Parsing Patterns: Developing Metrics to Characterize Spatial Problem-Solving Strategies Within an Environmental Science Simulation

• In Event: Emergent Methods for Studying Spatial and Embodied Dimensions of Learning

Abstract

Spatio-temporal methods for studying learning are particularly useful when the content being learned is itself a phenomenon that unfolds over space and time, as is true with topics like environmental science and urban planning. Many of the current challenges we face as a society occur when human systems (e.g., settlements, roads, infrastructure) interact with natural systems (e.g., groundwater, habitats). The extent and nature of these interactions are dependent on the spatial patterns found within the human systems, the patterns within natural systems, and the intersections of these patterns. Urban planning and environmental science practice involves understanding and solving problems that arise from these intersections, but current educational practices delay opportunities to grapple with problems until well into graduate-level studies. Our research project has developed an interface (EcoCollage) that allows learners (undergraduate and AP high school students) to recognize these patterns and mindfully rearrange them to affect emergent outcomes. The context provided is the challenge of integrating green infrastructure (e.g., native vegetation swales) within an urban environment, where the task is to balance the needs of humans with the sustainability of environmental resources.

We have augmented an urban storm water management simulation developed for the Illinois EPA with a computer vision input system to support a paper-based Tangible User Interface (TUI). TUIs may confer special advantages for spatial problem-solving (Antle, Droumeva, & Ha, 2009), and our paper-based approach is also designed to be cost- and time-effective for schools, as it allows students to interact with complex system simulations without requiring computer programming expertise or multiple computers. Rather, students solve environmental science problems through the hands-on placement of paper tiles (representing swales) on a large map, which mimics authentic planning practice. The paper map is then read and interpreted as input for the simulation so that students can test how each pattern affects the urban ecosystem.

By observing student interaction and problem-solving as they used EcoCollage, we qualitatively witnessed how learners developed ad-hoc spatial problem-solving strategies. However, we soon realized that there were not very many quantitative approaches for documenting these strategies and their evolution. The task required that learners be sensitive to both univariate patterns (the location of swales vis-a-vis other swales) as well as to multivariate patterns (the location of swales vis-a-vis man-made infrastructure elements like impermeable road surfaces and sewers). We found that the ontologies of “spatial knowledge” that have been developed to describe how students learn geospatial concepts (Marsh, Golledge, & Battersby, 2007) were inadequate for characterizing the sophistication of student patterns, but the field of plant ecology has a rich array of spatial analytic techniques for detecting spatial inter-relationships, like clustering and over-dispersion, that can be used to characterize swale placement strategies (Dale, 2004). We validated the utility of these metrics by reconciling the descriptive metrics generated for each map with the categorical judgments made by urban planning and ecology experts. We are developing this analytic approach into a method for quantitatively characterizing student solutions to spatial problems and the evolution of their spatial problem-solving strategies.

Authors

• Leilah Lyons, University of Illinois at Chicago

• Emily Minor, University of Illinois at Chicago

• Moira Zellner, University of Illinois at Chicago

• Brian Slattery, University of Illinois at Chicago

• Tia Shelley, University of Illinois at Chicago