We explore students' sources of self-generated evidence and their meta-cognition when they solve physics and non-physics questions. Our sample comprises 50 freshmen taking or having taken introductory physics. Each student participated in a one-hour interview to complete five open-ended reasoning questions taken from published instruments. Two questions are non-physics that deal with correlation of variables. The three physics questions pertain to the topic of energy. Results indicate that for the non-physics questions, the students mainly used given information in the task as source of evidence. They realized that everyday experiences or assumptions were informal ideas and hence assigned them less weight in generating the evidence. For the physics questions, the students did not realize that they used informal ideas. They packaged informal knowledge in the form of formal physics knowledge. These outcomes may be explained by the ease with which the students dealt with the context of the questions coupled with a high cognitive load associated with processing multiple pieces of information when students tackled qualitative questions.