructional and a learning strategy to "facilitate the acquisition of procedural knowledge and problem-solving skills in difficult top- ics" (1988, p. 22). Again, students learn by des ign ing- -deve lop ing an expert system individually or in groups, on their own or under the guidance of a teacher. According to Lippert, the strategy can be used with students as young as grade six and in any domain whose knowledge base can be expressed in productions.

Like the Harel and Papert approach, developing an expert system forces students to construct a meaningful representat ion of the domain. Most expert systems are systems which reduce a content domain to a set of "if . . . then" rules. According to Lippert's scheme, the knowledge base is the key component and includes four parts: decisions that define the domain, questions that extract information (answers) from the user, rules that relate the answers to the decisions, and explanations (of questions or rules) that require the developer to understand the relationships among the various elements of the domain- - the learner must unders tand " w h y " and "when," not merely "what." The developer constructs the knowledge base, which the system then evaluates; if the system finds incon- sistendes or redundancies, the developer must revise the knowledge base. In doing so, the learner must be reflective and articulate his or her implicit knowledge. Developing such a system helps students confront their misconceptions of the content.

In designing the system, students can experiment, trying out ideas and revising them as necessary until they are satisfied with their representation of the domain. They can ini-


tially restrict their system to one component of a domain and expand it to accommodate their expanding knowledge base. As in Harel and Papert's Logo environment, when students work in groups they are exposed to and stimulated by different representations of the domain; and, like the students in the Logo environment, students who design an expert system learn much incidental information and often become enthusiastic about learning the content.

Several issues remain unresolved with respect to learning through design activities. Like many prototype programs, factors con- tribufing to the success of the programs are largely embedded within the expertise of the researchers and teachers. The critical factors in the instructional design need to be further explicated; this will allow the approach to be more easily identified and exported to different settings. A key issue is the question of effi- ciency: Can learning through design make

• more efficient use of time and resources, or does a design-oriented approach require extra time and resources? Theoretically, learning through design represents a fundamental alternative to traditional instructional designs, and therefore deserves continued attention among ID theorists.

IMPLICATIONS FOR INSTRUCTIONAL DESIGN

In reviewing the set of teaching models de- scribed above, one is struck by the diversity of approaches and features, and particularly by the key features not common to all the model programs.

Problem-Solving versus Skill Orientation

solving. It is worthy of note that the LISP Tutor is the only program that does not include explicit guidance to encourage metacognitive thinking.

Spiro's KANE hypertext program is another unusual problem-solving environment. Spe- cific problems are not posed within the program; learners do not "practice" solving interpretation problems per se. Instead, learners are provided a rich environment to explore and study. Problem solving in this setting depends heavily on the predisposition and prior knowledge of the learner. Thus, the explicit instructional guidance for problem solving is relatively weak, which would likely result in failure for the learner who lacks metacognitive and preexisting content knowledge. The rest of the programs reviewed include higher levels of guidance for problem solving and practice on procedural skills.

A problem-solving orientation toward instruction is not implied by the conditions-of- learning paradigm; in fact, the reverse seems to be true. In the words of an anonymous reviewer of a previous draft of this article, "this is a content issue, not a method issue." Tra- ditional instructional designers "would say that the needs analysis should determine whether or not you have a problem-solving orientation or a skill orientation (or both, or nei ther- - perhaps an understanding orientation)." Hence, from an ID point of view, problem- solving instruction is only one of several kinds, each meeting different needs. A more constructivist point of view, however, would suggest that virtually all instruction should be embedded somehow within a probfem-solving context (e.g., Bransford & Vye, 1989; Wilson, 1989). The heavy bias toward problem solving in the models reviewed seems to be evidence of the viability of such an emphasis.

Overwhelmingly, the models approached instruction from a problem-solving orientation. The main exception is Anderson's LISP Tutor, which essentially teaches procedural skill, with little need for metacognitive reflection. Near transfer is the goal of Anderson's tutoring systems. "Problem solving" in this environment is something like working out problems at the end of a textbook chapter. We do not consider this kind of activity to be authentic problem

Detailed versus Broad Cognitive Task Analysis

Most of the models base instruction on a careful and detailed cognitive task analysis. These analyses go beyond behavioral observation; subtle and covert mental heuristics are sought and identified. Again, however, there is considerable variation among the teaching models. Intelligent tutors demand an exhaustive

60 ETR~, Vol, 39, No. 4

representation of the content; by contrast, the skills taught by group methods such as reciprocal teaching or the fractions software design project are not so well defined in advance.

There is no pretense among any of the pro- jects' designers that the goals of instruction are completely captured by analysis. Rather, the instructional materials and interactions provide a setting wherein students can develop the subtle and often unidentified knowledge needed to succeed in the given task. The distinction between the identified learning goals and the knowledge actually acquired through learning activities is an important one that is often blurred in traditional ID theories. Given the right design, incomplete learning specifi- cations can nonetheless lead toward complete learning outcomes. This attitude toward instructional outcomes helps deflect the criticism that ID is reductionistic because it pretends to capture and transmit expertise outside of the expert (cf. Bereiter, 1991; Dreyfus & Drey- fus, 1986).

Cognitive science has opened up an array of tools for performing task/content analyses, including think-aloud protocols, computer modeling, networks, maps, production systems, and cognitive process analysis (Gard- ner, 1985; Nelson, 1989; Smith & Wedman, 1988). Instructional designers can represent content with greater precision and accuracy than before. Yet, as important as these ana- lytic tools are, of greater importance to ID are theories of knowledge and learning that are presently being developed within cognitivist and philosophical traditions (e.g., Bereiter, 1991; Lakoff, 1987; Paul, 1990). Faulconer and Williams (1990), for example, outline a Hei- deggerian perspective that helps to resolve the dualistic bias inherent in the current "objec- tivism" versus "constructivism" debate (Duffy & Jonassen, in press). Instructional designers can do more to be sensitive to the variety of conceptions of knowledge and learning, and to develop teaching models in line with those varying conceptions.

Authent ic versus A c a d e m i c Contexts

Brown, Collins, and Duguid (1989) define authentic activity as "the ordinary practices of

the culture" (p. 34). We take this to mean that they have some bearing on real-life, everyday activities. There is a problem, however, in determining what counts as authentic activity. Apparently, the theorizing and inquiry activities of a nuclear physicist or historian qualify as authentic. For the term to have any precise referent, there needs to be a boundary drawn between authentic and unauthen- tic activity, and we are not sure where that boundary lies. Upon examining the set of teaching models, however, there seems to be a continuum of authenticity. Some programs relate more closely to real-life situations (e.g., the Jasper series), but most are more typically academic work (e.g., Anderson's LISP tutor). Some programs seem to fall in the middle (Spiro's KANE or Harel and Papert's fractions project). Clearly, the resemblance between the instructional setting and real-life settings is not a constant among the models.

ID theories have tended toward academic contexts. For example, the Reigeluth volumes (1983a, 1987) contain surprisingly little discussion of simulation as an instructional strategy. The irony is that large numbers of ID practi- tioners are heavily involved in developing "authentic" training, either in job sites or through simulations. Clearly, more attention needs to be given to prescriptions for designing simulations that go beyond present models for direct instruction.

Learner versus System Control

There is also considerable variability on the learner control dimension. Some programs are tightly controlled (e.g., LISP Tutor) while others are very flexible (e.g., GL~DON, KANE), There seems to be a correspondence between the degree of learner control and the metacognitive goals of the instruction. Where metacognitive development is a goal, there is some flexibility allowing learners to assume responsibility over the "zone of proximal development" (Scardamalia & Bereiter, 1991). Where metacognitive skill is not a goal of instruction, learner control is less critical.

Yet even this generalization is not complete. Schoenfeld's math program is very much oriented toward metacognitive development, yet


prob lem cases are carefully p l anned and sequenced in advance by the instructor. Within the framework of the cases, however, learners are able to negot ia te control wi th the instructor. The p lanned sequencing of cases may be accounted for by how well defined the content is, which allows for a systematic progression from simple to complex cases. The same can be said for White and Frederiksen's (1986) progression of mental models. Learn- ers may experiment and control actions within levels, but movement from level to level is p l a n n e d in advance according to conten t parameters . With less defined content (e.g., reciprocal teaching), sequence of cases is, at least on the surface, much less important . Another view of reciprocal teaching, however, is that the text being read is not the content; rather, the content lies in the skills being modeled and practiced, and the sequencing, under the information control of the teacher, does proceed from simple to complex.

Error-Restricted versus Error-Driven Instruction

GUIDON is the pr ime example of a program whose instructional presentations are triggered by the learner ' s errors. Presenta t ions and explanations thus take the form of feedback to learner actions. In contrast, several programs seek to control errors through careful presentations and sequencing. Errors within the LISP Tutor or White and Frederiksen's troubleshooting tutor would be due to mismatches between the learner a n d the ins t ruct ional plan; the learner can then be corrected and brought back on track. Most of the programs seem to fall in the middle on this feature; for example, errors in Schoenfe ld ' s ma th p rog ram are expected, even modeled by the instructor, but the instruction is not completely keyed around diagnosis and response to s tudent errors.

Feedback is an old topic of research, but we know surprisingly little about it. It has long been recognized that the best way to ensure learners' unders tanding is to have them re- spond actively. At the same time, old maxims such as " 'provide immedia te knowledge of resul ts" are be ing chal lenged (Corbet t & Ander son , 1991; Lepper & Chabay, 1988; Salmoni, Schmidt, & Walter, 1984). There is

)

some evidence that somet imes cont inuous , detailed feedback actually may serve to sup- plant the self-reflective processes that a learner would otherwise apply to a problem. Error- driven (i.e., feedback-heavy) instruction should be designed so as not to interfere with such learning processes.

CONCLUSION

This article has been concerned with the rela- t ionship between two related disciplines: cognitive psychology and instructional design. ID, the more applied discipline, faces the challenges of constantly reinventing itself as new research in basic psychology sheds light on learning processes. It is no cause for concern that ID models need continual self-review; indeed, there would be cause for concern if the situation were otherwise. To provide some sense of context, however, we would like to call attention to some basic principles of cognitive apprenticeship articulated by the phi losopher Erasmus more than 500 years ago. In a letter to a s t ude n t fr iend, Erasmus offers some advice:

Your first endeavor should be to choose the most learned teacher you can find, for it is impossible that one who is himself no scholar should make a scholar of anyone else. As soon as you find him, make every effort to see that he acquires the feel- ings of a father towards you, and you in turn those of a son towards him . . . . Listen to your teacher's explanations not only attentively but eagerly. Do not be satisfied simply to follow his discourse with an alert mind; try now and then to anticipate the direction of his thought . . . . Do not be ashamed to ask questions if you are in doubt; or to be put right whenever you are wrong. (Erasmus, 1974, pp. 114-115)

At a time when terms like metacognition, scaffolding, and cognitive apprenticeship are being invented to describe the learning process, it is humbl ing to be reminded of the insights of former paradigms. ID theorists may chafe at the continuing need to revise their theories in light of advances in psychological theory, but it is good for both fields for the dialogue to continue. Indeed, the interaction between basic psychology and appl ied instructional design can be expected to continue for some time to come. [ ]

62 ETR~, VoL 39, No. 4

We would like to thank Jim Cole for calling our attention to the works of Erasmus. We also thank anonymous reviewers for their suggestions. Please send requests for reprints to Brent Wilson, University of Colorado at Denver, Campus Box 106, P. O. Box 173364,Denver CO 80217-3364. Telephone: (303)556-4363. E-mail: bwilson@cudenver.

REFERENCES

Anderson, J. R. (1990). Cognitive psychology and its implications (3rd ed.). New York: Freeman.

Anderson, J. R., Boyle, C. F., & Yost, G. (1985). The Geometry Tutor. In Proceedings of the International Joint Conference on Artificial Intelligence (pp. 1-7). Los Altos, CA: International Joint Conference on Artificial Intelligence.

Anderson, J. R., Farrell, R., & Sauers, R. (1984). Learning to program in LISP. Cognitive Science, 8, 87-129.

Andrews, D. H., & Goodson, L. A. (1980). A com- parative analysis of models of instructional design. Journal of Instructional Development, 3(4), 2-16.

Bereiter, C. (1991, April). Implications of connec- tionism for thinking about rules. Educational Researcher, 10-16.

Bereiter, C., & Scardamalia, M. (1987). The psychology of written composition. Hillsdale, NJ: Lawrence Erlbaum.

Bereiter, C., & Scardamalia, M. (1989). Intentional learning as a goal of instruction. In L. B. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Gtaser (pp. 361-392). Hillsdale, NJ: Lawrence Erlbaum.

Bransford, J. D., Goin, L. I., Hasselbring, T. S., Kinzer, C. K., Sherwood, R. D., & Williams, S. (1988). Learning with technology: Theoretical and empirical perspectives. Peabody Journal of Educa- tion, 64(1), 5-26.

Bransford, J. D., & Vye, N. J. (1989). A perspective on cognitive research and its implications for instruction. In L. B. Resnick & L. E. Klopfer (Eds.), Toward the thinking curriculum: Current cognitive research. Alexandria, VA: Yearbook of the Associa- tion for Supervision and Curriculum Development.

Branson, R. K., & Grow, G. (1987). Instructional systems development. In R. M. Gagn6 (Ed.), Instructional technology: Foundations. Hillsclale, NJ: Lawrence Erlbaum.

Brown, J. S., Collins, A., & Duguid, P. (1989, January-February). Situated cognition and the culture of learning. Educational Researcher, 32-42.

Brown, A., & Palincsar, A. S. (1989). Guided, cooperative learning and individual knowledge acquisition. In L. B. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Glaser (pp. 393-451). Hillsdale, NJ: Lawrence Erlbaum.

Bunderson, C. V., & Inouye, D. K. (1987). The evaluation of computer-aided educational delivery sys-

terns. In R. M. Gagn6 (Ed.), Instructional technology: Foundations. Hillsdale, NJ: Lawrence Erlbaum.

Burton, R. R., & Brown J. S. (1979). An investigation of computer coaching for informal learning activities. International Journal of Man-Machine Studies, 11, 5-24.

Case, R. (1978). A developmentally based theory and technology of instruction. Review of Educa- tional Research, 48, 439-463.

Case, R., & Bereiter, C. (1984). From behaviourism to cognitive behaviourism to cognitive development: Steps in the evolution of instructional design. Instructional Science, 13, 141-158.

Chi, M. T. H., & Bassock, M. (1989). Learning from examples via self-explanations. In L. B. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Glaser (pp. 251-282). Hitlsdale, NJ: Lawrence Erlbaum.

Chi, M. T. H., Glaser, R., & Farr, M. J. (1988). The nature of expertise. Hillsdale, NJ: Lawrence Erlbaum.

Clancey, W. J. (1986). From GUIDON to NEOMYCIN and HERACLES in twenty short lessons: ONR final report, 1979-1985. AI Magazine, 7(3), 40-60.

Cognition and Technology Group at Vanderbilt. (1990, August-September). Anchored instruction and its relation to situated cognition. Educational Researcher, 2-10.

Cognition and Technology Group at Vanderbilt. (in press). Anchored instruction approach to cognitive skills acquisition and intelligent tutoring. In J. W. Regian & V. J. Shute (Eds.), Cognitive approaches to automated instruction. Hillsdale, NJ: Lawrence Erlbaum.

Collins, A. (1991). Cognitive apprenticeship and instructional technology. In L. Idol & B. E Jones (Eds.), Educational values and cognitive instruction: Implications for reform. Hillsdaie, NJ: Lawrence Erlbaum.

Collins, A., & Brown, J. S. (1988). The computer as a tool for learning through reflection. In H. Mandl & A. Lesgold (Eds.), Learning issues for intelligent tutoring systems~ New York: Springer.

Collins, A., Brown, J. S., & Newman, S. E. (1989). Cognitive apprenticeship: Teaching the crafts of reading, writing, and arithmetic. In L. B. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Glasser (pp. 453-494). Hillsdale, NJ: Lawrence Erlbaum.

Collins, A., & Stevens, G. (1983). Inquiry teaching. In C. M. Reigeluth (Ed.), Instructional-design theories and models: An overview of their current status (pp. 247-278). Hillsdale, NJ: Lawrence Erlbaum.

Corbett, A. T., & Anderson, J. R. (1991, April). Feed- back control and learning to program with the CMU Lisp Tutor. Paper presented at the meeting of the American Educational Research Association, Chicago.

Dreyfus, H., & Dreyfus, S. (1986). Mind over machine: The power of human intuition and expertise in the era of the computer. New York: The Free Press.

Duffy, T., & Jonassen, D. H. (in press). Instructional


principles for constructivist learning environments. Hillsdale, NJ: Lawrence Erlbaum.

Eckert, P. (1990). Adolescent social categories-- Information and science learning. In J. Gardner, J. G. Greeno, E Reif, A. H. Schoenfeld, A. A. diSessa, & E. Stage (Eds.), Toward a scientific practice of science education. Hillsdale, NJ: Lawrence Erlbaum.

Erasmus, D. (1974). The correspondence of Erasmus. Volume 1: Letters 1 to 141, 1484 to 1500. Trans- lated by R. A. B. Mynors and D. E S. Thomson. Toronto: University of Toronto Press.

Faulconer, J. E., & Williams, R. N. (1990). Recon- sidering psychology. In J. E. Faulconer & R. N. Williams (Eds.), Reconsidering psychology: Perspec- tives from continental philosophy. Pittsburgh: Du- quesne University Press.

Gagn6, R. M. (1966). The conditions of learning (1st ed.). New York: Holt, Rinehart, & Winston.

Gagn6, R. M. (1985). The conditions of learning (4th ed.) New York: Holt, Rinehart and Winston.

Gagn6, R. M., & Driscoll, M. P. (1989). Esssentials of learning for instruction (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.

Gagn6, R. M., & Merrill, M. D. (1990). Integrative goals for instructional design. Educational Technol- ogy Research & Development, 38(1), 23-30.

Gardner, M. K. (1985). Cognitive psychological approaches to instructional task analysis. In E. W. Gordon (Ed.), Review of Research in Education, 12, 157-195.

Glaser, R. (1982). Instructional psychology: Past, present, and future. American Psychologist, 37(2), 292-305.

Glaser, R., & Bassok, M. (1989). Learning theory and the study of instruction. Annual Review of Psy- chology, 40, 631-666.

Gott, S. P. (1988a, April). The lost apprenticeship: A challenge for cognitive science. Paper presented at the meeting of the American Educational Research Association, New Orleans.

Gott, S. P. (1988b). Apprenticeship instruction for real-world tasks: The coordination of procedures, mental models, and strategies. In E. Z. Rothkopf (Ed.), Review of Research in Education, 15, 97-169.

Harel, I., & Papert, S. (1990). Softwar6 design as a learning environment. Interactive Learning Envi- ronments, 1, 1-32.

Hayes, J. R., & Flower, L. S. (1980). Identifying the organization of writing processes. In L. W. Gregg & E. R. Steinberg (Eds.), Cognitive processes in writing. HiUsdale, NJ: Lawrence Erlbaum.

Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: Uni- versity of Chicago Press.

Larkin, J. H., & Chabay, R. W. (1989). Research on teaching scientific thinking: Implications for computer-based instruction. In L. B. Resnick & L. E. Klopfer (Eds.), Toward the thinking curriculum: Current cognitive research. Alexandria, VA:

Yearbook of the Association for Supervision and Curriculum Development.

Lepper, M. R., & Chabay, R. W. (1988). Socializing intelligent tutors: Bringing empathy to the computer tutor. In H. Mandl& A. Lesgold (Eds.), Learning issues for intelligent tutoring systems (pp. 242-257). New York: Springer-Verlag.

Lewis, M. W., Wilson, R., & Anderson, J. R. (1987). Designing an intelligent authoring system for high school mathematics. In G. Kearsley (Ed.), Artificial intelligence and instruction: Applications and methods. Reading, MA: Addison-Wesley.

Lippert, R. C. (1988, March). An expert system shell to teach problem solving. TechTrends, 22-26.

Lippert, R. (1990). Teaching problem solving in mathematics and science with expert systems. Journal of Artificial Intelligence in Education, •(3), 27-40.

McDaniel, M. A., & Schlager, M. S. (1990). Discov- ery learning and transfer of problem-solving skills. Cognition and Instruction, 7(2), 129-159.

Merrill, M. D. (1983). Component display theory. In C. M. Reigeluth (Ed.), Instructional-design theories and models: An overview of their current status. (pp. 279-333). Hillsdate, NJ: Lawrence Erlbaum.

Merrill, M. D., Kowallis, T., & Wilson, B. G. (1981). Instructional design in transition. In E Farley & N. Gordon (Eds.), Psychology and education: The state of the union. Chicago: McCutcheon.

Nelson, W. A. (1989). Artificial intelligence knowledge acquisition techniques for instructional development. Educational Technology Research & Development, 37(3), 81-94.

Newman, D., Griffin, E, &Cole, M. (1989). Thecon- struction zone: Working for cognitive change in school. Cambridge: Cambridge University Press.

Palincsar, A. M., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and monitoring activities. Cognition and Instruction, 1(2), 117-175.

Paris, S. G., & Winograd, P. (1990). How metacognition can promote academic learning and instruction. In B. E Jones & L. Idol (Eds.), Dimensions of thinking and cognitive instruction. Hillsdale, NJ: Law- rence Erlbaum.

Paul, R. W. (1990). Critical and reflective thinking: A philosophical perspective. In B. F. Jones & L. Idol (Eds.), Dimensions of thinking and cognitive instruction (pp. 445-494). Hillsdale, NJ: Lawrence Erlbaum.

Perkins, D. N. (1990). The nature and nurture of creativity. In B. E Jones & L. Idol (Eds.), Dimen- sions of thinking and cognitive instruction. Hillsdale, NJ: Lawrence Erlbaum.

Posner, G. K., Strike, K. A., Hewson, P. W., & Gertzog, W. A. (1982). Accommodation of scientific conception: Toward a theory of conceptual change. Science Education, 66, 211-227.

Reigeluth, C. M. (Ed.). (1983a). Instructional-design theories and models: An overview of their current status. Hillsdale, NJ: Lawrence Erlbaum.

Reigeluth, C. M. (1983b). Meaningfulness and

EIR&D, Vol, 39, NO. 4

instruction: Relating what is being learned to what a student knows. Instructional Science, •2(3), 197-218.

Reigeluth, C. M. (Ed.). (1987). Instructional theories in action: Lessons illustrating selected theories and models. Hillsdale, NJ: Lawrence Erlbaum.

Reigeluth, C. M., Merrill, M. D., Wilson, B. G., & Spiller, R. T. (1980). The elaboration theory of instruction: A model for structuring instruction. Instructional Science, 9, 195-219.

Reigeluth, C. M., & Stein, R. (1983). Elaboration theory. In C. M. Reigeluth (Ed.), Instructional- design theories and models: An overview of their current status (pp. 335-381). Hillsdale, NJ: Lawrence Erlbaum.

Reiser, R. A. (1987). Instructional technology: A history. In R. M. Gagn~ (Ed.), Instructional technology: Foundations. Hillsdale, NJ: Lawrence Erlbaum.

Resnick, L. B. (1981). Instructional psychology. Annual Review of Psychology, 40, 631-666.

Resnick, L. B. (1989, February). Learning in school and out. Educational Researcher, 13-20.

Riesbeck, C., & Schank, R. (1989). Inside case-based reasoning. Hillsdale, NJ: Lawrence Erlbaum.

Rogoff, B., & Lave, J. (Eds.). (1984). Everday cognition: Its development in social context (pp. 95-116). Cambridge, MA: Harvard University Press.

Salmoni, A. W., Schmit, R. A., & Walter, C. B. (1984). Knowledge of results and motor learning: A review and critical reappraisal. Psychological Bulletin, 95(3), 355-386.

Salomon, G., & Perkins, D. N. (1988). Rocky roads to transfer: Rethinking mechanisms of a neglected phenomenon. Educational Psychologist, 24(2), 113-142.

Scardamalia, M., & Bereiter, C. (1985). Fostering the development of self-regulation in children's knowledge processing. In S. E Shipman, J. W. Segal, & R. Glaser (Eds.), Thinking and learning skills: Research and open questions (pp. 563-577). Hillsdale, NJ: Lawrence Erlbaum.

Scardamafia, M., & Bereiter, C. (1991). Higher levels of agency for children in knowledge building: A challenge for the design of new knowledge media. The Journal of the Learning Sciences, 1(1), 37-68.

Schank, R. C., & Jona, M. Y. (1991). Empowering the student: New perspectives on the design of teaching systems. The Journal of the Learning Sci- ences, I(1), 7-35.

Schoenfeld (1985). Mathematical problem solving. New York: Academic Press.

Sherwood, R. D., & The Cognition and Technol-

ogy Group at Vanderbilt. (1991, April). The development and preliminary evaluation of anchored instruction environments for developing mathematical and scientific thinking. Paper presented at the meeting of the National Association for Research in Science Teaching, Lake Geneva, WI.

Smith, E L., & Wedman, J. E (1988). Read-think-aloud protocols: A new data source for formative evaluation. Performance Improvement Quarterly, 1 (2), 13-22.

Spiro, R. J., Coulson, R. L., Feltovich, E J., & Ander- son, D. K. (1988, October). Cognitive flexibility theory: Advanced knowledge acquisition in ill-structured domains (Technical Report No. 441). Champaign, IL: University of Illinois at Urbana-Champaign, Center for the Study of Reading.

Spiro, R. J., & Jehng, J-C. (1990). Cognitive flexibility and hypertext: Theory and technology for the nonlinear and multidimensional traversal of complex subject matter. In D. Nix & R. J. Spiro (Eds.), Cognition, education, and multimedia: Exploring ideas in high technology (pp. 163-205). Hillsdale, NJ: Law- rence Erlbaum.

Strike, K. A., & Posner, G. J. (in press). A revisionist theory of conceptual change. In R. Duschl & R. Hamilton (Eds.), Philosophy of science, cognitive science, an educational theory and practice. Albany, NY: SUNY Press.

Van Haneghan, J. P., Barron, L., Young, M. F., Wil- liams, S. M., Vye, N. J., & Bransford, J. D. (in press). The Jasper Series: An experiment with new ways to enhance mathematical thinking. In D. Halpern (Ed.), The development of thinking skills in the sciences and mathematics. Hillsdale, NJ: Law- rence Erlbaum.

Wertsch, J. V. (1985). Vygotsky and the social for- mation of mind. Cambridge, MA: Harvard Univer- sity Press.

White, B. Y., & Frederiksen, J. R. (1986). Progres- sions of quantitative models as a foundation for inteUi- gent learning environments (Technical Report #6277). Cambridge, MA: Bolt, Beranek, & Newman.

Wilson, B. G. (1989). Cognitive psychology and instructional design. In S. Thiagarajan (Ed.), Pro- ceedings of the National Society for Performance and Instruction, 317-333.

Wilson, B., & Cole, P. (1992, April). Returning the "'theory" to Elaboration Theory: Strategies for organizing instruction based on cognitive conceptions of learning. Paper presented at the annual meeting of the American Educational Research Associa- tion, San Francisco.

Documents

A review of cognitive teaching models