The study of students' preconceptions and how they affect their learning in science, technology, engineering and mathematics (STEM) fields is of nationally recognized importance. There are, however, various and contradictory theoretical approaches to conceptual change, and none of them have been rigorously applied in the context of engineering education. This paper is part of a larger study drawing on existing sets of data from a wide range of engineering content areas to develop a theoretical explanation of conceptual change in engineering education. In the work reported here we re-analyze students' understanding of concepts about axially loaded members (from mechanics of materials) and Boolean logic (from digital logic). Previously published analyses of these data argue that the context of a problem or question effects students' reasoning about that concept. These contexts can range from the presence or absence of figures or diagrams to the social contexts of the problem. We explored three potential theoretical explanations for the context-sensitivity of student reasoning: (1) a perceptual cues theory, (2) a domain specificity theory, and (3) a language-based theory. It is argued that these competing theoretical explanations do not contradict each other as much as they overlap, and potentially productive syntheses of the theories are proposed as directions for future work.