Meet Gautam Biswas

Back to Perspectives

CIRCL perspectives offer a window into the different worlds of various stakeholders in the cyberlearning community — what drives their work, what they need to be successful, and what they think the community should be doing. Share your perspective.

Gautam Biswas

Gautam Biswas is a professor in the School of Engineering at Vanderbilt University.

How did you get started in cyberlearning?

My early education was in engineering, and one of the areas I worked in was modeling and simulation of systems. I realized that the general approach to modeling and simulation of complex systems could be translated into a form that was useful for teaching others about systems, especially physical processes. That’s how I really got interested in seeing how we get students to really understand concepts and phenomena in science and engineering in a much deeper way by actually getting them to build models of these systems.

What really got me into education was that after coming to Vanderbilt, I met two very well known education researchers, John Bransford now at the University of Washington in Seattle, and Dan Schwartz who is at Stanford. We talked about how they might implement some of the work they were doing in computer based environments to facilitate problem solving, as well as data gathering and analysis. They were not computer scientists, therefore, they looked to me to figure out ways in which we design and analyze such tools. My initial goals were not very lofty. If we got students to work in a computer environment, we could capture and analyze data of their work more easily than if one relied on paper and pencil activities. That’s how I got started and worked with them on a number of systems, including the Teachable Agents project.

If we walking into a learning environment that was using your technology, what would it look like, including the role of the teacher?

Our overarching approach is get students to learn science phenomena by building computational models. Not only do students use the computer tool to build their model, they can use their model to answer questions. This allows them not only to learn by building, but also monitor their own learning processes, and improve and correct their models, if they produce erroneous answers to questions. For example, this is the case when students build a causal model to teach Betty about a science topic in Betty’s Brain. Environments such as Betty’s Brain, are embedded in a project-based curriculum, and we are working on addressing the new science standards where students have to do more critical thinking, analysis, and problem solving. For example, we get students in 5th and 6th grade to build models of climate change. They answer questions like what are the human causes that contribute to the greenhouse effect and what are the implications of this (greenhouse effect) on climate change.

The teacher leads these activities in the classroom and the students take 4-5 days to build their complete model. The students teach Betty, their agent, so she can correctly answer all questions on quizzes generated by the Mentor agent, Mr. Davis. But their larger goal is to find situations in the school and home environment that add to the greenhouse effect and explain why using that model. Then they look for solutions and ways to reduce the carbon footprint. Students work individually on their models, and then work in small groups on different problems of their choice. We allow them to revisit the model they built as many times as they want, to relate the activity they are studying to the greenhouse effects, the consequences, and how they may modify the activity to reduce the carbon footprint. At the end of the project, students prepare their presentations that describe the problem and potential solutions to reduce the greenhouse effect to the rest of the class. It is working out to be quite successful. In the future, we want to formalize this process to build problem-based learning units.

What are the different domains that you are currently working in, and what do your teams look like?

In the Betty’s Brain system we have developed curricular units that cover climate change, body temperature regulation, and natural ecological processes that include ecocolumns, pond and desert ecosystems. In another project, CtSiM, students use a visual programming interface (similar to Scratch) to build simulation models of science phenomena.
In this project, we are currently working in 2 different science domains, kinematics and ecology, because these are common topics covered in 5th and 6th grade science classrooms. We are gradually moving towards teachers developing their own units––teachers come up with the terminology they want the students to use to teach a science topic, and we work with them to build the underlying computational units that form the building blocks for the simulation models. Whereas our current focus has primarily been to promote science learning, the environment can also be used to simultaneously facilitate learning of mathematical concepts, and developing capabilities for engineering problem solving. The big gains that we observe can be attributed to the synergy between science and computational thinking. Students seem to learn the science in a deeper manner, and pick up computational concepts as they work in the CTSiM environment. In this way, students are learning fundamental computer science concepts in a science classroom, with teachers who were not originally trained in computer science. We believe this will help students tremendously in the future, primarily because they develop computational thinking ideas, and secondarily because they are getting a head start in computing even before their high school years.

For the CTSiM project, we are a team of 3 faculty members. My background is in AI, cognitive science, educational technology, and systems engineering. Pratim Sengupta, a collaborator from Peabody School of Education at Vanderbilt University is involved in research on agent-based modeling and simulation, and analysis of complex systems. Doug Clark, also a faculty from the Peabody School of Education, works primarily in science education, and has been developing games to help middle school students learn kinematics concepts. We all have a strong interest in science education and design-based research, but we each bring a different set of expertise to the table. Currently, we want to focus on building scaffolding and support into the CTSiM environment to help students overcome the challenges they face in understanding science phenomena and in computational modeling. We have identified a number of challenges that students face during learning, and we want to incorporate teacher input in developing scaffolds that can support the students even better.

What would you say is your most significant project right now, with the most impact?

There are 2 significant projects I am working on now. The first is Betty’s Brain, a project that we have worked on for a number of years now. Our initial goal was to build a system that exploited learning is a social process. The literature says that one way to really learn something in a deep manner is to teach it, and we have worked on emulating that in a computer-based “learning by teaching” environment. A lot of the work we did initially was learning how to design systems where we could exploit the social interactions between a student and a computer agent, and also provide them with scaffolding and support so that they could go back and learn for themselves. We found this to be very successful. Students were willing to spend more time in the learning process.

More recently, we are doing more work in Computational Thinking in Simulation and Model-Building (CTSiM). Here, we have decided to go one step further and not only ask students to build models of science phenomena, but also ask them to build them in a way in which they can actually simulate the model and see the behaviors that these models generate. They can then relate these models to real-life situations. We are currently working on developing and implementing scaffolds within the CTSiM environments to support guided discovery.

Share this: