Neurocomputational Mechanisms of Sense of Agency: A Scoping Review Integrating Predictive Coding, and Optimal Control in Human-Machine Interfaces

Background/Objectives: The sense of agency (SoA)—the feeling of control over one’s actions and their consequences—is a fundamental aspect of volition, learning, and self-awareness. Disruptions in SoA are implicated in various neuropsychiatric conditions, including functional neurological disorders (FND). In emerging human-machine interfaces (HMIs), preserving SoA is critical for usability and therapeutic impact. This scoping review synthesizes computational models that explain the neural mechanisms underlying SoA and explores their application in the design and optimization of HMIs for both rehabilitation and skill learning. Methods: A narrative synthesis of peer-reviewed literature was conducted, focusing on models rooted in predictive coding, Bayesian brain theory, active inference, and linear-quadratic-Gaussian (LQG) optimal control. Simulation studies were also included to illustrate theoretical mechanisms in practical XR-based rehabilitation contexts. Results: The review highlights the role of internal forward models, efference copies, and sensory feedback in the generation and regulation of SoA. It shows how Kalman filter and LQG control frameworks model belief updating and motor planning, explaining disrupted SoA in FND and its restoration via hypnotic suggestion and virtual sensory perturbations (exafference). EEG microstate dynamics and directed brain connectivity studies reveal distinct SoA-related patterns differentiating novice and expert performance in skill learning. Key regions implicated include the supplementary motor area, parietal cortex, cerebellum, and prefrontal cortex, connected via structural pathways. Conclusions: Integrating computational frameworks such as active inference and Kalman filtering with causal reasoning (e.g., Ladder of Causation) offers a powerful lens to understand and modulate SoA with exafference. These insights support the co-design of adaptive XR-based HMI systems for neurorehabilitation and cognitive-motor skill acquisition.