Paul J. Camp, Jackie Gray, Harriett Groves, and Janet L. Kolodner

College of Computing

Georgia Institute of Technology

Atlanta, GA 30332-0210

{pjcamp, grayj, harriett, jlk}

The Learning-by-Design group at Georgia Tech has been developing science units for middle school that engage students in design problem solving as a vehicle for learning science (Kolodner, et al, 1998). While achieving design challenges, students identify what they need to learn and why they need to learn it, engage in investigation activities, and have myriad experiences applying what they are learning in attempts to achieve the design challenge. They have ample opportunity for practicing science process (e.g., experiment design, measuring, observation), experiencing scientific phenomena they are studying, both succeeding and failing in their applications of what they are learning, and practicing communication, planning, design, and collaboration skills. Materials we provide for them, cases, classroom rituals, and journaling prompts (design diaries) scaffold inquiry, investigation, and application, as well as the reflection needed to explain failures, to extract from their experiences what they are learning, to develop their skills, and to acculturate themselves into the practices of science and design.

Initial assessments suggest that students are learning science content as well or better than their peers in more traditional classrooms, that their science process skills are as good as those of students in the best gifted classes, and that, in addition, LBD students are more attentive to the need for planning and better able to collaborate to solve problems and design experiments than are their peers from traditional classes.

But our successes are in the area of physical science, where it is possible for students to work on design challenges in which they can construct, test, and refine real working artifacts. It is far more difficult to design such challenges for earth science and life science. There are serious problems of scale involving size, complexity, and/or time that render such designs intrinsically difficult to accomplish. Many artifacts that we could have students design don’t afford testing easily. For example, a retaining wall to control erosion is difficult to test since we cannot control the weather. Other artifacts we could have them design (e.g., tunnels) are so large and difficult to construct (not to mention test) that they are simply impractical as classroom activities.

In physical science, we have learned a great deal about creating and implementing collaborative design activities that may be leveraged for earth science. Since we cannot pragmatically study actual systems, however, we turn to modeling -- both for learning science concepts and for trying out ideas. Models must satisfy a set of criteria intended to insure a close match with real phenomena. Design activities can focus on a well-done model for feedback. As well, their modeling activities provide a vehicle for helping students learn the important science skill of interpreting models and understanding their intrinsic limitations. We turn as well to use of expert cases, which show the issues involved in complex design challenges and how experts have applied science to achieve design challenges. Use of expert cases helps students understand the applicability of what they are learning and experience the complexities involved in solving real-world problems.

We continue by presenting a short overview of learning by design, our progression of earth science units, and then the principles we’ve derived that have allowed us to transfer what we’ve learned about learning from design of physical devices to learning from design of complex plans for complex real-world problems.

Learning by Design in Brief

The typical sequence of activities in an LBD unit (Kolodner et al, 1998) has students encountering a design challenge and attempting to solve it. Students engage in a system of classroom rituals that promote reflection and scaffold reasoning and learning. "Messing about" activities help students make connections to what they already know and promote curiosity and question asking. "Whiteboarding" (borrowed from PBL (Barrows, 1986)) promotes articulation of what they already know, their understanding of the challenge, their ideas about how they might go forward in achieving it, and the questions they need to find answers to in order to be successful. "Rule of thumb generation" promotes extraction of what they are learning from investigative activities in the form of concrete guidelines for application. Rules of thumb serve the purpose of theory abstraction from specific observations. Question asking is followed by investigation. Investigation can take a variety of forms — designing and running experiments, reading in books, talking to experts, or reading cases. After each investigation, students return to the challenge and consider how they might apply what they've learned. "Solution exploration" provides time for generating solution ideas based on investigation. Students consider several ideas, choose one to move forward with, and begin to plan for how to make it work. "Pin-up sessions" provide an opportunity for students to share their design ideas with classmates and hear their classmates' suggestions. "Gallery walks" provide opportunities to show off works in progress. Students iteratively construct, test, analyze, and refine their solutions, refining both their solutions and their understanding of important concepts in the process. Students engage in the practices of both scientists and designers. "Design diaries" are for recording questions, experiment designs, results of experiments, ideas, and what’s been learned, and for keeping a running history of one’s design in progress. The system of activities balances planning, doing, and reflection and balances individual, small-group, and whole-class work.

This integrated system of rituals and tools are manageable by middle-school students and motivates the kinds of reasoning and behavior that promote learning. We’ve learned that an important component of learning science from design activities is the use of science language and concepts in explaining one’s decisions. Thus, the emphasis in all collaborative activities is on communicating one’s ideas well enough so others can understand them and justifying one’s suggestions, opinions, and decisions with knowledge gained through investigative activities. Important to success, as well, is that teachers model values that make teacher and students a team.

Our Progression of Earth Science Units

We continue to learn and refine how to make design work in the physical sciences, but we have been struggling with how to make it work in the earth sciences. We have worked with one earth science teacher for several years. From this teacher’s persistence and trials in her classroom, we began a collaboration with a number of researchers to further develop her ideas. From this collaboration we have identified several developmental considerations that we wanted to address.

If we could not ask our students to design and build a structure or a tunnel, what would serve to motivate their need to know the science? We had worked with other earth science teachers over the summer to develop a unit on erosion, and that effort included the use of models to give students an understanding of the dynamic forces involved in erosion and reading of expert cases to gain an understanding of the ways the science of geology comes into play in designing our landscape. Our next unit was to deal with the earth’s formations and map reading. We began to work on a way students could grapple with the earth’s formation and it’s incredible variety.

We decided that we had the advantage of presenting a developmental progression across our earth science units. Models and their use could be a theme across the units. Using maps and other representations (e.g., cross sections) would also be developed across the units. Finally, we could use cases more systematically in these units (Dahger, 1997, Kolodner, 1997). We are using the term "model" in a precisely defined operational sense. It has the properties of similarity, reliability, and scale, and anything that satisfies those properties, we are willing to admit as a model. Maps serve as a first step in the direction of moving from concrete models, such as were used in the erosion unit (with all their intrinsic limitations on accuracy and flexibility), into abstract models, such as are more common in science and engineering. It may be that this can eventually lead to the goal of purely mathematical modeling.

The three units we are developing and testing this year are erosion, earth formations, and plate tectonics. We provide a brief sketch of these three units integrated by models and cases.

The Erosion Management unit focuses on two major issues in addition to erosion — how to make and use models and the possibility of negative "downstream" consequences of our actions. These are addressed in the context of a design challenge in which students are trying to manage erosion around a basketball court that their school is proposing to build at the bottom of a hill. We help students develop an operational definition of models and modeling that specifies the characteristics a model must possess to be useful in understanding a design problem. Two of those characteristics, addressed in the erosion unit, are similarity (does it look like what you see in the world?) and scale (are the pieces of it the correct relative size?). Similarity is supported by an initial activity observing erosion and erosion management in the students’ neighborhood. Models are used first to reproduce what was seen and later to explore potential management methods. The whiteboarding episodes during this unit will end up leaving enough unanswered questions regarding rocks and minerals to serve as a lead-in to the next unit.

The Tunneling unit addresses three big ideas with respect to models and representation, in addition to its focus on rocks, water, and geologic formations. It continues the theme of operational definitions by using physical characteristics (such as hardness or luster) to identify rocks and minerals rather than their enormously variable superficial appearance. It also introduces the third criterion for models -- reliability. There is no one correct model for a physical phenomenon since all models are approximations. The only way we know we can trust conclusions based on models, then, is if different people using their own models arrive at the same conclusions. This idea is scaffolded through the use of personally-created models as well as by discussion of the variety of models used (and reactions to them) in a Nova videotape, The Fall of the Leaning Tower. Lastly, the unit extends the students' notion of models to a more abstract level by noting that maps fit our three criteria for models. Students create, as well as use, existing geologic, topographic, satellite, relief, cross section and bathymetric maps, mostly in combinations to see relationships between the different things represented by each map. In addition, there are some simple physical models constructed for various purposes such as exploring how water migrates through the ground into aquifers and what characteristics of rock make for good tunnels that don’t collapse as soon as you dig them. Their ultimate goal is to produce a geologic report detailing the potential problems to be encountered in the construction tunnels for a train across Georgia. They will need to pick out the places that are likely to be problematic and also the places where the geology is sufficiently problematic that core samples should be taken to gather further information.

After presenting the challenge to students, a modeling activity concerned with scale is presented as a messing about ritual. The purpose of this activity is to have students explore what happens when different materials are placed over a tube (representing a tunnel) and the tube is removed. What happens? Does the material leave a wave-like dip when the tube is removed (i.e., when the earth is widdled out below it)? Does it remain stable? How does it change with different formations of materials?

Cases are presented in two ways after the initial modeling experience: (1) a textual case of the Bay Area Rapid Transit tunnel and (2) a video documentary on tunnels and tunneling. In addition to several cases of tunnels and tunneling, the video shows the advancement of the technology of tunneling and the geological forces that had to be overcome to successfully build tunnels.

Students return to the challenge. The whole class identifies facts, ideas, and learning issues that need to be examined to solve the challenge. We use whiteboarding to capture and record students' natural curiosity, providing a framework for moving forward and doing investigation. Their questions lead to a need to investigate. Questions students ask are expected to center on what the land is like (e.g., rock types, water, natural features, soil), properties of rocks, how maps can help, minerals, what cross-sections are, impacts of water, and modeling. Questions that center on the type of rock formations found in the tunneling areas will lead students to the rock lab where they can notice, identify , and better understand the characteristics of various rocks and minerals. Questions will remain about where to find these types of minerals and rocks, and the geologic maps will be needed for further study. Multiple representations and maps are now addressed systematically as students are curious to know where to find the various formations. Additional cases are read to understand more about actual tunneling efforts. A set of modeling activities helps students make the translation from two dimensions (on some maps) to the three dimensions on others. The cases lead to additional questions about water. Questions arise as to how does the water get there? The cross-section map is revisited and the water tables are noted. More physical modeling helps students learn how water makes its way through the ground to create a water table. Aquifier maps are studied. And so on. Learning how to model is integrated throughout all the activities and moving back and forth from the challenge to the cases to modeling experiences is encouraged.

After cases are read and modeling is done, a rules of thumb table is created. These rules codify what's been learned and serve as a guidelines for making decisions about design. As well, they provide a way of indexing the cases and experiences students are having, allowing them to be reminded at appropriate times. Rules also provide a concrete way of forming a connection between science and design. "If rock is fractured, then water will flow through it" and "tunneling below the water table can cause flooding problems" are powerful examples.

Students are now ready to engage in solution exploration. This consists of pulling together what's been learned to formulate a solution or set of potential solutions. Student groups produce pinups of their ideas, specifying justifications for each idea they've proposed. Different students have different expertise because of the modeling they've done or the cases they've explored. Important to these pinup sessions is the critiques other students can provide based on what they know. Students iteratively refine their proposed solutions and continue with investigations they still need to carry out, holding pinup sessions periodically to share their expertise and provide each other with feedback. In the end, they produce a geological report, cross section of the geology encountered, site plan on a topographic map, and structural sketches of the tunnel they propose.

The questions still remaining deal with why is there the diversity we’ve seen in the land formations? Why would certain rock formations be in one place and not another? What causes the folds in the cross sections we seen? What causes the fractures that water is found in? What could account for all this amazing variety? Plate tectonics comes next and is our next effort.

Principles of Design

What we’ve struggled with most in designing our earth science units is how to provide students with useful feedback that would help them identify their misunderstandings and gradually develop deeper and more accurate conceptions. In physical science, that happens primarily through constructing and running working examples of the item they are designing. But in earth science, we had to find other ways to help students apply what they are learning in ways that afford useful feedback. We’ve come up with five modes for feedback, all authentic to the design workplace. None, by itself, is as good as building a device and seeing how it behaves, but we hope their use as a system will provide the kind of feedback students need to identify their misconceptions and deepen their understanding.

Pinups and gallery walks, in the past, have been primarily for the purpose of sharing ideas with others, and helping others to experience more diversity in use of science concepts. They also serve as a venue for feedback, but it was not as critical a role. For our earth science units, pinups and gallery walk sessions are an essential forum for feedback. We encourage students to play the role of critical advisor for others by distributing their expertise. Students do different investigative activities and read different cases. While they share them with others, those who investigated something fully will usually be more expert than others and be able to provide informative critiques of other students’ work.

Modeling activities are a cornerstone of our approach in earth science. Students can’t design, construct, and test tunnels, but they can model the way different kinds of soil or rock formations effect structural integrity when a tunnel is drilled, the way water travels through formations, the effects of rain on soil formations, and so on. Indeed, this kind of modeling activity is integral to the work of engineers and architects. They can’t build a chemical processing plant or tunnel or bridge and then try it out and see what happens; they must design based on understanding of scientific principles and then test using models. Models allow scientists, engineers, and our students, to adjust the physical or temporal scale of processes so that they can be easily observed. Our students use modeling activities both to further their understanding of such processes and to test their design ideas. Student use of modeling as a tool of science helps them learn relationships between representations of things and things themselves. Benchmarks (AAAS, 1993) tells us that modeling is an important science skill for students to learn.

Exploration of multiple perspectives and representations is another cornerstone in earth science, afforded by the wealth of maps available from the US Geological Service. Each highlights different earth characteristics. We introduce students to maps as models of some aspects of the earth — each type of map focuses on different earth characteristics and therefore highlights different characteristics in its representation. Each affords answering different questions. Students might reason based on one set of maps and critique based on another set. With growing expertise, they will someday be able to reason based on larger and larger sets of representations.

Reading about and reasoning about real-world cases is another cornerstone of our approach in earth science (Kolodner, 1997; Dahger, 1997). Our earth science design challenges require a good deal of knowledge about geologic processes. The cases we provide for them help students to draw connections between earth’s characteristics and the structures we build on it. We write the cases to focus on three issues (Kolodner, 1993): the problem the geology posed for the tunnel builders (or landscapers or engineers); the way the geological challenges were handled by those professionals; and what happened as a result. We write them in such a way that students can extract rules of thumb that connect our knowledge about the earth to our ability to manipulate its formations and surfaces. Some cases are for the purpose of helping students generate questions; others for the purpose of helping them generate solution ideas. As suggested earlier, if some students are more expert at some cases than others, the same cases read by one group that helps them come up with ideas become grist for feedback to other groups about the viability of their ideas.

Commentary and critique by experts is one other optional source of feedback for students that can be quite powerful. An expert visiting the classroom partway through a design challenge can critique students’ ideas in progress and tell additional stories to make the challenge more concrete for students, helping them focus on what else they need to learn. Later in the challenge, an experts’ critique can clarify things students have already learned, helping them fill in gaps in their knowledge.

As we’ve learned from working on physical science units, we can’t expect students to know how to do modeling or to use cases right away, or even to be expert in useful critiques of their peers’ design ideas. We therefore take a developmental approach to helping students learn these skills (Vygotsky, 1978). Opportunities for refining their skills are interwoven in the erosion and tunneling units. They use the modeling, case reading, pinup sessions, and gallery walks as opportunities for developing more expertise in those important skills while they are learning about earth science. The books we are developing to go with these units focus both on the science concepts and the process skills needed to learn science well. We are currently putting our earth science units into classrooms. We have six teachers piloting these units with 750 students beginning in November, 1999, and running through February, 2000.


This research has been supported in part by the National Science Foundation (ESI-9553583), the McDonnell Foundation, the BellSouth Foundation, and the EduTech Institute (with funding from the Woodruff Foundation). The views expressed are those of the authors.


American Association for the Advancement of Science. (1993). Benchmarks for science literacy: Project 2081.New York: Oxford Press.

Barrows, H. S. (1985). How to design a problem-based curriculum for the preclinical years. NY: Springer.

Dagher, Z. R. (1997). The case for analogies in teaching science for understanding. In J. J. J. Mintzes, H. H. Wandersee, and J. D. Novak (Eds.). (1997). Teaching Science for Understanding: A human constructivist view. (195-211). NY: Academic Press.

Kolodner, J. L. (1993). Case-Based Reasoning. Morgan-Kaufmann Publishers. San Mateo, CA.

Kolodner, J. L. (1997). Educational implications of analogy: A view from case-Based Reasoning. American Psychologist, 52, 35-44.

Kolodner, J.L., Crismond, D., Gray J., Holbrook, J., & Puntambekar, S.(1998). Learning by Design from Theory to Practice. Proceedings Third International Conference of the Learning Sciences '98, pp.16 - 22.

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, Mass.: Harvard University Press.