This thesis expands and details the resources model of cognition through the use of two metaphors, maps and graphs. Maps and graphs, complementary representations of the resources model, provide organizational structure and illustrate core properties of the model. Maps show which resources are relevant to a given situation and graphs show how those resources can be connected to each other.

A nearly-novel situation is one that forces students into an area outside of established conceptions, but near many resources. Being outside of established conceptions forces students to make sense on-the-fly, while being near many resources means students have opportunities to build graphs by linking resources together to help make sense of a new situation.

The physics of diode design is an ideal nearly-novel situation in which to study epistemology and metacognition in upper-level physics students: rich in physics ideas, not mathematically complex, and understudied by the population. Through clinical interviews and an iterative survey, students were questioned about the functions of diodes in circuits, then asked to design a diode given a charge source. Following their design, students were asked questions intended to probe their previous studies of diodes and to suggest possible effects to consider in a redesign of their diodes. Students could then redesign their diode.

Nine of twenty-five respondents were incapable of designing a diode. Nondesigners usually indicated they could not remember how, despite having never studied diode construction. Epistemologically, these students appear to use knowledge-as-rememberable to the exclusion of knowledge-as-derivable in this context.

We find two constraints on successful reasoning in nearly-novel situations. To see a situation as nearly-novel, students must both be familiar with and see the necessary material as relevant. Furthermore, the epistemological resource knowledge-as-derivable must not be blocked from activating.