Our understanding of the nature of student ideas informs our instructional and research agendas. In this paper, I characterize student ideas in terms of five observable properties (determinacy, coherence, context-dependence, variability, and malleability) and describe how those observable properties correspond to the "misconceptions" and "pieces" models of student reasoning. I then analyze instructional materials and student thinking in a particular topic area (special relativity) in terms of each of those two models. I show that specific instructional strategies reflect specific theoretical orientations, and explore the extent to which observed student behavior corresponds to predictions made by the theoretical models. The analysis suggests that while both the misconceptions and pieces models are flexible enough to accommodate all of the data, some aspects of student thinking seem best described in terms of pieces, and others seem better characterized as misconceptions. The purpose of the analysis is to illustrate the effect of theoretical orientation on instruction, instructional research, and curriculum development.