Investigating student misconceptions and alternative representations is a longstanding and important concern in science education (Resbiantoro & Setiani, 2022). These misconceptions, particularly early in the learning process, often reflect persistent ontological commitments that are difficult to shift (Chi, 2005). Their identification and remediation are not only central to conceptual development, but also offers a valuable lens into learners’ sense-making processes (DiSessa, 1993).

This study investigates how confronting learners with their own misconceptions, visually and explicitly, can foster epistemic agency and support conceptual change. It draws from conceptual change theory (Posner et al., 1982) as well as affective and metacognitive models of learning (Duit & Treagust, 2003), situating cognitive conflict within a broader perspective that includes emotional discomfort, uncertainty tolerance, and learner empowerment. While traditional approaches to conceptual change often remain invisible or instructor-led, this study leverages an interactive tool to make learners’ knowledge structures explicit and manipulable by themselves.

Building on work in science education and learning analytics (Viberg et al., 2018), I propose a new method that transforms learner explanations into network graphs, capturing not just isolated misconceptions but structural patterns of understanding. These conceptual networks are then projected onto a disciplinary reference map constructed by experts, enabling a precise visual comparison. The resulting discrepancies reveal which concepts are misused, missing, or incorrectly related.

Importantly, this confrontation is not framed as evaluative, but as exploratory. Learners are encouraged to reflect on the visual feedback, consider possible revisions, and negotiate meaning. This process aims to support the emergence of epistemic agency (Elgin, 2013), operationalised as the learner’s capacity to reflect, take ownership of, and revise their own understanding. In doing so, this study shifts the focus from “fixing” misconceptions to recognising them as productive entry points for the development of conceptual understanding.

This study therefore addresses three intertwined goals: (1) to develop new methods for identifying and visualising learners’ conceptual structures; (2) to study how learners respond when confronted with these structures; and (3) to explore how this process can support not only conceptual change but also a more agentic relationship to learning in the natural sciences.

The study combines network-based analysis of knowledge with qualitative interviews and observational protocols. At its core is a newly developed web interface that enables learners to interact with their own conceptual structures in a 3D environment (Sherbetjian, 2026). This interface visualises graph networks derived from written or oral explanations of cloud physics, a multidisciplinary topic rich in abstract scientific processes and common misconceptions (Weihs, 2026). Concept networks are constructed using a method presented by Weihs et al. (2025), in which nodes represent scientific concepts and directed edges represent causal, temporal, or logical relations inferred from learner explanations. These learner networks are then compared with a validated disciplinary reference graph built from several expert inputs. Using visual overlays, the interface highlights missing links, conceptual inconsistencies, and peripheral versus central ideas. Network metrics (e.g., degree, betweenness, eigenvector centrality) help identify influential or problematic nodes, including those that often serve as misconception anchors. Learners are invited to explore these visualisations during individual sessions, in which they can zoom, rotate, and interrogate nodes and connections. During these sessions, a concurrent think-aloud protocol (Cowan, 2017) is used to capture learner responses, reflections, and emotional reactions. These sessions are followed by semi-structured interviews that provide additional insight into learners’ evolving understanding and their perception of the tool. By linking graph-based diagnostics with learner introspection, this methodology makes it possible to study not only what learners know, but how they engage with representations of their own knowledge. The combination of network-based quantitative and reflective qualitative data offers a multimodal picture of conceptual change in action.

This study is expected to yield several interrelated findings. First, it will provide empirical evidence for how network-based concept visualisations can reveal the structure of learner understanding in complex scientific domains. These structural insights can offer a more nuanced diagnosis of learning difficulties than conventional assessments. Second, by confronting learners with their own conceptual networks and inviting reflective engagement, the study investigates how this approach can foster epistemic agency. Preliminary data suggest that learners respond not only with surprise or discomfort, but often with curiosity and a desire to revise and clarify their thinking. This supports theoretical models of conceptual change that emphasise the learner’s active role in navigating cognitive conflict (Chi, 2005; Duit & Treagust, 2003). Third, the findings will inform pedagogical practices in science education, particularly the design of tools that integrate diagnostic feedback with opportunities for learner control. By making misconceptions visible and interactive, such tools can shift the classroom discourse from correction to inquiry. Ultimately, the project contributes to a growing body of research that uses network analysis to study conceptual development, and extends it by embedding these methods in learner-active environments designed to cultivate agency, instead of the historical focus on epistemological accuracy. In doing so, it bridges the gap between learning analytics, science education, and the design of transformative learning experiences.

Chi, M. T. H. (2005). Commonsense Conceptions of Emergent Processes: Why Some Misconceptions Are Robust. Journal of the Learning Sciences, 14(2), 161–199. Cowan, J. (2017). The potential of cognitive think-aloud protocols for educational action-research. Active Learning in Higher Education, 20(3), 219–232. DiSessa, A. A. (1993). Toward an epistemology of physics. Cognition and Instruction, 10(2–3), 105–225. Duit, R., & Treagust, D. F. (2003). Conceptual change: A powerful framework for improving science teaching and learning. International Journal of Science Education, 25(6), 671–688. Elgin, C. Z. (2013). Epistemic agency. Theory and Research in Education, 11(2), 135–152. Posner, G. J., Strike, K. A., Hewson, P. W., & Gertzog, W. A. (1982). Accommodation of a scientific conception: Toward a theory of conceptual change. Science Education, 66(2), 211–227. Resbiantoro, G., & Setiani, R. (2022). A Review of Misconception in Physics: The Diagnosis, Causes, and Remediation. Journal of Turkish Science Education, 19(2), 403–427. Sherbetjian, S. (2026). 3D Visualisations of Complex Graph Networks on a Web Platform. Bachelor thesis. Lucerne University of Applied Sciences and Arts. Viberg, O., Hatakka, M., Bälter, O., & Mavroudi, A. (2018). The current landscape of learning analytics in higher education. Computers in Human Behavior, 89, 98–110. Weihs, J.-P., Gjerde, V., & Drange, H. (2025). From Novice to Expert in Cloud Physics: A Graph-Based Analysis of Learner Conceptual Understanding. Weihs, J.-P. (2026). Investigating Understanding, Knowledge, and Conceptual Difficulties Using Graph Network Analysis: A Study of Learners and Epistemologies in Cloud Physics. PhD thesis. University of Bergen. Weihs, J.-P. (2026). Investigating understanding, knowledge, and conceptual difficulties using graph network analysis: A study of learners and epistemologies in cloud physics. PhD thesis. University of Bergen.