Gaze strategy and sense of ownership in learning prosthetic control: a comparative study using wearable eye tracking

Prosthetic control requires not only motor execution but also the development of adaptive visual strategies. Myoelectric prostheses provide limited sensory feedback and therefore rely more heavily on visual monitoring. However, learning-related changes in gaze behavior-including fixation patterns and physiological indices such as blink rate-remain underexplored. This study aimed to investigate how gaze behavior changes and the sense of ownership change during the learning of body-powered and myoelectric prosthetic control, and how these effects differ depending on hand dominance.

Thirty-six healthy adults (18 males and 18 females) were randomly assigned to four groups: body-powered prosthesis with dominant hand, body-powered with non-dominant hand, myoelectric with dominant hand, and myoelectric with non-dominant hand. Participants performed a simulated prosthetic control task (Coin Task from the Southampton Hand Assessment Procedure) before and after training. Gaze behavior was recorded at 50 Hz using Tobii Pro Glasses 3. Primary outcomes included gaze fixation percentage, blink rate, task completion time, and self-reported sense of ownership. To enhance the ecological validity of the findings, two participants with upper limb amputation who regularly used prostheses also completed the task using their own devices.

All groups demonstrated reduced task completion time and increased sense of ownership following training (p < .01). The body-powered groups exhibited increased fixation on the target (jar) during the lift phase, suggesting predictive gaze use. In contrast, the myoelectric groups maintained gaze on the hand or object, indicating compensatory strategies. Blink rate did not change significantly. The participants with upper limb amputation also showed high jar fixation and a strong sense of ownership. The participant using a body-powered prosthesis demonstrated a gaze pattern consistent with the predictive gaze observed in able-bodied users of the body-powered simulator, whereas the participants using myoelectric prostheses showed high jar fixation that differed from the hand-centered fixation typically seen in able-bodied myoelectric users. No significant effects of hand dominance were observed in any of the gaze or blink rate measures.

Short-term prosthetic training improved task performance and increased the sense of ownership across all groups. Body-powered and myoelectric prosthesis control showed characteristic differences in gaze strategies; however, these differences did not emerge as significant main effects of prosthesis type. Hand dominance also had no significant effects on gaze or blink-related measures. The findings suggest that training-induced changes, rather than prosthesis type or side of control, primarily shaped gaze behavior during early prosthetic learning. Participants with upper limb amputation demonstrated efficient gaze allocation and a strong sense of ownership, indicating possible adaptations associated with long-term prosthesis use.