Chapter 3.5
Computer Contexts for Supporting Metacognitive Learning
Xiaodong Lin
Teachers College, Columbia University
Florence R. Sullivan
University of Massachusetts, Amherst
Abstract: A major challenge for both educational researchers and practitioners is to understand why some people seem to learn more effectively than others and to design tools that can help less successful people improve their abilities to learn. In this chapter, we describe the most frequently documented metacognitive learning outcomes including: recall/memory; content learning/problem solving; and social interactions as knowledge acquisition. We then use each of these metacognitive learning outcomes to examine how today’s computer tools have or have not reached their fullest potential to support these learning outcomes and we suggest ways that computers tools can be designed to achieve these outcomes.
Key words: metacognition; metacognitive learning; metamemory; content knowledge; problem solving; social interaction; adaptive expertise
3.5.1 Common Metacognitive Learning Outcomes
Some 30 years ago, Brown and Flavell introduced the concept of “metacognition” to the educational research community (Brown, 1975; Flavell, 1976). Metacognition is defined as an awareness of one’s own thinking processes and the ability to control, monitor and self-regulate one’s own learning behaviors so effective problem solving and deep understanding can be reached. In 1983, Brown, Bransford, Ferrara and Campione did a comprehensive summary and analysis of metacognitive research. They concluded the analysis by suggesting that a variety of learning outcomes can be produced when people are engaged in metacognitive experiences. For instance, people who are aware of the limitations of their own memory and deliberately use rehearsal strategies recall more than those who are not aware of their own limitations (Wellman, 1977). In terms of content learning and problem solving, the research shows that people are able to apply what they learn in new situations if they are involved in intentional instruction where they understand how, why, when and where the new information and strategies are useful (Brown et al., 1983).
A third learning outcome, that has not been given enough attention, is the relationship between social interactions and metacognition. This is particularly important in terms of classroom teaching. Researchers have found that teachers interact with students with good and poor reading skills quite differently. Good readers are questioned about the meaning behind what they are reading, asked to evaluate and criticize materials, and so on. By contrast, poor readers primarily receive drills (McDermott, 1978). What kinds of metacognitive understanding get developed from these different kinds of social interactions for both students and teachers? This is an interesting question to explore.
In this chapter, we discuss how different types of metacognitive learning outcomes can be developed from different situations and how different situations require different metacognitive skills. We focus on the following learning outcomes: (1) simple recall and memorization of facts; (2) more complex learning outcomes, such as problem solving; (3) domain subject learning; and (4) social knowledge. We then examine how today’s computer tools have or have not reached their fullest potential to support these learning outcomes and we suggest ways that computers tools can be designed to achieve these outcomes.
3.5.2 Recall and Memory What Research Says
Among the learning outcomes, recall seems to get the most attention for a variety of reasons. The first is that the ability to recall or memorize is sensitive to developmental and learning material changes. Older children remember better than younger ones and typical children recall better than children with developmental delays. The research also shows that when the materials are familiar and the items are distinct, age differences are minimal (Myers, Clifton & Clarkson, 1987). The second reason that recall receives a great deal of attention is that it is one of the most frequently used assessment measures by teachers, school systems, and national testing agencies.
Metamemoryrefers to learner awareness about his or her own memory systems and memory strategies. Research indicates that young students and novice learners have difficulty accurately estimating their comprehension and that metamemory strategy instruction should focus on specific strategic knowledge. Metamemory can be divided into two types: explicit and conscious knowledge and implicit and unconscious knowledge (Brown et al., 1983). An example of explicit metacognitive knowledge, that even preschoolers are consciously aware of, is that it is easier to remember a simple and short word than a long and complex word. Such self-monitoring enables people to generate a feeling of knowing that can help them predict how well they will remember later on. However, often, metacognitive knowledge is unconscious. For instance, good readers slow down their reading when the texts become difficult without realizing they are doing so (Siegler & Alibali, 2005).
Research on the relationship between memory and metacognition has been motivated by the assumption that children’s increasing knowledge about their own memory and about the strategies they use to facilitate memorization can help them choose more effective strategies for memory. Whether or not metacognition facilitates memory is a somewhat tricky question. On the one hand, research shows that young or learning disabled children tend not to use rehearsal or other strategies to facilitate their memory because they may not know that their memory capacity is limited (Brown et al., 1983). But once they are trained to use effective strategies, they greatly improve their memory performance. If older students are prevented from using effective memory strategies, they produce levels and patterns of performance that are very similar to younger children or children with learning disabilities. In addition, knowing the relative usefulness of strategies could improve children’s strategy choices in a wide range of situations (Brown, et al., 1983; Siegler & Alibali, 2005). This is one of the most robust findings in the developmental literature (Belmont & Butterfield, 1971; Brown, 1975; Kail & Hagen, 1977).
However, metacognition alone may not improve memory – other ingredients need to be in place. These ingredients include developmental capabilities (the ability to associate and recognize things), use of broadly applicable memory strategies (such as rehearsal, organization, and selective attention), and knowledge about the specific content (Siegler & Alibali, 2005). Metacognition can considerably assist memory performance only when each of the ingredients is present (Siegler & Alibali 2005).
3.5.3 Ways to Improve Memory Performance
There are several ways to help learners become effective in memory and recall tasks. One way is simply to rehearse the facts until they are remembered. This approach usually does not lead to understanding, especially when a task requires application of the facts learned (Brown et al., 1983). More effective ways are to employ different kinds of metacognitive and planful memory strategies, such as elaboration, identifying main ideas and categorization strategies (Brown et al., 1983). The most frequently cited research on metamemory regard interactions between understanding and strategies, and learning facts as they are applied in varied-contexts.
Many researchers argue that the application of elaboration, categorization, and generation strategies are important for comprehension and thus memory performance (Anderson & Reder, 1979; Bransford, et al., 1982; Brown et al., 1983). However, the degree to which any of these strategies are successful in improving memory is influenced by the availability of relevant content knowledge (Chi, 1978). Nitsch (1977) showed when students study the same concept in varying contexts; they are better able to understand the concept in new situations. Research by Hatano & Inagaki (1986) also shows that experiencing varied contexts is important to the development of adaptive expertise. Adaptive expertise is characterized as procedural fluency complemented by explicit conceptual and principle understanding that allows people to adapt what they learn to widely varied situations.
3.5.4.Computers as Metacognitive Tools to Enhance Memory
A program developed by Bransford and his colleagues (Cognition and Technology Group at Vanderbilt, 2000) - the Knock Knock™ game, offers a promising example of using computers as metacognitive tools to enhance literacy and memory. Knock Knock™ helps children become aware of constraints on their own learning that they need to address in order to be successful with the game. For example, to achieve the best results children have to use broadly applicable memory strategies, such as rehearsal, organization, generation, categorization, and selective attention strategies. They also need to generate simple stories based on the letters they hear or read. The children will also develop knowledge about the specific content that they are learning - letters, sounds, and story writing. To facilitate metacognitive development, children are asked to estimate how well they will apply the letters to a variety of different situations and discuss their applications with peers. The discussions among peers and with teachers also offer students social support and help students recognize the usefulness of the strategies in helping them perform the memorization and application tasks. Knock Knock™ illustrates an approach of using computers to support recall and learning that should help students develop skills that are important for future success.
3.5.5.Content and Domain Subject Learning: What Research Says
In this section, we examine issues concerning the importance of acquiring content knowledge of any given discipline from the perspective of adaptive expertise development. Hatano and his colleagues introduced the concept of adaptive expertise in relation to masters in using the abacus. They proposed that abacus masters should be termed routine experts if they have only developed procedural knowledge and skills about the abacus they learned. Whereas, adaptive experts understand the principles and concepts underlying the content and skills learned. He and his colleagues contrasted routine experts with adaptive experts, and asked the educationally relevant question of how “novices become adaptive experts – performing procedural skills efficiently, but also understanding the meaning and nature of their object.” (Hatano & Inagaki, 1986, pp. 262-263). Procedural knowledge is often only useful for limited types of problems and situations. Comprehending principles underlying problems and content learned enables people to flexibly apply this knowledge to various new situations (Hatano & Inagaki, 1986).
As such, adaptive experts usually verbalize the principles underlying one’s skills, judge conventional and non-conventional versions of skills as appropriate, and modify or invent skills according to local constraints. Wineburg (1998) and others (e.g., Bransford & Schwartz, 1999) have added to this list by pointing out that adaptive experts are also more prepared to learn from new situations and avoid the over-application of previously efficient schema (Hatano & Oura, 2003).
A second perspective Hatano and Inagaki suggested is that in stable environments, participation in one’s culture typically provides sufficient resources for learning and executing routine expertise. People have many pockets of routine expertise where they are highly efficient without a deep understanding of why. To develop adaptive expertise, people need to experience a sufficient degree of situational variability to support the possibility of adaptation. This variation can occur naturally, or people can actively experiment with their environments to produce the necessary variability. Hatano and Inagaki (1986) proposed three factors that influence whether people will engage in active experimentation.
One factor is whether a situation has “built-in” randomness or whether technology
has reduced the variability to the point where there is little possibility for exploration. Much software we reviewed often eliminates situational variability to help students focus on the procedural skill. This is particularly true of software aimed at helping students develop literacy and numeracy. For example, many math programs, such as Math BlasterTM ( present students with a storyline or game-like interface, but these conceits are meant as a means of motivating students only, and in fact, math learning is presented in a drill and skill format, wholly divorced from any meaningful context in which math may be learned. Likewise, math-tutoring programs, such as Wayang Outpost ( ) (Beal & Lee, 2005), while providing a motivating storyline and individualized and helpful feedback to students on the procedure of solving a problem, do not provide varied situations in which the math skills may be needed. This may have the unintendedconsequence of preventing students from developing variations in that procedure in response to new situations.
The second factor involves the degree to which people are enabled to take risks in approaching a task. When the risk attached to the performance of a procedure is minimal, people are more inclined to experiment. “In contrast, when a procedural skill is performed primarily to obtain rewards, people are reluctant to risk varying the skills, since they believe safety lies in relying on the ‘conventional’ version” (Hatano & Inagaki, 1986, p. 269). Game-like software that provides rewards for successful performance of the procedure or skill will limit risk-taking, thereby limiting students’ ability to adapt their understanding to new situations.
The third factor involves the degree to which the classroom culture emphasizes either
understanding or prompt performance. Hatano & Inagaki (1986) state, “A culture, where understanding the system is the goal, encourages individuals in it to engage in active experimentation. That is, they are invited to try new versions of the procedural skill, even at the cost of efficiency” (p. 270). They proposed that an understanding-oriented classroom culture naturally fosters explanation and elaboration, compared to a performance-oriented classroom culture. Their views also echo the research findings by Bereiter & Scardamalia on the importance of engaging students in a knowledge and understanding-oriented society and their impact on adaptation and human development (Bereiter & Scardamalia, 2000; Scardamalia & Bereiter, 1996). Central to these concerns is people’s ability to self-monitor their own understanding at a deep principle level.
3.5.6 Ways Metacognition can Improve Content Learning and Adaptive Expertise
Neither metacognitive monitoring skills nor content learning alone will do the job of improving people’s deep understanding of the subject matter leading to adaptive expertise in a specific domain. Rather, the two work in concert with one another in the following ways. First, utilizing familiar content knowledge improves the effectiveness of using different metacognitive strategies. Second, familiar content facilitates learning of new strategies such as elaboration (Bransford et al., 1982). Familiar content may also serve as “a kind of practice field upon which children exercise emerging memory strategies” (Siegler & Alibali, 2005, p. 262). Third, content knowledge facilitates people’s metacognitive development by offering specific data and a context in which to monitor and revise their strategies and procedures. Research shows that metacognition works best when an individual has specific issues to work through (e.g., Chi, DeLeeuw, Chiu, & LaVancher, 1994; Lin & Schwartz, 2003). This is because people think best when they have a known specific context to work with (Gay & Cole, 1967). Indeed metacognitive monitoring is often retrospective, capitalizing on a specific past as opposed to a vague future.
Ample research shows that effective metacognitive interventions can improve people’s understanding of deep principles that underlie content and problems in a given domain. The majority of metacognitive interventions involve either a strategy-training approach, or a contextualizing knowledge and tools approach aimed at supporting students metacognitive monitoring and revision of understanding. In recent years, researchers have also started to recognize the importance of creating social interactions to support metacognition. Each of these approaches will be discussed below.
Metacognitive strategy training. The main purpose of strategy training research is to explore: (a) how specific sets of metacognitive strategies help people monitor conflicting thoughts and build a coherent understanding of a subject domain; (b) how specific metacognitive strategies will help people develop deep principles about the concepts learned; and (c) how different types of instructional supports for metacognitive strategies influence students’ engagement in metacognitive activities. Metacognitive strategy training is usually used during the acquisition of either domain-specific or self-as-learner knowledge. Students usually stop at fixed intervals while learning specific subject domains to reflect on and revise their work. The interventions usually do not involve changing the existing school curriculum and classroom culture. The most effective approach to strategy training seems to be prompting students to self-explain or self-question as a way to engage in metacognitive thinking and modeling through social interactions.
The act of explanation helps students become aware of the strategies they are using and the content they are learning. For instance, Siegler and Jenkins' (1989) found children who were aware of using a new strategy subsequently generalized it more to other problems. However, research also indicates that students often fail to check and monitor whether or not they understand the content knowledge they are learning if they are not explicitly trained to do so (Brown et al., 1983). Chi et al., (1994) found that prompting self-monitoring in students leads to such awareness and stronger learning outcomes. Moreover, the prompted students who generated a large number of self-explanations (the high explainers) learned with greater understanding than the low explainers. Chi et al., (1994) reported that such monitoring through self-explanation helped students recognize principles underlying the content and procedures learned, not just the procedures. This provides an important basis for the development of adaptive expertise (Hatano & Inagaki, 1986).
Researchers have also used video technologies to model effective strategy applications. For instance, Bielaczyc and her colleagues used video to model effective learning strategies employed by good problem solvers in the domain of LISP programming (Bielaczyc, Pirolli & Brown, 1995). Students were exposed to specific metacognitive strategies and received explicit training in their use. They found that mere exposure to good learning models was not sufficient. The key to the success in their design was to have students experience these strategies in their own learning, explicitly compare their own performance with that of the model, and take actions to revise ineffective learning approaches.