Model-based learning to conceptualize complex biological systems

Background/Question/Methods Understanding complex systems is fundamental to science and science education, especially in the biological sciences. Unfortunately, the very nature of complex systems also defines the barriers to student learning and understanding. Complex systems are typically characterized as hierarchical and scalar with processes that are frequently dynamic and unseen. As part of the reformation of a large, introductory majors biology course towards a learner-centered pedagogy, we incorporated model-based learning to help students conceptualize and reason about complex biological systems and to provide a method for students to conceptualize complex biological systems. This approach is not without pitfalls. Science students frequently define models as miniature versions of reality rather than as representations of scientific processes. Resulting student models are simple and tend to be linear and literal. Such representations cause students to compartmentalize knowledge and limit a student's capability to understand complex biological systems. We investigated the development of student modeling skills and compared our pedagogical approach to a second section of reformed biology where modeling played a lesser role in the classroom. Results/Conclusions Prior to instruction, students created models of gene transfer as part of a larger unit on genetics. Just over half of initial models were linear box and arrow models, lacking processes and complex interactions (n=103). The remaining student models were evenly divided between literal representations (n=38) and simplified methods lists (n=37). By the time of the first mid-term exam, students were formally introduced to box and arrow models and had several opportunities to generate and critique this type of model. An open-response question on the exam directed students to create a model of their choice to represent and link several genetic concepts. Students in section one overwhelmingly chose box and arrow models (n=166) to represent this system. We measured model complexity as the number of branches and the sophistication of the language used to describe processes. Over half of the models included one or more branches and biologically accurate language (n=109). In contrast, student models in the second section of biology were nearly split between box and arrow modeling (n=100) and pictorial representations (n=85). These results suggest, with instruction, students rapidly migrate from simplistic, linear models to more complex models with one or a few branches. However, student models continue to lack accurate representations of hierarchies and unseen processes. With time and instruction, we expect student models will grow to better incorporate the properties of complex biological systems.