Hypothesis Testing Governs an Efficiency-Flexibility Trade-off in Strategic Motor Learning

It remains unknown how people discover an effective movement strategy when the environment changes (e.g., when adapting to a new computer trackpad). We propose that strategic adaptation operates through hypothesis testing: learners generate candidate hypotheses, discard those inconsistent with feedback, and iteratively refine their actions through practice. A core prediction of this account is an efficiency-flexibility trade-off. In constrained environments, where few hypotheses are viable, learning slows as people eliminate competing hypotheses but supports broader generalization. In unconstrained environments, where many hypotheses are viable, learning accelerates as learners adopt one of many expedient hypotheses but yields poorer generalization. In two reaching experiments (N = 560), we varied the arrangement of target positions to manipulate how tightly the hypothesis space was constrained. As predicted, the constrained group learned more slowly but generalized more--an efficiency- flexibility trade-off that highlights hypothesis testing as a novel process governing human motor learning.