Embodied pedagogies have repeatedly shown their effec- tiveness, even in STEM domains where content is inher- ently abstract. Multiple studies indicate that these pedago- gies—whether through physical actions, observation, or men- tal imagination—can enhance learning. Sensorimotor simu- lation theory explains that abstract concepts are grounded in concrete, meaningful representations. However, it remains un- clear whether embodied experiences lead learners to actively use visuospatial mental representations. In real learning envi- ronments, prior knowledge levels vary, meaning external em- bodied representations may not affect all learners equally. This study used mediation analyses to examine how learners with different prior knowledge leverage mental representations in problem-solving. Results showed that high-prior-knowledge learners benefited more from the indirect effect, where em- bodied pedagogies enhance learning by fostering visuospa- tial representation use. In contrast, the direct effect—which stands for other mechanisms beyond measured mental sim- ulation through which embodied pedagogies enhance learn- ing—was more pronounced for low-prior-knowledge learners. These findings not only lend further support to the sensorimo- tor simulation account of embodied learning but also highlight how this mechanism differs for learners with different levels of prior knowledge.