Our ability to transfer motor skills across tools and contexts is central to how we adapt to an increasingly technological world. From smartphones to wearable robotics, modern technologies demand flexible motor control that generalizes beyond specific learning contexts. Robotic limbs for body augmentation are designed to increase the body's degrees of freedom while coordinating with our biological limbs.1,2,3,4,5,6,7,8 While these devices perform well in controlled settings, their real-world effectiveness remains uncertain.9,10 This raises a fundamental question: can the brain form flexible motor representations of extra robotic limbs that support skill generalization? We tested whether motor augmentation learning relies on body-specific or body-independent sensorimotor mappings. While body-dependent learning may be more intuitive, it tends to be context-bound and thus limited in its transferability.11,12,13,14,15,16 Learning based on higher-level, body-independent representations may promote flexibility17,18 but can increase cognitive demands. Participants trained for 7 days to use an extra robotic thumb (the Third Thumb, Dani Clode Design), worn on the right hand and controlled by the toes. They showed broad skill generalization, spanning tasks, body postures, and body parts used to wear/control the device, indicating the development of flexible, body-independent motor representations. Training also reduced cognitive demands and enhanced the sense of agency. The toe-control mechanism mildly affected balance performance, reflecting behavioral trade-offs. Yet, this broad generalization did not predict participants' inclination to use the device when given a choice, suggesting that factors beyond skill generalization, cognitive effort, and embodiment must be addressed to support the real-world adoption of these technologies.
Molina-Sanchez et al. (Sun,) studied this question.