This project explores the synergy between artificial intelligence (AI) and education by retasking the AI sensitivity to the utility of formal representations for automated reasoning and learning to the analysis and design of external representations for promoting human reasoning and learning. We are exploring the utility of a knowledge representation, called Structure-Behavior-Function (SBF), for helping students learn some of the core ideas related to complex systems. Our basic approach is to focus students’ critical inquiry skills around building SBF models of complex systems, allowing them to express their understanding in a visual SBF knowledge representation language that not only makes explicit some of the targeted core ideas related to complex systems but also serves as a stimulus, scaffold, and coordinator for various learning interactions within and among students. This representation appears to serve well for this purpose, both as a canonical form of explanation in biological phenomenon and as characteristic of expert reasoning about biological systems.
We are designing and evaluating the Aquarium Construction Toolkit (ACT), an interactive learning environment targeted towards middle-school students, which supports learners in building SBF models in the domain of ecosystems, using an aquarium as an example. ACT is envisioned for use in conjunction with a learning approach that integrates model-driven and design-based inquiry. In this approach, students design, build and maintain aquaria in their classrooms. Learners observe different phenomena that need to be explained and pieced together. They document their observations and iteratively construct and refine SBF models that account for those observations.
ACT includes an SBF modeling tool called SBFAuthor. SBFAuthor provides a visual tool for SBF modeling called Visual-SBF. Visual-SBF adds a visual syntax to the SBF language that partitions an SBF model space into three views corresponding to Structure, Behavior and Function. Each view contains a palette of icons that makes salient the crucial elements we want learners to acquire in terms of the SBF ontology. The process of modeling can begin top-down (from the overall system to its lowest-level subsystems; e.g., from aquarium system to fish system to kidney system and so on), bottom-up, or from any intermediate level of abstraction. At each level, students learn about the system, its function, its behavior for achieving that function, and the structures that participate in the behavior.
Once an SBF model has been created in the SBFAuthor, learners can simulate that model in NetLogo, which provides a dynamic visualization of the model in a simulated world. This integration also provides an evolving graph of the values of the different parameters of the model. The simulation can be stopped and resumed at any stage. Once the simulation is completed, the SBF model underlying the simulation can be evaluated by (1) comparing the generated simulation to an expert’s simulation or (2) by comparing the graph obtained from the simulation to the graph obtained from the data collected by observing the real aquarium maintained in the classroom. Any serious differences indicate a gap in the model and in the modeler’s understanding. This can lead to another cycle of inquiry, resulting in the refinement of the SBF model.