From Obstacle to Catalyst: A Narrative Review of Learning from Errors in Science Education

Abstract

Learning from errors (LFE) constitutes a complex, multilevel process whose pedagogical value depends on how errors are elicited, interpreted, and corrected. This narrative review critically synthesizes research on LFE and its implications for science instruction in school and higher education. The reporting of the purposive, non-exhaustive search was strengthened with reference to the Scale for the Assessment of Narrative Review Articles (SANRA) criteria. The synthesis traces the historical transition from error avoidance to conditional error-as-resource perspectives and integrates five complementary lenses: error type, negative knowledge, cognitive load and prior knowledge, affective and self-regulatory processes, and the Generating Errors/Detecting and Correcting Errors model. It then evaluates instructional approaches including productive failure, erroneous examples, peer instruction, formative feedback, conceptual-change strategies, Predict–Observe–Explain, refutation texts, and gamification. Evidence is strongest when errors are embedded in structured cycles of manageable challenge, explanation, feedback, and revision; however, effects are heterogeneous and moderated by learner expertise, task demands, feedback design, classroom error climate, and safety constraints. Domain-specific evidence is most developed in physics, chemistry, biology, and health-professions simulation, whereas Earth and climate sciences and teacher professional development remain comparatively underexamined. The review concludes with an integrative framework and actionable design principles for science teacher education, while identifying methodological, cultural, and empirical gaps for future research.