Challenges in Learning about Complex Systems
Cindy E. Hmelo-Silver, Rutgers University
My research on complex systems began as part of the learning-by-design project (Hmelo, Holton, & Kolodner, 2000). In this work, I initially focused on human biological systems and our research group was investigating how to use design to help support science learning in middle school science. I observed that the children tended to think about structures, with little understanding of functions and behavior. The idea of structure-behavior-function (SBF) as a conceptual representation for thinking about systems originated in work in artificial intelligence that demonstrated that such a representation could be used to effectively reason about designed systems (Goel et al, 1996). To investigate whether this SBF representation might account for expert understanding of complex systems, I studied experts and novices in two domains: human biology and aquarium ecosystems. The results of this study demonstrated that experts thought about complex systems in terms of SBF whereas novices represented these systems largely in terms of structures, occasionally in terms of functions and only rarely in terms of behavior (Hmelo-Silver, Marathe, & Liu, 2007). Novices tended to have simple mental models that revolved around a single structure (e.g, lungs, fish). Experts tended to think about these systems either hierarchically if they were scientists or pragmatically, depending on their goals. For example, the hobbyist experts in the aquarium domain considered what it would take to maintain a healthy aquarium in terms of keeping the fish healthy, breeding, etc. The scientists experts considered how the entire system was driven by energy. One conjecture is that the hobbyist model might form a bridge to a hierarchical scientist model. Another conjecture is that organizing instruction around SBF might help learners push beyond just considering structures.
The first proof of concept study compared structure-oriented hypermedia with function-oriented hypermedia (Liu & Hmelo-Silver 2009). The results demonstrated that the function-oriented hypermedia helped learners develop a deeper understanding of phenomena that were occurring at a microlevel. These were phenomena that novices never considered in the Hmelo-Silver et al, 2007 study. The results suggest that organizing text in terms of a conceptual representation can be a powerful tool for making the invisible visible. However, behaviors were still not well represented, This made sense as the hypermedia was a static medium. Helping learners understand system behaviors was a challenge that required the use of dynamic models.
To address the challenges of helping learners understand the behavioral and functional dynamics in ecosystems, we used the aquarium as a model system. We made behaviors and functions visible through the use of NetLogo simulations (Wilensky & Reisman, 2006) that allowed us to focus on the emergent behaviors and functions at both macro and micro level (Hmelo-Silver, Liu, Gray, & Jordan, 2010). The macro level simulation modeled fish reproduction and carrying capacity. Water quality was a black box in this simulation. This created a problem of understanding that motivated the use of the microlevel simulation. The microelevel simulation modeled nitrification in the aquarium and opened up the black box, allowing learners to explore the relationships between fish population, bacterial populations, and nitrification. A continuing challenge is to encourage learners to make connections between the virtual world and the real world of the aquarium and to take the lessons learned from a model system out into the world. One approach we are currently exploring is using a sequence of curricular units that move from the closed aquarium system to increasingly open ecosystems such as ponds and estuaries. The use of a new tool, the Aquarium Construction Toolkit (Vattam et al., 2009), helps make the SBF conceptual representation explicit. A variety of other challenges remain such as the tension between inquiry and content and teacher's understanding of both inquiry and content.
Goel, A. K., Gomez de Silva Garza, A., GruŽ, N., Murdock, J. W., Recker, M. M., & Govinderaj, T. (1996). Towards designing learning environments -I: Exploring how devices work. In C. Fraisson, G. Gauthier & A. Lesgold (Eds.), Intelligent Tutoring Systems: Lecture notes in computer science. NY: Springer.
Hmelo-Silver, C. E.. Liu, L., Gray, S., & Jordan, R. (2010). Using Representational Tools to Learn About Complex Systems: A Tale of Two Classrooms. Manuscript submitted for publication.
Hmelo-Silver, C. E., Marathe, S., & Liu, L. (2007). Fish swim, rocks sit, and lungs breathe: Expert-novice understanding of complex systems. Journal of the Learning Sciences, 16, 307-331.
Liu, L., & Hmelo-Silver, C. E. (2009). Promoting complex systems learning through the use of conceptual representations in hypermedia. Journal of Research in Science Teaching, 46, 1023-1040.
Vattam, S., Goel, A. K., Rugaber, S., Hmelo-Silver, C. E., & Jordan, R. (2009). From conceptual models to agent-based simulations: Why and how Paper presented at the Fourteenth International Conference on AI in Education, Brighton UK.
Wilensky, U., & Reisman, K. (2006). Thinking like a wolf, sheep, or firefly: Learning biology through constructing and testing computational theories – an embodied modeling approach. Cognition and Instruction, 171-210.