Bodong Chen

Crisscross Landscapes

Notes: Hewitt, J., & Scardamalia, M. (1998). Design principles for distributed knowledge building processes



Citekey: @Hewitt1998

Hewitt, J., & Scardamalia, M. (1998). Design principles for distributed knowledge building processes. Educational Psychology Review, 10(1), 75–96. doi:10.1023/A:1022810231840


A great review article (sort of) that summarizes design principles of knowledge-building communities in the late 90s. Interesting insights into the historical development of KB as an ed approach.


In this paper we explore various interpretations of the term “distributed cognition,” then turn our attention to communities grounded in the practice of collaborative knowledge building. We discuss CSILE (Computer-Supported Intentional Learning Environments), a technology designed to support contributions to a communal database. Shared responsibility for this community resource extends to aspects of school practice typically handled exclusively by teachers, and engagement in improving and connecting the contents of the database makes the process of knowledge building self-sustaining. (p. 1)

role of tech
role of tech is mentioned here. (p. 1)

Was this paper the first one to mention design principles? a precursor of principle-based design? (p. 1)

design principles for distributed knowledge building processes. (p. 1)

The conventional notion that cognition resides “in the head” is currently being challenged by a perspective of cognition as distributed over both individuals and their surrounds. (p. 1)

Lave (1988) for example, suggests that the relationship between human thought, human action, and the environment is so tightly interwoven that the mind cannot be studied independently of the culturally organized settings within which people function. (p. 1)

For over a decade our research has been exploring the implications of distributed cognition for schooling and educational practice. What kinds of distributions facilitate learning? And how can technologies be harnessed to support educationally effective distributed processes? (p. 2)

distributed cognition
Interesting that in this earlier paper distributed cognition is strongly emphasized, whereas self-organization and complex systems are mentioned more nowadays. Wondering how the transition happened and why? (p. 2)


But what does it mean to say that a group of people know something? And what does the word distributed mean? An investigation of the literature reveals many conflicting responses to these questions. For clarity, we examine the concept of distributed cognition from three perspectives: a situative perspective, a cognitive perspective, and a combined perspective. (p. 2)

Distributed Cognition: A Situative Interpretation (p. 2)

Proponents of the situative perspective (Lave, 1988; Lave and Wenger, 1991) claim that much or all of what is learned is tied to the specific situation in which the learning takes place. (p. 2)

The issue is not simply that in-the-head knowledge is selectively tailored to meet the particular needs of each situation, but that the affordances and constraints of a situation are an inseparable part of the cognitive process. (p. 3)

This tight binding between in-the-head representations and in-the-world activity suggests that cognition should be viewed as distributed over mind, body, and the surrounds. The term “distributed,” in this case, does not mean “divided up” in the sense that candies are distributed to children at a party. Rather it means “spread over,” much in the same way that weather systems cover a geographical area. (p. 3)

In a similar fashion, the mind, the setting, activity, artifacts, signs, symbols, social processes, and cultural factors comprise a mutually interacting, interdependent, and indivisible system of cognition (Greeno, 1997). Thus, from a sociocultural point of view, individual mental processing is better understood as a complex system involving the individual and the whole personal environment. All cognition is fundamentally situated and distributed. (p. 3)

a complex system of mental processing
‘complex system’ is mentioned here. (p. 3)

An important extension of the situative perspective is a line of thinking that focuses almost exclusively on social distributions. Grounded in Vygotskian theory, this approach suggests that knowledge exists in the way that social groups communicate, make use of symbols and tools, and organize their belief systems. Understanding is no longer a process of coming to know the entities and attributes that exist in the world, but one of successfully negotiating the meaning of these objects with others. This is achieved by taking a legitimate role in the ongoing activities of a community and gradually moving to fuller participation (Brown, Collins, and Duguid, 1989). (p. 3)

Collins et al. (1989) have begun to investigate the educational ramifications of Vygotskian theory by devising an instructional model called cognitive apprenticeship. At the heart of cognitive apprenticeship is the notion of a “zone of proximal development” (Vygotsky, 1978)—a zone of cognitive processes that are just beyond the student’s immediate capabilities. The idea is that the teacher and student collaboratively work on activities in this zone, with the teacher gradually transferring responsibility for activities to the student as competence develops. (p. 3)

cog appreticeship
It seems it will be contrasted with KB later in this paper. (p. 3)

Distributed Cognition: A Cognitive Interpretation (p. 4)

The focus is on the individual’s internal intellectual structures, with in-the-world activity serving as input to mental processes through the sensory-motor channels. This view is consistent with the situative position, that activity is an important part of learning, but in the cognitive case a strong distinction is made between “knowing” and “doing” rather than viewing the two as completely interdependent and inseparable. Some knowledge is specific to in-the-world situations, while other knowledge is contextually independent (Anderson et al., 1996). (p. 4)

From a strictly cognitive perspective, the notion of distributed cognition is ill-defined. If cognition is interpreted as an in-the-head phenomenon, then how can cognition be distributed across people or objects? (p. 4)

For instance, a child can know X (that the Earth is round) and not-X (that everyone walks on the same seemingly horizontal plane) at the same time. Similarly, a researcher may adopt the pragmatic perspective that knowledge and learning are fundamentally situated for some purposes, and in-the-head for other purposes (Cobb, 1994). (p. 4)

psychological underpinning
distributed cognition serve as a psych underpinning of KB? or at least in its early stage? (p. 5)

Discourse is dynamic; none of the participants know with absolute certainty where a particular thread will lead or what new ideas will emerge. This co-construction of a progression-of-thought is one interpretation of the phrase “distributed cognition” from a cognitive point of view. Each person’s individual cognitions are continually reorganized in an effort to construct meaning out of the other person’s speech acts. (p. 5)

The notion that there may be different kinds of distributed cognition has been suggested by Salomon (1993) who proposed that there are at least two classes. The first class consists of cognitions that are distributed through shared activity, such as the problem-solving conversation described above. The second class, called off-loading, is the one more commonly used to describe material distributions (although one also can off-load onto other humans). For example, a shopping list supports the cognitive task of remembering what groceries need to be purchased. A calculator helps with arithmetic operations. A folder placed by the door serves as a reminder to bring the folder to work. A car dashboard reduces some of the complexity of driving. And so on. (p. 5)

Artifacts mediate cognition, but they do not do cognition (Pea, 1993), and it is not clear in what sense they contain knowledge (see Anderson et al., 1995, for further discussion of this last point). (p. 6)

Distributed Cognition: A Combined Interpretation (p. 6)

The tension that exists between the cognitive and situative interpretations is rooted in a disagreement concerning the nature of cognition. The former presupposes a container model (Bereiter and Scardamalia, 1996), in which the mind is viewed as a container for cognition. Situative theory, on the other hand, proposes that cognition emerges when people, objects, and situations combine. Cognition is spread over the entire situation, and it can not be reduced to chunks of cognition that are located in the head or in an object. (p. 6)

Distributed intelligence focuses on situated, moment-by-moment interactions—what people do in their day-to-day lives. It involves planning, inventing artifacts, reorganizing existing artifacts to facilitate future personal and communal activity, and so forth (Pea, 1993). (p. 7)

Both Pea and Salomon propose theoretical frameworks that interweave individual cognitions with cognitions (or “intelligences” in Pea’s case) that are situationally-based and grounded in activity. (p. 7)


At the same time, the notion that students should take charge of the highest-level regulatory functions of the classroom—functions such as evaluation of progress in understanding, curriculum coverage, and so forth—are still considered the exclusive domain of teachers and curriculum experts. (p. 7)

regulatory functions/processes
interesting that in this earlier paper ‘regulatory functions/processes’ is used instead of ‘cognitive responsibility’. Distribution is emphasized (in addition to collective), and less used in later literature. (p. 7)

Greater distribution of regulatory processes—including those related to curriculum goals and assessment of progress—havebeen the concern of the CSILE (Computer-Supported Intentional Learning Environments) initiative. (p. 7)

The Knowledge-BuildingCommunityModel (p. 8)

In a broad sense, a Knowledge-Building Community is any group of individuals dedicated to sharing and advancing the knowledge of the collective. (p. 8)

What is defining about a Knowledge-Building Community is not formal association (e.g., “Department,” “Club,” “Company”) or physical proximity (although that is often important) but a commitment among its members to invest their resources in the collective upgrading of knowledge. (p. 8)

Extended to a classroom setting, the KnowledgeBuilding Community model is different from contemporary school practices in several respects: (p. 8)

Classroom Activity Defined by Advances in Knowledge Rather than Completion of Tasks (p. 8)

Conventional school tasks may still be involved, but these are now subordiante to engagement in a collaborative research program with the goal of advancing both individual and collective understanding. (p. 8)

Much like an academic research community, this involves talking to more knowledgeable colleagues, reading relevant resource materials, posing questions, offering theories, conducting experiments, and generally working with peers to make sense of new ideas. (p. 8)

Greater Access to Distributed Expertise (p. 8)

Within a typical classroom of 20-30 students there is a wealth of untapped knowledge and expertise. (p. 8)

The Knowledge-Building Community model attempts to solve this problem through technological supports. As networked computer systems make their way into school classrooms, they provide us with opportunities to redesign the kinds of distributions that go on there. (p. 9)

role of technology
an important point about the role of tech for KB — making distributed expertise easier to support (p. 9)

This “many-to-many” form of communication bypasses many of the logistical limitations associated with large group, face-to-face discourse, providing opportunities for distributed processes that would normally not be possible. (p. 9)

it also brings about a challenge to make sense of, regulate, and advance such a many-to-many discourse. (p. 9)

Student-Created Artifacts as Mediators of Distributed Cognition (p. 9)

This is the initiating, or designing, end of distributed cognition. The culture of schools, however, is one that de-emphasizes this design process (Pea, 1993). (p. 9)

In either case, they are not likely to be referred to again. Thus, while students go through the motions of creating new artifacts, they fail to reap the benefits of using them as cognitive tools. Rather than “teaching for the design of distributed intelligence,” as Pea (1993) recommends, schools set up artificial situations that fail to convey the personal and communal advantages of actively distributing knowledge (p. 9)

To address this deficiency, student-created artifacts in a Knowledge-Building Community are made available to the entire class. Using a collaborative online environment, each student’s ideas are stored as electronic documents in a publicly-accessible database, and can be subsequently used as a foundation for other, more advanced artifacts. In this manner, student work is valued as an intellectual resource for the entire classroom community. (p. 9)

Resnick (1991) points out that many efficient systems in nature are of a form in which none of the individual parts manages the activity of the whole. He suggests that people tend to bring a “centralized mindset” to problems of organization and task completion, and overlook the possibility of building more autonomous, self-organizing systems. (p. 10)

self-organizing systems
mentioned… (p. 10)

CSILE: Computer Supported Intentional Learning Environments (p. 10)

CSILE (Scardamalia et al., 1989) (p. 10)

Figure 1 presents a typical discussion window, with a student’s problem represented at the top of the window, and classmate responses recorded in chronological order below. (p. 10)

tech design of CSILE is quite different from KF nowadays. moving from problem-based to the problem field rarely used. What happened?
looks much like threaded discussion around a single problem.
not clear how links among different notes are drawn. (p. 10)

Database browsing occurs through a “Knowledge Map” utility which provides a graphical overview of the notes written in a database topic (Fig. 2). Rectangles on the Knowledge Map represent discussions, squares represent graphics, and circles represent text notes. Different colored lines represent different types of connectivity between notes. (p. 10)


Over time, we have identified a number of general strategies that seem particularly effective in fostering educationally-beneficial distributed practices. (p. 12)

  1. Support Educationally Effective Peer Interactions (p. 12)

One goal, therefore, is to develop a framework of social support in the classroom that optimizes the conditions necessary for educationally effective peer relations. (p. 12)

computer-supported discourse gives all students a voice, and provides time for reflection and revision before an idea is submitted for public scrutiny (Harasim, 1990). (p. 13)

While online discourse may promote equality, it is arguably less conducive to maintaining an optimal level of conflict. Sometimes it produces too little. (p. 13)

“Grounding a conversation” (Clark and Schaefer, 1989; Clark and Wilkes-Gibbs, 1986) is a difficult task across media that lack copresence, visibility, audibility,and simultaneity(Clark and Brennan, 1991). Users of such environments need to invest more effort in formulating their contributions than they would otherwise. (p. 13)

teacher’s role?
grounding educationally meaningful peer interactions? (p. 13)

CSILE teachers often discuss collaboration strategies with their class. They encourage students to read and comment upon other people’s work. And they illustrate effective and ineffective ways of disagreeing and criticizing. For instance, teachers might write sample sentence starters on the blackboard (e.g., “I really like your idea, but I wonder if …”). Students copy these down in their notebooks, and then refer to them during their CSILE sessions. The goal, of course, is not to avoid student disagreement or argumentation but to maintain a balance between too much and too little conflict. (p. 13)

  1. Integrate Different Forms of Discourse (p. 13)

CSILE teachers try to be sensitive to the different strengths of faceto-face conversation and online discourse, and look for effective ways of combining the two in the classroom. (p. 13)

teacher’s role (p. 13)

or instance, they may encourage groups of students to gather together around a common computer and coconstruct notes. Or, they orchestrate classroom activities so that in-person conversations are more likely to make their way into the database, (p. 13)

Therefore, their conversations “get” into the database … ; otherwise it would be lost/shared in a sense to just the few rather than the community of learners. (Cedar Rapids Iowa, U.S.) (p. 14)

also role of technology (p. 14)

  1. Focus Students on Communal Problems of Understanding (p. 14)

The need to come to a shared understanding places greater demands on students to clarify ideas, refine theories, answer each other’s questions, and negotiate meaning with one another. One of the ways that the CSILE interface provides affordances for problem-centered discourse is through its discussion facility (Figs. 1, 3). Any class member may start a discussion by recording a personal problem of understanding in the “problem area” of the note. (p. 14)

embodied design
principle embodied in tech design. in this case, problem-based learning. (p. 14)

The discussion environment also contains a special branching feature that allows students to create subdiscussions as they develop more specific queries (Fig. 4). Miyake (1986) noticed in a study of people trying to understand the functioning of a sewing machine that participants learned in an iterative fashion. (p. 15)

Questions inspire new explanations, and explanations, in turn, inspire more questions. The resultant cascade of discussions is visible to students on the Knowledge Map as a web of interrelated notes. (p. 15)

some teachers report that it is sometimes difficult to keep students focused on making advances on their (p. 15)

communal problems. Many students find it easier to study broad topic areas than to find answers to specific problems. As one CSILE teacher explains: (p. 16)

I do spend quite a bit of time with individualsasking them what problem they’re working on when they’re doing their research. The answer often is, “I’m working on energy” (for example), so I take that opportunity to redirect them to a problem which they are trying to solve. I explain to them how difficult it is to find information related to a specific problem and how they will have to consult many sources before they are likely to be successful. It’s another attempt to move them away from the model of source material determining the direction of research, rather than the problem determining the direction of research. (Teacher, Toronto Ontario) (p. 16)

role of teacher
focusing students on problems. (p. 16)

Thus, the teaching task is not simply one of supporting distributed processes, but is also a matter of focusing distributions in the direction of sustained collaborative inquiry by relating individual activity to pressing communal problems. (p. 16)

  1. Promote Awareness of Participants’ Contributions (p. 16)

This is an attitude that CSILE teachers continually try to change by highlighting interesting student work and encouraging database exploration. For students to take advantage of their classmates’ knowledge productions they have to know that those resources exist. (p. 17)

teacher’s role (p. 17)

Several facilities have been constructed to promote awareness and exploration of CSILE notes. (p. 17)

  1. Encourage Students to Build on Each Other’s Work (p. 18)

teacher’s role (p. 18)

CSILE teachers place a great emphasis on making connections with other students. (p. 18)

Teachers, for their part, encourage students to take up each other’s ideas, conjectures, and queries, and extend them through online discourse. Their goal is not simply to encourage collaboration and the efficient distribution of resources, but also for students to view their ideas as useful communal artifacts. This is consistent with the notion that students come to appreciate the personal and social advantages of designing distributed intelligence into their environment (Pea, 1993). (p. 18)

  1. Emphasize the Work of the Community (p. 18)

more successful CSILE teachers have adopted the strategy of focusing students on advancing the class’s knowledge. The reason for this is twofold. First, by placing the individual’s work in the context of a larger group mission, the teacher illustrates that the learner’s own writings have value that extends beyond assessment. Second, by trying to make advances on what the group knows collectively, the student must first learn what the group already knows—and that process leads to a deeper understanding of the content and deeper appreciation of peer contributions. (p. 18)


Our objective in identifying design principles is to clarify means for engaging students in processes of knowledge advancement. This requires a significant reconceptualization of classroom operations. It is not enough to carry out just one or two of the above approaches—rather, it requires the coming together of many new practices. Such dramatic changes can be difficult to make. (p. 19)

implementation of principles
interesting comment that this is hard to achieve at once. (p. 19)

Most novice CSILE users have entrenched preconceptions that interfere with the goal of establishing a classroom-based Knowledge-Building Community. The challenge, therefore, is not simply to provide opportunities for distributed processes, because students may not even recognize them as such. Instead, the challenge is to design situations and tools that have a deliberate bias toward shared activity and rewarding interactions with other knowledge builders. (p. 19)

community of practice
interesting that it has been used in earlier KB literature to describe KB communities. (p. 19)

CSILE research is an ongoing process of finding new ways to enculturate students into a knowledge-centered community of practice. (p. 19)

Thus, the Knowledge-Building Community idea is being tested both in the classroom and in the online community of CSILE teachers. In both cases, the goal is to make the drive for improvement become self-sustaining, (p. 20)

To foster a classroom-based Knowledge-Building Community, we suggest that a careful interweaving of computer supports and new educational (p. 20)

practices can bring about a more effective use of distributed resources in the classroom. (p. 21)