A pilot study on the effects of a physio-avatar EB experience on motor performance

Neurorehabilitation using virtual reality (VR) and electromyography has garnered attention in recent years. The authors developed a system based on these technologies that enables participants to experience an avatar (physio-avatar EMG biofeedback, PAEB) with arbitrary dynamics in VR environments, controlled by their own EMG signals. Using this system, the study examined whether experiencing an avatar that feels slower than the real body can promote the relearning of cerebellar internal models and improve motor performance.

Experiments were conducted in 30 healthy adults. Participants were assigned to four groups: the PAEB adaptation group (n = 6), which underwent prolonged PAEB training; the PAEB nonadaptation group (n = 7), which received only short-term PAEB exposure; the control group (n = 8), which experienced a real-body avatar (RBA) reflecting actual movements without EMG control; and the resistance training (RT) control group (n = 9), which performed isometric contractions comparable to the PAEB adaptation group without visual avatar feedback. Motor performance was evaluated based on angular velocity during a real-body reaching task. Each group performed the task before and after the adaptation phase to assess the effects of PAEB exposure. Subjective questionnaires were administered to evaluate the sense of ownership (SoO) and sense of agency (SoA).

Subject-level one-tailed Mann-Whitney U $\left(\alpha =0.05\right)$ indicated significant improvements in motor performance in five of six participants in the PAEB adaptation group. In contrast, no significant improvements were observed in the other groups, except for one participant in the PAEB nonadaptation group and one participant in the control group and one participant in the RT control group. Notably, no participant in the RT control group showed a significant improvement; instead, most exhibited a decrease in angular velocity. For within-group comparisons, the one-tailed Wilcoxon signed-rank test $\left(\alpha =0.05\right)$ showed that only the PAEB adaptation group demonstrated a substantial increase (25.4%) in angular velocity from baseline to post-adaptation. Questionnaire results further indicated that only the PAEB adaptation group showed a tendency toward increased SoA with repeated PAEB exposure.

The improvement in motor performance observed in the PAEB adaptation group suggests that exposure to an avatar with altered bodily dynamics modifies motor control strategies through error-driven updating of internal models. This interpretation is supported by the RT control group, which showed no improvement despite experiencing comparable physical load. These findings indicate that the enhancement in motor speed is more likely attributable to internal model recalibration via error learning than to acute neuromuscular responses to physical training.

These results indicate that experiencing the PAEB-an avatar that appears to move slower than one's own body-can potentially improve motor performance. However, as the number of participants who underwent PAEB training was limited, further studies with a larger sample size are required to confirm these findings.