Bold claim: our sense of how our bodies move can be as much a hurdle as a helper when learning new skills, and robotic prosthetics reveal this paradox in a striking way. A new study shows that the way people perceive their own movement changes differently when they learn to walk with a robotic leg, compared to how they perceive normal walking.
“ When people first start walking with a prosthetic leg, they think their bodies are moving more awkwardly than they really are, ” explains Helen Huang, the study’s corresponding author. “ With practice, as performance improves, people still misjudge their movements, but the nature of the inaccuracy shifts in a fundamental way. ” Huang is the Jackson Family Distinguished Professor of Biomedical Engineering at North Carolina State University and the University of North Carolina at Chapel Hill.
The idea behind this research is that everyone carries a personal body image—a mental map of how our bodies are built, how they move, and how they should feel while moving. That internal model helps us execute new physical skills, like dancing or catching a ball, but it’s often not perfectly aligned with our actual movements. Over time, this mental image converges toward reality, and our performance improves.
The researchers asked whether people who use robotic prosthetics adjust their body image to include the prosthetic device itself, and how this evolves with practice. They also explored whether linking the device more closely with one’s internal body map relates to better use of the prosthetic.
In the experiment, nine able-bodied volunteers practiced walking on a treadmill while attached to a knee-ready robotic prosthesis set at a right angle. Each day, participants repeated the task and, after practice, reviewed a computer animation showing a range of possible walking gaits. They then picked the gait that most closely matched their recent performance with the device.
Early on, participants felt their gait was off-balance and stiff, more than was actually the case. After four days, they perceived their gait as smoother and more natural than it truly was, even as their actual performance improved significantly. In other words, their confidence grew even as their accuracy remained imperfect.
One notable finding was that participants tended to focus on the torso’s position when judging their own gait, paying less attention to how the prosthetic device behaved. This likely reflects the limited direct feedback they receive about the device’s actions, since they can’t see themselves move with the prosthesis in real time. Huang suggests that adding visual or other feedback during training could help calibrate the body image and improve gait with the prosthetic.
Another important takeaway is the overconfidence phenomenon. If people believe they’re already doing well, they may be less motivated to put in the effort needed to improve, even when there is substantial room for advancement. The team advocates developing ways to provide users with a clearer, more accurate assessment of how their bodies are really moving during prosthetic training.
The findings appear in the open-access journal PNAS Nexus under the paper “ Projecting the New Body: How Body Image Evolves During Learning to Walk with a Wearable Robot, ” with I-Chieh Lee as first author and contributions from NC State colleagues Huan Min and Ming Liu. The research was supported by NIH grant R01HD110519 and NSF grant 2211739.
Controversy aside, these results open doors to better prosthetic training by aligning feedback, body perception, and device control. Do you think real-time visual feedback or sensor-based cues during prosthetic training would meaningfully boost learning, or could they overwhelm users with information? Share your thoughts below.
