Support for Cognitive Apprenticeship in ObjectOriented Model Construction Jakob Tholander and Klas Karlgren Department of Computer and Systems Sciences Stockholm University/Royal Institute of Technology Electrum 230, 164 40 Kista, Stockholm, Sweden + 46 8 16 20 00
[email protected],
[email protected] ABSTRACT We discuss a design solution for a learning environment for students of object-oriented modeling. The learning environment consists of three tools and is explicitly based on work on situated learning and work on cognitive apprenticeship. An overall goal in this work has been to ensure a design that is well grounded both theoretically and empirically. The design of the tools relate to central issues in the theories such as: authenticity in problem-solving, observation of experienced practitioners, the language use of experienced practitioners, and self-reflection on one’s own problem-solving. Theoretical and practical design considerations are presented for each of the three tools. Two of the three tools directly focus on authentic activity and are therefore based on studies of how experienced practitioners solve problems. The third tool is designed to support aspects of reflection and meta-cognition. We also present a short example of students using the learning environment. Keywords: Design of learning environments, theories of learning, situated learning, cognitive apprenticeship.
INTRODUCTION Theories of learning are important for the design of computer based learning environments. However, theories cannot simply be applied and used in the process of design. Theories of learning need to be discussed and adapted to particular circumstances of computer supported learning in general and to specific learning domains in particular (Jonassen & Land, 2000). The focus of this paper is on our approach for incorporating cognitive apprenticeship and situated action in the design of a learning environment for computer science students of object oriented (OO) modeling. In order to illustrate how the environment was used we will also present examples from a study that we have conducted with this environment. Detailed presentations from this study is presented in (Tholander 2001, Karlgren 2001). The learning environment consists of three tools aiming to support cognitive apprenticeship learning (Collins, Brown, & Newman, 1989). Two of the tools support aspects of authentic activity, i.e., an activity that help “students to foresee their participation in activities that matter beyond school” (Greeno, 1997, p.11), and are therefore based on studies of how experienced OO-modelers solve problems. The third tool support aspects of reflection and metacognition and uses analysis of situated action to achieve this. The three learning tools were integrated in the Rational Rose modeling environment:
• • •
Expert problem solving tracks are based on how experienced modelers solved similar tasks. (Picture 2) A library of modeling pattern to support students’ practical use of abstractions. (Picture 1) An assistant to encourage reflection and meta cognitive thinking. (Picture 3)
Picture 1: The Pattern library
Picture 2: Expert track
Picture 3: The assistant
DESIGN GROUNDED IN THEORIES OF LEARNING Theories that emphasize the situated properties of human action and learning have been very influential on current understandings of these phenomena. Work on situated action by (Suchman, 1987), and on situated learning and cognitive apprenticeship by (Lave & Wenger, 1991), (Wenger, 1998) and (Collins et al., 1989) have been extensively used in analyses of learning and interaction (Jordan & Henderson, 1995). However from the perspective of a designer of learning environments we believe that there is still much to gain by further considering the consequences of these theories in design situations. Since these perspectives shift the focus of how learning and understanding should be understood, the designer of learning support is put in a new situation. The focus is shifted from considering “processes of individual cognition as basic” to considering processes of “interactions people have with each other and with the material and representational resources in their environment” as basic (Greeno, 1997). The current work explores this new focus and the conditions and possibilities it provides for designers of learning environments. According to (Suchman, 1987) the designer of interactive artifacts faces a problem which stem from an asymmetry of access to contextual resources between humans and machines: “Because of the asymmetry of user and machine, interface design is less a project of simulating human communication than of engineering alternatives to interaction’s situated properties.” (Suchman, 1987, p. 185) We believe that designers need to further explore what these alternatives might look like. Design approach An overall goal in this work has been to ensure a design that is well grounded both theoretically and empirically (Land & Hanaffin, 2000). This has led to two important design activities. The first was to explicitly state what aspects of the theoretical framework that we wanted to support in our design. The second activity came as a consequence of our theoretical framework which emphasizes learning in so-called authentic activities (Brown, Collins., & Duguid, 1989). This led us to conduct studies on experienced conceptual modelers in order to get an understanding of how they acted when solving problems. The results of these studies were used to design particular aspects of the learning environment and later served as a basis in evaluations of how learners interacted with the learning environment (Tholander, Karlgren, Rutz, & Ramberg, 1999), (Tholander, 2001). A similar approach was used by (Pea 1994) in the design of the Dynagram environment when from looking at an existing practice, implement a learning environment for it, and then evaluating and studying learner’s use of the learning environment.
DESIGN AND LEARNING FOCI The design of the learning environment is based on some of the central factors in cognitive apprenticeship. Collins, Brown, & Newman provide a framework for issues that designers of learning environments should consider (Collins et al., 1989). The four main dimensions of this framework are content, methods, sequence, and sociology. In our design we have identified the following aspects to be particularly important to consider in order to promote students to get engaged in the cognitive practices of conceptual modeling: 1. Learners should engage in authentic problem solving to develop skills that help them put knowledge into use. Knowledge of concepts and methods must not be learnt as abstract notions, but in a context where the practice of their use is uncovered. The term authenticity has been discussed in relation to the distinction between inschool and out of school reasoning (Brown et al., 1989; Resnick, 1987). Here, we refer to an authentic activity as an activity that help students to see the relation of their learning activities to the activities of a professional context (Greeno 1997), (Barab, Squire, & Dueber, 2000). 2. Observation of experienced practitioners’ problem solving help student to develop their own problem solving strategies. It is emphasized that a central aspect of expertise is the use of meta-cognitive strategies to guide problem solving. Such strategies help to organize the execution of tasks, relate to new knowledge and feedback from mentors and teachers, and support further investigation into a knowledge domain. 3. Learners should practice to use experienced modelers’ language, concepts, and tools since this will help learners to see the role of these concepts in practice. Ramberg and Karlgren argue that the transition from novice to expert may in part be the transition from knowing the abstract explanations to being involved in the language games of the area, i.e., being able to use the language of the domain (Ramberg & Karlgren, 1997). 4. Learners should reflect on their own problem solving and understanding, and on their use of tools and concepts in relation to how experienced practitioners use these. Two important aspects of cognitive apprenticeship that we have considered less relevant for our purposes have been left out; fading and sequencing. First, we have left out fading since this is something that we believe does not have to be explicitly supported in open-ended learning environments since learners often choose to pay less attention to
the learning support as their skills increase. Second, we have not explicitly supported a particular sequencing of the learning material since the learning environment is part of an undergraduate course where such aspects are taken under consideration. The students are first introduced to the subject through traditional lectures and literature, and later to the tools, so the users of the tools are not complete novices in the area.
LEARNING OBJECT ORIENTED MODELING The purpose of object-oriented modeling is to create models of an enterprise that can be used for the subsequent design and construction of supporting communication and information systems. Such models are usually drawn graphically and consist of classes, relations, and attributes that represent information in an enterprise. OO-modeling is open-ended in nature with multiple solutions which all could be appropriate depending on the perspective taken. We found cognitive apprenticeship particularly suitable to support learners of OO-modeling because of the focus of non-explicit skills in practical action. This was confirmed when we in the design process studied experienced modelers and the character of their skills. We have found that a lot of the aspects of how experienced OO-modelers act are left out in classroom situations, i.e., the way things are carried out in classrooms is very different from how they are carried out in professional contexts. Theorists of situated learning have taken this division as a starting point for their analysis (Greeno, 1997), (Resnick, 1987). Therefore we argue that in our case it is powerful to use a theory in the design process where these aspects are emphasized.
STUDYING EXPERIENCED MODELERS AS A BASIS FOR DESIGN The first three aspects of our design focus above (authentic problem-solving, observing experienced practitioners, tool and language use) all include aspects of how experienced modelers go about in their problem solving. To be able to design tools that really support these aspects it is essential to ground the design in an understanding of how experienced modelers reason and carry out tasks. Therefore a think-aloud study with experienced modelers was conducted. The goal of the study was to find out characteristics about the different ways modelers solve problems in order to understand important elements of the cognitive practice they work in. Three modelers were video-recorded when solving a modeling problem on a white board. The modelers could interact with a person acting as stakeholder and domain expert. The sessions took 45-60 minutes each to complete. The modeling problem they were given described a rather complex scenario with ambiguities and incomplete statements. This was in order to make sure that the modelers would have a challenging problem to engage, and not only use pre-packed solutions. The most important findings found for the design are summarized below (for a closer presentation of the results see (Karlgren, Tholander, Dahlqvist, & Ramberg, 1998)). • In the problems they face, they tend to identify familiarities with other problems which they have experienced and use these to solve the current problem, i.e., they engage in case based reasoning. • They often go back and reflect on the overall nature and goal of the task, i.e., they show a high degree of metacognitive thinking. • They discuss with stakeholders how important concepts in the problem domain should be understood. They do not presuppose certain interpretations of the concepts based on their own ideas, i.e., they are aware of that in object oriented modeling everyday concepts are given local specific meanings. • They refer to general problems and solutions that they often face in their everyday practice, i.e. they use analysis patterns. In the object-oriented community within system development this notion of analysis patterns has become important. This is further elaborated below (Fowler, 1997) • Because the meaning of concepts depend upon each other they proceed iteratively by solving sub-problems that they move back and forth between.
DESIGNING FOR OBJECT-ORIENTED MODELLING PRACTICE We have designed three tools to support students of object oriented modeling practice. The expert problem solving tracks are designed to support authentic problem solving (the first learning focus), observation (the second learning focus), and use of experienced practitioners language, concepts, and tools (the third learning focus). The library of modeling patterns is designed to support authentic problem solving and also use of experienced practitioners language, concepts, and tools. The assistant is designed to support students to reflect (the fourth learning focus) upon their problem solving in order to develop their meta-cognitive skills (Collins, 1991). The tools are integrated with the Rational Rose modeling tool and the students use the modeling and the learning tools to solve modeling cases as project work.
Expert Based Problem Solving Tracks The purpose of the tool is to present modeling tasks that provide the same kind of problems and complexity which conceptual modelers face in the 'real world'. Through this tool we want the students to get exposed to scenarios that resemble authentic environments including how experienced modelers go about to solve complex problems and how they talk about them. Emphasis is on the authentic practices of experienced modelers and on the language used by the modelers. As a starting point for the design of the tracks, as mentioned above, experienced modelers were video-recorded when solving a complex conceptual modeling task. The video-recordings were then used to create authentic scenarios emphasizing use of the language of experienced modelers. In order to achieve a high degree of authenticity we used the recordings in two ways. First, to create tracks illustrating how experts solved the current problem. The tracks consist of 'steps', each corresponding to different actions taken, or decisions made, by the experts during problem solving. Each step shows what part of the task the modeler was working on, how the modeler modified the model, and the comments made about this phase. Since the tracks are designed from the activities of different modelers they can provide multiple perspectives on the same problem, as discussed by (Collins et al., 1989),(Spiro, Feltovich, Jacobson, & Coulson, 1992). In the tracks, students can follow the activities of one modeler or jump between the solutions of different modelers. Second, we used the recordings to extract critical points in the problem solving process. We saw that the experienced modelers ran into various difficulties or stages in their modeling which demanded making decisions and choices about how to proceed, e.g., that some generalization should be made or that some object should be split into several. The tracks were therefore developed into two different levels; a Problem level and an Authentic level. On the Authentic level the user can follow exactly how a modeler constructed a model step by step and also see exactly what the modeler said at each stage. The Problem level consists of the difficulties and problems that the experienced modelers ran into during their modeling without showing any solutions. On the Problem level users are introduced into the problem and are then prompted to choose between alternative ways of proceeding. One or two of these are alternatives that were actually taken by the modelers and thereby take the user back to the Authentic level. The other alternative ways of proceeding are measures that were decided against by the modelers. If these are chosen, the user can take part of the modelers’ comments to these alternatives and follow links to alternatives preferred by the modelers. It is thus not obvious how to go from the problems on the Problem level to the Authentic level. The aim is however not just to present such problems for a passive learner, the learner is, on the contrary, encouraged to actively reflect on the problem and come up with a solution before taking part of the experienced modelers' solutions. On each step on the Authentic level users can choose to deal with the next problem on their own, i.e., move on to the next step on the Problem level or choose to follow how a modeler solved the current problem or went on to solve the next one. The Pattern Library The purpose of the pattern library is to support the students to use the language, concepts, and tools that experienced modelers use. Our study on experienced modelers showed that these aspects are often referred to through analysis patterns. An analysis pattern is a prototypical OO-model, which can be applied in different situations with appropriate modifications. For instance, patterns like product, ordering, booking, and contract may be reused in many order management systems (Fowler, 1997). We view analysis patterns as representations of the language, concepts and tools, which the students should practice to use, not as special constructs that they should memorize. The pattern library provides support to learners to practice use of, and articulation of models in, the particular language with the concepts used by the practitioners of the field - the language game (Ramberg & Karlgren, 1998). (Morrison & Collins, 1995) uses the notions of epistemic games and epistemic fluency to describe how learning is a process of developing an ”ability to recognize and practice a culture’s epistemic games, with their associated forms, goals, rules and strategies”. (ibid., p43). Similarly, (Pea, 1992) uses the term ’learning discourse’ to denote the importance in learning to practice the use of the language, concepts, and tools that are central to the domain of expertise one is being trained for. Creating Abstractions through Analysis Patterns One of the goals of the project was that students should learn to create models at a level of abstraction that makes them reusable in future situations. Our way of supporting learners in this is through analysis patterns, which are abstractions of common knowledge in object oriented modeling.
The role of abstractions in learning is widely debated, and a common misinterpretation of the situated learning arguments is that they claim that abstractions is of little use in learning situations (Anderson, Reder, & Simon, 1996), (Greeno, 1997). However, abstract knowledge is not less situated, since abstraction in itself is a social activity, with specific goals and customs, and with certain expectations about success of the abstraction. Therefore, the learning of abstraction must be meaningfully connected to the situations where such abstractions are used (Greeno, 1997). Consequently, our goal here is to support students to make abstractions, not to learn specific abstractions. Noss & Hoyles use the term “situated abstractions” to describe how mathematical abstractions become meaningful in a particular situation of use (Noss & Hoyles, 1996). They characterize situated abstractions as ”a process which develops in activity, which - like all activity – is situated. Like plans and actions, abstraction is not necessary a guide for and determinant of mathematical activity, it is a resource for activity” (Noss & Hoyles, 1996, p. 124). Two issues are central to their description of situated abstractions, activity and resource. Based on this, it was central to the design of the pattern library to make it a resource that students can utilize in their ongoing problem solving activities. The abstractions that students work with must become an integrated part of the other activities that are part of their work. To encourage such active use of analysis patterns, it is integrated with the modeling tool so that patterns can be inserted for use and modification in the model currently worked with. The patterns are also accompanied by examples and descriptions of different ways of using them. An Assistant to Support Reflection The third tool, the assistant, provides comments with the purpose to encourage the students to reflect and think critically. Students should not take their preconceptions about their solutions, their knowledge, and the problem domain they are modeling for granted. As the students construct their models the assistant asks questions and gives critical remarks about why the students have created some particular objects or relations, comments about the way the student approaches the problem, or gives general advice about good ways of approaching such a problem. Analyzing and understanding somebody else’s design and providing advice and critique on problematic points is a mostly qualified task requiring a lot of experience and skill. At first sight, it might seem improbable that an automated tool could be able to criticize such models, due to the open ended nature of OO-modeling,. Such a tool would require a deep analysis of the user and the task carried out. However, in a learning situation like the current one, our goal is not that the assistant should come up with critique of the solutions that are valid from the point of view of independent experts. The purpose is rather to give comments to students which they experience as meaningful and which encourage them to think critically about their own work. The design is therefore based on analysis of interaction between humans and machines and the consequences thereof for the design of interactive artifacts (Suchman, 1987). We have focused on one of her main arguments; that the intelligibility of an interaction is not a property of each of the participants in the interaction but rather an effect that arises in the system as a whole out of the social and material relations (Suchman, 2000). In designing our assistant we closely considered ways of achieving the purposes of the assistant in alternative ways. We have tried to design the comments of the assistant so that they relate to the activities that the students are engaged in at particular points in the modeling process. Similar to the way the instructions of the Xerox machine expert system studied by Suchman were “read by the machine’s users as comments on the activity underway that should be intelligible” (Suchman, 2000, p. 6, original italics), we give comments that are designed to be meaningful from the point of view of the learner. This can help students to think about their models and provide explanations of them. Comments for Reflection We have created three types of comments. First, comments about some particular objects and relations in the model being constructed. For instance: “Why have you connected the object 'Plane’ to the object 'Airport' with the relation 'lands-on'?" Second, comments about important issues in the enterprise that must be represented in the model. For instance: “Are you aware of that you have to be able to represent that flights might be redirected to another destination than the scheduled one?” Third, comments about some general, problematic issues from modeling theory that are also found in the model. For instance, “You have a many-to-many relationship between the classes Plane and Flight. Are you aware that such relationships might cause redundancy problems when the model is implemented in a database?”
Triggering the comments Since OO-modeling is very much about formalizing an understanding of an enterprise, a lot of information about a user’s view of the enterprise can be extracted from details of a model such as names given to classes and relations. By identifying the user’s current area of activity, e.g., flight scheduling, flight routing, or passenger information, comments relevant to this part of the enterprise can be constructed dynamically, such as "Why did you introduce the object flight route?" or "Are you able to represent that flights might be overbooked, in your model?". In systems that critique or comment upon a user’s activities the timing of the comments have shown to be of central importance (Fischer, Nakakoji, Ostwald, Stahl, & Sumner, 1993). In order to provide comments on the subtopic a student is working on at a current moment, the system identifies the classes, relations, and attributes the user is currently working on. From these classes, relations, and attributes a subtopic area of the problem domain is identified. Based on this, a comment is created from pre-constructed templates and then posed to the user (see Figure 1). In the case of the first comment above, what the system does is to look at the last two classes worked on by the user, Plane and Airport with the relation lands on. From that a comment is constructed from a template where Plane, Airport, and lands on are filled in. For the third type of comment, the system searches the student’s solution for constructions that in their form might be problematic, e.g., many-to-many relationships or deep hierarchies of inheritance. If the learner is currently working on such a construction the assistant can construct a comment about general problems that might occur in such cases. Remark regarding subtopic:
Model under construction
Currently manipulated classes, relations: Airport, Flight, Airplane
Subtopic problem domain: Flight l i
of
1. 2.
3.
XXX Can you represent the fact that flights might be redirected to another destination XXX
Figure 1: Comment selection The analysis of the users’ actions is of course limited to interface manipulations. When users are moving between parts of the enterprise information or switching context without manipulating anything that the system can detect, the comments might become irrelevant or disturbing.
COLLABORATIVE USE OF THE LEARNING ENVIRONMENT Two studies of students using the learning environment were conducted. In the studies students solved modeling problems in the learning environment for 60 minutes. The results indicate that the tools promote aspects of the cognitive practice we designed for even though the long-term effects were not clear (Tholander et al., 1999), (Tholander, 2001). An important finding was the collaboration that students engaged in when using this environment. The different tools became mediating artifacts between the students and something to use as support for discussion and to come up with new solutions from. The learning environment is used by students collaborating in pairs or in smaller groups. The students were given a case that involved a hijacking organization that needed to handle information regarding different aspects of airline flight scheduling and planning. In the studies the students were instructed to use the tools actively as they solved the case they were given. However, they were not given specific tasks to solve with the tools. Instead they were encouraged to use in situations where they did not know how to proceed or needed support in other ways. The following transcript aims to illustrate how two students used the learning tools. We will also relate their interaction to our theoretical foci (italics are used to indicate how they refer to aspects in the learning environment). In this particular episode they are trying to model information regarding the resources, such as the amount of a particular item or what kind of item, required to carry out a hijacking operation. At the start of the transcript (line 1) the assistant comments on the large number of classes in the students’ model. Note that the comment only points at a possible problem in the model, it does not state that this is the case. In the following turns (line 2-6) the
students’ discuss if all the classes they have are relevant for their solution by going through each class and evaluate its relevance. The assistant’s comment clearly made them stop and reflect upon the problem they were currently solving. They later realize that they need to adjust the resource allocation analysis pattern they have inserted into their model (lines 4-11) and they discuss how the pattern should be changed in order to suite their particular problem and that they even might need two versions of the pattern (line 6). The comment of the assistant is interpreted as central to their problem (line 2 and 9) and it makes them slow down their progression, as they have to evaluate and articulate their solution to themselves. (1) (2) (3) (4)
Assistant: B: B: C:
(5) B: (6) C: (7) B: (8) C: (9) B: (10) C: (11) B:
Do you really need all those classes? Well, do we really? Well, let’s see! Implement, ehh, resource allocation, yes we want that We want resource allocation, quantity is good, we want resource type, and then we can make isa from that to all people and then maybe Yes And even, we could need two of this pattern, one that goes to hijacking and one to flights. Proposed and implemented action, what is that anyway? An action, well an action? Yes, what is that? The question is if we need it. He asked that. If we really needed all those classes An action could be like controlling the airplane and that is good. But that’s not really the resource allocation we wanted. Were there any more of those?
A couple of turns later they are still not sure about the level of detail of the resources that their model should describe; if they should go down to the level of specific items or just refer to the kind of items. In order to solve this they make use of the way the two experienced modelers, Paul and Maria, solved a problem in the expert tracks. This helps them to better understand how to use the pattern in their own solution. Even though they are not convinced that the cases from the expert tracks is particularly useful to their problem, they use the expert tracks (line 20 and 22) as a means to articulate aspects of their own solution. In line 20 they draw an analogy between the allocation of resources for a hospital to the resources needed for the flight planning. In parallel they also learn to understand solutions of others and how to relate those to their own problems. Furthermore, they do not only borrow a solution but takes a rather critical stance towards the solutions in the tracks. In order to make use of it they adapt it to their own problem which requires them to think about how the experienced modeler had solved a similar problem. Hence the expert tracks give them possibility to engage in the solutions of experienced modelers from a critical point of view. (17) C: But, we can check this with how Paul did it (18) B: Yes, but his one wasn’t really good (19) C: How about the way Maria did it? (20) B: I think there should be a resource object or instantiation that specific allocation uses, if one uses the particular nurse Eva then that is different from any nurse at all. Paul had not included that, at least not the way I wanted Eh, resources [looks at the tracks] (21) C: Yes, right because he did a really general one (22) B: Our studies have been in the form of field experiments where students are given course like exercises to solve with the learning environment we have designed. The environment became a means to build their discussions and activities around in that it expanded the conceptual apparatus that the students could use to reason with and helped them to see their solutions from new perspectives. Even though the results from the studies indicate that the learning environment supports the cognitive practice that we have designed for it is not clear what kind of learning that would take place if it were taken into full-scale use. Salomon, Perkins, & Globerson (1991) distinguish between effects “with” technology and effects “of” technology. Effects “with” refer to how technology can have amplifying effects on humans. However, it is not sure if these effects are only momentarily. Effects “of” technology refer to the cognitive transformation that most often is desired in learning situations. In these transcripts there are clear indications of effects “with” technology, but it is not possible to say anything about any effects “of “ the learning environment. However, we believe that if we start changing the practices of students course work this will have long-term effects on their learning and their future working practices.
SUMMARY Research in design of software for learning usually emphasizes either theoretical or practical aspects. In this work we have taken a well-defined theoretical position based on cognitive apprenticeship and situated learning. The point of this work has been to explicitly design solutions based on these models of learning in order to investigate how such models can and must be adapted to particular circumstances. The main point has been to discuss how theoretical issues have been taken under consideration in our design. We believe that by thoroughly considering theories of learning, we have been able to design a learning environment that supports the cognitive practice that we aim for.
REFERENCES Anderson, J., Reder, L. M., & Simon, H. A. (1996). Situated Learning and Education. Educational Researcher, 25(4), 5-11. Barab, S. A., Squire, K., & Dueber, B. (2000). Supporting authenticity through participatory learning. Educational Technology Research and Development, 48(2), 37-62. Brown, J. S., Collins., A., & Duguid, P. (1989). Situated Cognition and the Culture of Learning. Educational Researcher, 18(1), 32-42. Collins, A., Brown, J. S., & Newman, S. (Eds.). (1989). Cognitive Apprenticeship: Teaching the Crafts of Reading, Writing, and Mathematics. New Jersey: Lawrence Erlbaum Associates. Collins, A. (Ed.). (1991). Cognitive Apprenticeship and Instructional Technology. Hillsdale, New Jersey: Lawrence Erlbaum Ass. Fischer, G., Nakakoji, K., Ostwald, J., Stahl, G., & Sumner, T. (1993). Embedding Critics in Design Environments. The Knowledge Engineering Review Journal, Special Issue on Expert Critiquing., 8(4), 285-307. Fowler, M. (1997). Analysis Patterns: Reusable Object Models: Addison Wesley Publishing. Greeno, J. (1997). On Claims That Answer the Wrong Questions. Educational Researcher, 26(1), 5-17. Jonassen, D., & Land, S. (2000). Theoretical Foundations of Learning Environments: Lawrence Erlbaum Associates. Jordan, B., & Henderson, A. (1995). Interaction Analysis: Foundations and Practise. The Journal of the Learning Science, 4(1), 39-103. Karlgren, K., Tholander, J., Dahlqvist, P., & Ramberg, R. (1998). Authenticity in Training Systems for Conceptual Modelers. Paper presented at the International Conference of the Learning Sciences, Atlanta, Georgia. Land, S., & Hanaffin, M. J. (2000). Student Centered Learning Environments. In S. Land & D. Jonassen (Eds.), Theoretical Foundations of Learning Environments (pp. 1-24): Lawrence Erlbaum Associates. Lave, J., & Wenger, E. (1991). Situated Learning: Legitimate Peripheral Participation: Cambridge University Press. Morrison, D., & Collins, A. (1995, ). Epistemic Fluency and Constructivist Learning Environements. Educational Technology, 39-45. Noss, R., & Hoyles, C. (1996). Windows on Mathamatical Meanings. Learning Cultures and Computers: Kluwer Academic Press. Pea, R. D. (1992). Augmenting the Discourse of Learning with Computer Based Learning Environments. In E. De Corte, M. C. Linn, H. Mandl, & L. Verschaffel (Eds.), Computer Based Learning Environments and Problem Solving. (pp. 333-343). Berlin: Springer Verlag. Pea, R.D. (1994). Seeing what we build together: Distributed multimedia learning environments for transformative communications. The Journal of the Learning Sciences, 13 (3), 285-299. Ramberg, R., & Karlgren, K. (1998). Fostering Superficial Learning. Journal of Computer Assisted Learning, 14, 120-129. Resnick, L. (1987, ). Learning in school and out. Educational Researcher, December, 13-20. Salomon, G., Perkins, D. N., & Globerson, T. (1991). Partners in cognition: extending human intelligence with intelligent technologies. Educational researcher. 20(3), 2-9. Spiro, R. J., Feltovich, P. J., Jacobson, M. J., & Coulson, R. L. (1992). Cognitive Flexibility, Constructivism, and Hypertext: Random Access Instruction for Advanced Knowledge Acquistion in Ill-structured Domains. In T. M. Duffy & D. H. Jonassen (Eds.), Constuctivism and the Technology of Instruction . Hillsdale, NJ: Lawrence Erlbaum. Suchman, L. (1987). Plans and Situated Actions: Cambridge University Press. Suchman, L. (2000). Human/Machine Revisisted : Department of Sociology, Lancaster University. Tholander, J. (2001). Students interacting through a Cognitive Apprenticeship Learning Environment. Paper presented at the Euro CSCL, Mastricht, The Netherlands.
Tholander, J., Karlgren, K., Rutz, F., & Ramberg, R. (1999). Design and Evaluation of an Apprenticeship Setting for Learning Object-Oriented Analysis. Paper presented at the International Conference on Computers in Education, Chiba, Japan. Wenger, E. (1998). Communities of Practice. Learning, Meaning, and Identity: Cambridge University Press.
