Roadmaps of Biochemical Knowledge: Analyzing Neural Networks of Nucleic Acids and Protein Structure Formed by Undergraduate Chemistry and Biochemistry Students
Undergraduate level biochemistry is an upper division course that can be difficult to understand, as it heavily relies upon visual representations. Interpreting these can exceed the cognitive load for students. The cognitive load theory explains that our preexisting frameworks are used to process and assimilate new information. Only a specific amount of information can be transferred to long term memory at a time. By exceeding this information capacity, learning is hindered as the excess acts as extraneous cognitive load. Improving students’ ability to understand visual representations and to incorporate this with content could aid in assimilating new information into pre-existing frameworks. This will lessen the cognitive load and then lead to increased conceptual understanding. Students form neural networks after observing visual representations, connecting various topics with one another. As of now, there is little research regarding how students interpret biochemistry visuals as well as store the information they decode in their memory. There are also very few tools that allow researchers to assess students’ neural networks, or how they organize the information they extract from visual representations. Recently, a study was conducted in an undergraduate general chemistry course where students ranked the relatedness of words and phrases to measure students' neural networks. This study created a similar instrument that measures students’ organization of nucleic acid information. Rather than using words and phrases, however, students ranked the relatedness of biochemical representations. This study aims to determine if students' neural networks become more expert-like regarding nucleic acid vertical translation visual literacy skills as they progress through a sequential chemistry and biochemistry curriculum. In addition to nucleic acids, a sister survey was deployed for analyzing protein structure representational schemas. We analyzed our data with Pathfinder, a tool that allows us to examine qualitative and quantitative aspects of students' neural networks, relative to an expert referent network. The quantitative data include eccentricity, neighborhood similarities, coherency, degree values, and path length correlations. These values allow us to quantitatively compare student organization of knowledge with expert neural networks. We also qualitatively analyzed images of the neural networks of students by identifying patterns and similarities with expert neural networks to identify chunking of knowledge in relation to the biochemical representation employed. Our findings may allow educators to better develop their courses for optimal student learning and retention of information. Using the data to guide development of visual literacy skills will lead to increased student comprehension as comprehension is tied to understanding visual biochemical representations. Courses can then be designed to alleviate cognitive load arising from these representations, aiming for long term retention.