Abstract

This dissertation comprises a two-part longitudinal inquiry of complex systems thinking as a means of facilitating conceptual change. Study 1 employed a "posttest only control group" random assignment experimental design. Changes in ontological frameworks were assessed using an ontological coding taxonomy (OMMT) adopted from two sources (Ferrari & Chi, 1998; and, Jacobson, 2000) and refined for this study. The hypothesis was that first year Cegep science students receiving a complex systems intervention using StarLogoT would employ more emergent causal explanations and fewer clockwork explanations as a general explanatory framework for problem solving. By contrast, students in the control group would not. The 25 students in the experimental group generated significantly more emergent framework explanations on both near and moderate far transfer questions than did the 20 students in the control group. Furthermore, they generated significantly fewer clockwork framework explanations on near transfer questions but not on moderate or far transfer questions than did the control students. Study 2 was a mixed method qualitative case study of nine students selected from the participants in Study 1 using a purposeful sampling procedure. Students' acquisition of an emergent causal framework was assessed using an ontological measure referred to as the Complex Systems Taxonomy (CST) adopted from Jacobson (2000) and refined for this research. The two central research questions were the following: (1) What aspects of students' ontological and epistemological beliefs facilitated or constrained their acquisition of an emergent causal framework? (2) What experiences with StarLogoT facilitated or constrained this learning process? The findings were as follows: (1) Although students experienced gains in four of the six component features of emergent causal processes, their difficulty with the concepts of "random actions" of agents and "nonlinear effects" of agents constrained their deeper understanding of emergent causal processes. (2) Although StarLogoT facilitated the acquisition of certain aspects of this knowledge, it provided no affordance for learning the concept of "non linearity". Furthermore, aspects of these multi-agents representations generated conflicting ontological explanations for the concept of "randomness". (3) Although the selected StarLogoT simulations demonstrated emergent causal processes, they represented different types of complex systems (i.e., tightly coupled and dissipative loosely coupled). Although most students had difficulty with the representations of dissipative systems, those who had a more advanced understanding of science concepts gained an understanding of emergent causal processes from dissipative representations. (4) Conceptual change required metacognitive scaffolding and ongoing metaconceptual prompts during the instructional phase. However, once students acquired synthetic mental models, maturation over time and experience with complementary domain curricula was sufficient for them to elaborate their understanding of emergent causal processes.