Embodied guidance fading: When embodiment supports STEM learning by optimizing cognitive load

Zhang et al.’s (Citation2026) Perform First Hypothesis proposes that learners often benefit from direct sensorimotor action early in expertise development and from observing, imagining, or using abstract representations later. This commentary sharpens that proposal with cognitive load theory. We argue that action helps when it makes concept-relevant relations easier to process, such as when tracing a graph helps novices coordinate slope, variables, and change. Action can also hurt when it adds motor demands, redundant cues, or irrelevant activity without improving understanding. The central instructional aim is not movement or lower load for its own sake, but better learning, transfer, and flexible use of formal representations. We therefore reformulate the Perform First Hypothesis as embodied guidance fading: use the least embodied, task-relevant support needed for the learner’s current knowledge and the task’s element interactivity, then fade that support as schemas develop. This account yields concrete design principles, measurement recommendations, and testable predictions for embodied instruction in STEM.