Motor imagery (MI) is widely used across various domains, including neurorehabilitation, sports training, yet many individuals struggle with imagery vividness, a limitation known as MI ambiguity. To address this, recent AI advances allow for highly pe...
Motor imagery (MI) is widely used across various domains, including neurorehabilitation, sports training, yet many individuals struggle with imagery vividness, a limitation known as MI ambiguity. To address this, recent AI advances allow for highly personalized virtual hands (P-VHI) that induce body ownership. This study developed an AI-based P-VHI system and compared its efficacy in inducing body ownership and MI engagement against generic (G-VHI) and customized (C-VHI) versions.
A preliminary study confirmed that P-VHI generated from a single 2D image significantly induced subjective body ownership (questionnaire-based) and induced neural activation related to body ownership in the premotor cortex, establishing the groundwork for further research.
The main study employed an upgraded system featuring full 3D reconstruction and automatic rigging, which enabled natural motion generation and texture interpolation. Results indicated that P-VHI yielded the fastest illusion time and the most frequent first-place rankings. During the VHI induction phase, EEG analyses focusing on ownership-related frontal channels (F3, Fz, and F4) revealed critical hemispheric differences. While all conditions activated left frontal regions, P-VHI showed a distinct pattern with marginally significant, strongest activation in the right-hemisphere (F4), an area linked to visual self-recognition processing.
In the MI phase, P-VHI produced the strongest C3 alpha ERD, which is related to MI enhancement. Rank-correlation analyses demonstrated that ERD was positively associated with right-hemisphere activation, whereas left-frontal channels showed negative correlations. These findings suggest that MI enhancement is driven by right-hemisphere networks that integrate body ownership with self-appearance processing. Overall, this research provides empirical evidence that AI-based P-VHI induces neural activation related to body ownership and self-recognition, enhancing ERD during MI, offering a solid foundation for various future applications.