Sense of Agency in Closed-Loop Muscle Stimulation

• Published in: IEEE access Vol. 13; pp. 105417 - 105433
• Main Authors: Gehrke, Lukas, Terfurth, Leonie, Gramann, Klaus
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Actuation; Augmentation; brain-computer interface; Closed loops; EEG; Electroencephalography; Environmental management; Feedback control; Hands; Hardware; Human-computer interface; Motors; muscle stimulation; Muscles; Neuromuscular electrical stimulation; Particle measurements; Presses; Propioception; Qualitative analysis; Real time; Reviews; sense of agency; Stimulation
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2025.3579043

To maintain a user's sense of agency (SoA) when using motor augmentation devices, system actuation must align with the user's intention. While stimulus-response paradigms in lab settings allow precise timing, real-world use requires detecting user volition directly. We developed a closed-loop system using a brain-computer interface (BCI) to detect readiness potentials (RPs) from EEG and trigger electrical muscle stimulation (EMS) at moments of volitional intent. The system distinguishes in real-time between idle and pre-movement states, allowing EMS to support finger movements aligned with the user's intention. We evaluated the system in a within-subject user study comparing three conditions: INTENTION (voluntary action), INVOLUNTARY (EMS-triggered without intent), and AUGMENTED (BCI-controlled EMS). We measured classifier performance (mean F1 score = 0.7), intentional binding, subjective control ratings, and collected qualitative interview data. Results showed that AUGMENTED preserved more agency than INVOLUNTARY, though less than INTENTION. Participants described moments of synchronized stimulation as collaborative, while misalignments reduced perceived control. These findings highlight the promise of BCI-controlled EMS for agency-preserving augmentation and identify design challenges for real-world systems.