Designing to Learn About Complex Systems (Hmelo, Holton, Kolodner)

Hmelo, C.E., Holton, D.L., Kolodner, J.L. (2000). Designing to learn about complex systems. Journal of the Learning Sciences 9(3), 247-298.

Design activities, which allow explorations of how systems work, can be an excellent way to help children acquire a deeper, more systemic understanding of such complex domains. In this article, the authors incorporate lessons from problem-based learning, structure-behavior-function theory, and case-based reasoning, to construct a two-week lesson plan for middle schoolers that involves both modeling and design. They call this Learning by Design.

Learning from design activities “has the advantage of more completely considering a system as a union of functionally related subparts, but it does not address how students can be motivated to be engaged in such learning, nor does it address how we can support students’ evolving understandings from superficial to deep.” The authors’ Learning by Design (LBD) approach takes those next steps by asking students to not only think of systems as designs, but, in addition, to understand some natural design (e.g., the lungs) well enough so that they can undertake the design of an artificial system that can carry out the same functions.

The LBD approach builds on Perkins’ “knowledge by design” approach. “Perkins suggested helping students view systems as designs: structures adapted to specific purposes. Viewing a system as a design goes beyond simply defining the parts, and also addresses their functional roles, the mechanisms by which those roles are carried out, and how those functions causally interact with each other” (p.248).

“Systems are dynamic entities, and many of their organizational levels are difficult to visualize, such as cellular respiration. Part of the difficulty individuals have in understanding complex systems can be related to the way students are introduced to them (Feltovich, Spiro, & Coulsen, 1993). Often, learners are introduced to complex systems in oversimplified static forms [reminds me Spiro and Jehng’s complaint in “Cognitive Flexibility and Hypertext”, leading to greater danger of “reductive bias” (Feltovich et al., 1996)], and these early conceptions form schemas that are then difficult to overcome. In life science, in particular, there is often an emphasis on understanding isolated concepts without introducing learners to the interrelations among various levels of the systems.”

The authors use SBF (structure, behavior, function) theory (Goel & Chandrasekaran, 1989) as a way to help student understand and describe systems. This involves building a representation that “highlights structures, functions, and behaviors of systems, and the connections between those parts. Structure refers to the physical structures of a system (the lungs are a physical structure in the respiratory system); Function refers to the purpose of the system or subsystem (the respiratory system transports oxygen throughout the body to the organs that require it); Behavior refers to the dynamic mechanisms and workings that allow the structures to carry out their function; that is, the mechanisms that cause changes in the structural state of a system (p.250).

Behavior of a system is probably the most difficult for novices to understand because it involves invisible (e.g., the invisible electrical impulses traveling through nerves that cause the respiratory system’s involuntary movements) and time-delayed causalities. “Novices tend not to consider that a system has behavior until some anomaly in the normal function of a system arises and they have a need to debug or explain it (Murayama, 1994)…Because structure is visible, novices typically begin to understand a system at a superficial structural level, only later beginning to understand how structures are related to each other through behavior and function. On the other hand, when experts reason about systems (e.g., for design or diagnosis), they typically reason first at the functional and behavioral levels (Chi, Feltovich, & Glaser, 1981).” (p.250-1).

Potential benefits of using design:

  • Design places students squarely in the process of constructing rather than receiving knowledge (Lehrer, 1993; Perkins, 1986).
  • The process of design initiates students into the discourse that typifies communities of practice in science and engineering (Roth, 1995) — students learn to use concepts and heuristics as tools for problem solving.
  • Discussion about the designed artifacts facilitates sharing of knowledge—the artifacts they are designing can serve as concrete referents that students focus attention on as they are communicating (Roth, 1996).
  • In the classroom, the process of designing affords the potential for learners to construct, apply, and evaluate models (Kolodner et al., 1998; Penner, Lehrer,&Schauble, 1998; Roth, 1996). Students may construct a model of the device they are designing, of some part of it, or of some mechanism they need to learn more about to achieve their design goal.
  • The design, construction, testing, and refinement of models provides a powerful means of coming to know our world and thus offers the potential to enhance understanding in scientific domains (Penner et al., 1998). As students design models, they need to describe, predict, or explain some phenomena, which requires them to discuss and invent objects and their relations to each other as well as to consider functions and causal behaviors of the components in their model.
  • Tradeoffs include:
    • Finding a balance between having students work on design activities and reflecting. Incorporating reflective activities is important to encourage an
      understanding-oriented approach;
    • Learning how to integrate real world knowledge without letting it overwhelm the class with irrelevant aspects of the world that might take the students on unproductive tangents;
    • Determining how to maintain extended student engagement in a manner that emphasizes principled understanding rather than task completion.

Types of scaffolding that may be needed:

  • Help with identifying new goals for learning
  • Making one’s thinking visible
  • Seeking answers to questions one has generated— with the ultimate goal of helping students anticipate and seek out better and better models to explain phenomena.

Modeling and design have many similarities—both lead to the construction of an artifact, and both require managing the interconnections between a variety of structural and functional parts. When the goal is to create a model, success is judged in terms of whether the model elucidates the phenomena being studied. When design is the goal, criteria for success are the satisfaction of functional constraints. We borrow from both traditions, asking students to use design practices to come to an understanding of the issues involved in achieving a large and complex design challenge (designing an artificial lung), to focus investigations on some of those issues (e.g., How do human lungs work? What artificial mechanisms can model the functions of human lungs? How do lungs connect to other organs?), and to construct and test models to answer those questions. [The authors later admit that this duality of (conflicting) goals produced some confusion among the students.]

The design challenge was used to promote engagement, questioning, and connection to the world. It was then assumed that students would recognize that to learn what’s needed to achieve the design challenge, they would use an investigative method engineers and scientists use—modeling. Unfortunately, this connection btw modeling and design was not made clearly enough. “The design challenge had not helped students sufficiently focus their earlier investigations as intended…Also problematic was the lack of explicit connection between the design challenge presented to the students and the modeling activity they actually did. It was ambiguous to us, and to the students as well, whether they were constructing a miniature artificial lung, a model of an artificial lung, or a model of a real lung or part of the respiratory system” (p.276).

Problem-based Learning (PBL):

  • Begins by presenting a group of students with a complex problem to solve.
  • Students pause to reflect on the data they have collected so far (FACTS); generate questions about those data, and hypothesize about underlying causal mechanisms that might help explain those facts or potential solutions to the problem they are working on (IDEAS). They identify concepts they need to learn more about to solve the problem (LEARNING ISSUES) and then they develop a plan for proceeding (ACTION PLAN).
  • During investigation—students divide up and independently research the learning issues they have identified. They regroup to share what they have learned, reconsider their ideas, and/or generate new ones in light of their investigations. They continue by attempting to solve the problem based on what they have learned, sometimes cycling several times through question generation and additional research. The dynamically changing lists of facts, ideas, learning issues, and action plan are recorded in columns of a public whiteboard, as shown in Figure 1. The whiteboard is revisited and updated on a regular basis. The whiteboard helps learners remember where they have been and where they are going in their learning and design. It helps focus negotiation of the problem and provides an anchor for students to co-construct knowledge. After completing their task, learners deliberately reflect on their experience to abstract the lessons learned and to consider how they performed in their self-directed learning and collaborative problem solving.
  • The facilitator (i.e., teacher) is responsible both for moving the students through the various stages of problem solving and for monitoring the group process—ensuring that all students are involved and encouraging them both to externalize their own thinking and to comment on each other’s thinking.
  • The facilitator plays an important role in modeling the needed thinking skills. Another important role the facilitator plays is helping students bring in their knowledge of the world as a contribution to problem solving while at the same time helping them to move forward without being overwhelmed by the magnitude of the problem. That can be carried out by allowing any and all ideas and questions to go onto the whiteboard initially and then guiding discussion toward identifying the more realistic and important ones, helping students create an action plan that has them investigating only the most important of the questions raised. Although the facilitator fades some of his or her scaffolding as the group gains experience with the PBL method, he or she continues to actively monitor the group, making moment-to-moment decisions about how best to facilitate.

Case-based reasoning (CBR):

  • Was developed as a way of allowing computer programs to solve complex problems.
  • Based on observations of experts—particularly their ability to learn from experience and re-use knowledge learned in one instance in another instance.
  • CBR, as a method for computer reasoning, focuses on storing problem-solving experiences (cases) in a case library, interpreting and indexing them so that they could be found and re-used at appropriate times, ways of searching that case library to find appropriate cases, and adapting and merging the solutions  to old cases to solve new problems.
  • CBR’s suggestions are in keeping with much of what is proposed by constructivist education and in keeping with the classroom experiences of researchers looking at learning in the classroom (e.g., Feltovich et al., 1992; Linn, 1995; Penner et al., 1998).
  • CBR shows that learning is an iterative process: One’s initial conceptions are often incomplete or faulty, and only by trying them out, seeing if results are as one expected, and attempting to explain ensuing anomalies can one discover the holes in one’s knowledge and be motivated to fill them. Thus, according to CBR, deep understanding requires the experience of one’s expectations failing and the subsequent need to explain why. One’s expectations fail when one tries something expecting one result and a different result ensues. When one fails at something one wants to accomplish or understand, one is motivated to explain the failure (i.e., to learn more). Conversely, it is difficult to recognize a lack of understanding without the opportunity to try out ideas and observe what happens.
  • CBR suggests that the better the connections between one’s goals, one’s solutions, and their outcomes, and the better one has articulated those connections in explanations, the better one will be able to re-use the results of one’s experiences.

Some lessons learned include:

  • Inadequate scaffolding. There was no scaffolding provided to help students plan the design of their lung models based on what they had learned. Also no scaffolding to help them debug and explain their models (p.273).
  • Students stopped at the point where each had a first partial working model, thus they missed out on the chance to apply their ideas in a new situation. “They had no opportunity to continue adding to their fluency (p.274)…”[or] iteratively make their models better — in parallel with refining and increasing their own understanding” (p.275).
  • Students’ research was focused on big issues, not on details needed to be able to design or build something that could work.
  • If a more explicit connection had been made btw design challenge and modeling activity, students would have had a chance to be introduced to the process [modeling], they could have had an interesting science process discussion and perhaps learned about modeling as something scientists and engineers do as a method of investigation” (p.276).

From p. 267: “Although the task was to design an artificial lung, this was overwhelming for these students, so they transformed the task into a modeling task that they could manage.” [Does this mean that the authors initially wanted more creative solutions?]

“Students had a hard time getting to a point where it made sense to grapple with the details of implementation. Yet it is in implementation that design challenges have their most powerful affordances for learning. When one can build and test a device or some piece of it, one can grapple with one’s conceptions. A device that does not work signals a lack of full understanding. The best design challenges for promoting learning are those that afford construction, testing, timely and authentic feedback, and revision” (p. 287).

Also, the materials they provided were “leading” and incomplete. The authors also suggested that “messing around” with the materials before fact-finding may also help students gain insight. “In addition to achieving focused investigation during a design challenge, it is important that the materials students “mess about” with and construct with have affordances and constraints that promote identification of the full range of important issues. The balloons and straws and plastic bottles we made available provided affordances for getting things in and out of something by blowing or pumping air. But we provided no materials that could be used for filtering, and we provided no materials that could be used to build a working pump. It should be no surprise then that most students focused on getting air into and out of the lungs and only one team considered issues of gas exchange. It should be no surprise, either, that only one team recognized a need to investigate issues associated with controlling the flow of air into and out of the lungs—the team that brought an electronic construction kit from home and built a working pump.” (p.288).

“Materials provided to students for messing about and construction can focus investigation through their affordances or poorly constrain what students think about through their lack of affordances. It is important to consider the range of issues we want students to investigate as part of a design experience, to choose materials that will afford identification of those issues and their productive investigation, and to give students experience messing about with those materials early on” (p.289).

“We’ve come to realize that if we expect students to model some phenomenon, we must include, as part of our classroom activities, a discussion of what a model is, what we can use them for, and what kind of fidelity is needed for a model to be useful…” (p.290). [Perhaps that same goes for simulations? (Molecules and Minds)]

A good design challenge for learning:

  • Affords construction, testing, timely and authentic feedback, and revision of something that works;
  • Makes clear the utility of learning targeted facts, concepts, and skills;
  • Focuses investigation at a concrete level;
  • Can serve as vehicles for promoting model design, model building, model running, and an understanding of modeling as an investigative method;
  • Should be supported with materials that promote the construction of working models; and
  • Can be orchestrated such that students will understand the relation between the challenge and any model building and testing activities they are doing.

The authors recommend the following sequencing of activities for good learning from design activities:

  1. Presentation of the design challenge with appropriate material to promote engagement.
  2. Engagement in some quick construction activity or an activity that has students investigating how some artifacts work to get students started recognizing specific issues they need to address to achieve the challenge (what we referred to as “messing about” earlier). It is important here that the materials they have available and the guidance students have for messing about be chosen such that students are able to identify a wide range of the issues that need to be addressed.
  3. Whiteboarding or some other reflective and planning activity that gets students together as a class to record ideas, learning issues, and plans.
  4. Planning and carrying out some of the investigations that will help them address some significant subset of the issues they have identified as important (e.g., through reading, experimentation, modeling), along with presentation of findings to the class, discussions of the applicability and implications of findings, and return to the whiteboards to articulate what they have learned and how far they have come; it is entirely appropriate here to have different groups in the class investigate different topics.
  5. Revisiting the design challenge to apply their new knowledge, planning their best solutions based on what they’ve learned so far, and presenting them to the class.
  6. Iterations of construction and testing of solutions interleaved with additional investigation as needed and with presenting to each other what they have done and learned with each iteration of their designs.
  7. Final presentations, reflection, and review.

Another important consideration: “[Teachers’] lack of experience with designing and modeling makes it difficult for them to feel comfortable staying with an activity for several iterations, to completely grasp design’s affordances, and to articulate to the children the reasons for what they are doing” (p. 293).

“When children conduct experiments, they often focus on creating outcomes rather than constructing understanding. This is understandable because the outcome (some artifact that works) is a concrete product, whereas understanding the internal workings of a system is more abstract and less tangible” (p.254).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: