In the ever-evolving landscape of neuroscience, the sense of agency (SoA) remains a captivating domain — the subtle feeling that "I am the one causing this action or effect." Recent groundbreaking research has now unpacked how this fundamental sensation is shaped through learning new motor skills. A study led by Tanaka and Imamizu, published in Communications Psychology (2025), offers unprecedented insights into the intricate processes underlying the emergence of SoA. Their findings illuminate how humans transform from relying on simple temporal cues to developing sophisticated internal models that endow us with the ability to predict and attribute agency to our own actions.
At the heart of their research lies the distinction between two crucial stages in SoA processing. Early in learning, individuals primarily depend on temporal contiguity—the close timing between performing an action and observing its outcome. This rudimentary reliance suggests that the brain initially uses temporal proximity as a proxy for causality: if the effect follows closely after the action, it must be caused by the actor. Yet, as practice continues and mastery develops, this temporal framework gives way to a more refined prediction-based system. The brain begins to utilize internal models tailored to the specific task to anticipate the outcomes of actions, thereby realizing a more robust, reliable sense of agency.
These internal models are particularly notable because the study emphasizes their structural nature, rather than operating by simple tabular mappings of discrete actions and outcomes. Structural models, in this context, refer to rule-based, continuous mappings that encapsulate the dynamics between motor commands and their resulting sensory feedback. Such models are not static lookup tables but dynamic frameworks that generalize across variations and enable fluid, adaptive motor behavior. This conceptualization challenges traditional views and suggests that the brain builds a nuanced representation of action-outcome relationships, critical for developing a stable SoA during motor learning.
To observe this progression, Tanaka and Imamizu employed tasks where participants acquired novel motor skills requiring continuous control rather than discrete, stepwise actions. In doing so, they could differentiate the emergent internal representations guiding behavior. The study reveals that the foundation of the SoA lies in this structural learning: as participants navigate the complex mapping from motor commands to outcomes, their brains forge internal models that transcend immediate feedback and predict results, deepening the subjective experience of agency.
One of the pivotal contributions of this study is the elucidation of the comparator model’s origins—a mainstream theoretical framework in agency research. The comparator model posits that the brain predicts sensory consequences of motor commands and compares them with actual sensory feedback to establish a sense of authorship over actions. However, before Tanaka and Imamizu’s work, the mechanisms constructing this comparator process during learning remained elusive. Their findings suggest that motor exploration is the crucible forging these internal models; through active experimentation and adaptation, individuals develop the predictive capabilities central to the comparator framework and, thus, agency itself.
The implications ripple far beyond theoretical neuroscience, touching on applied domains such as rehabilitation, virtual reality (VR), and brain-machine interfaces (BMI). In rehabilitation contexts, understanding how structural internal models form and relate to SoA can guide tailored therapies designed to restore motor functions with enhanced patient engagement through stronger feelings of control. Similarly, VR systems benefit from optimizing the user’s sense of agency, crucial for immersion and learning in simulated environments where action-outcome mappings may differ from reality. Brain-machine interfaces, which translate neural signals into device commands, also stand to gain from integrating these insights, potentially improving control precision and the user’s sense of ownership over external devices.
This study’s approach foregrounds the significance of motor exploration—not just passive observation—in shaping internal representations. In practice, this means that the brain thrives on active trial-and-error learning, using the continuous interplay of motor commands and sensory feedback to refine its predictive models. The transition from reliance on temporal contiguity to sophisticated internal predictions epitomizes the adaptability of neural systems in building coherent experiences of agency, a process fundamental for interacting with complex environments.
Moreover, the emphasis on continuous, rule-based mappings aligns with broader findings in motor control literature pointing toward the brain’s preference for generalized, scalable representations. Such representations facilitate transfer learning, enabling skills learned in one context to apply flexibly across others. This feature is essential for the fluid adaptation required in ever-changing environments, underpinning human dexterity and versatility.
Beyond the immediate context of SoA in motor learning, these results hint at the integration of multiple sensory modalities within internal models. While this study chiefly addresses motor-to-sensory mappings, the conceptual framework likely extends to understanding agency in multisensory and social contexts. It opens up new avenues for investigating how prediction and comparison mechanisms integrate proprioceptive, visual, and even auditory feedback to establish a coherent sense of self-produced action.
The research also prompts reconsideration of disorders characterized by disrupted agency, such as schizophrenia or certain movement disorders. If the formation of structural internal models is compromised or delayed, it may explain pathological experiences of alien control or diminished feeling of agency over one’s actions. Therapeutic strategies aimed at facilitating motor exploration and reinforcing predictive frameworks might thus hold promise for ameliorating such symptoms.
Technically, the study leverages advanced computational modeling alongside behavioral and neurophysiological measures to dissect the stages of SoA acquisition. By isolating continuous mapping structures and measuring their formation over time, the researchers provide a compelling account of the underlying mechanisms steering agency emergence. This methodological synergy exemplifies the power of integrative neuroscience—melding theory, computation, and empirical data to unravel complex phenomena.
Ultimately, Tanaka and Imamizu’s work enriches the theoretical tapestry of sense of agency, situating it firmly within the context of learning-driven internal model formation. Their insights underscore that agency is not an automatic given but a fragile, evolving construct built atop active motor exploration and sophisticated predictive architectures. This realization sets the stage for a more nuanced understanding of human motor cognition and the profound interplay between action, perception, and self-awareness.
As the field moves forward, this research invites further inquiry into how different types of motor skills—ranging from fine finger movements to gross limb actions—engage distinct internal models and shape agency perceptions. It also raises questions about developmental trajectories: how do children construct these structural internal models, and what role does experience play during critical periods of motor learning? Addressing these questions will deepen our comprehension of agency’s ontogeny and plasticity.
The broader ramifications extend into artificial intelligence and robotics, where embedding principles of human-like internal model formation could yield machines with more naturalistic and adaptable action control. Incorporating continuous, rule-based mappings may help robots better emulate human motor learning and agency, facilitating smoother human-robot interactions.
In conclusion, the emergence of the sense of agency during new motor skill acquisition is a dynamic interplay between temporal relationships and complex internal representations. Through active learning and structural modeling, individuals transcend simple cause-and-effect perceptions, cultivating a predictive framework that fosters robust agency experiences. Tanaka and Imamizu’s seminal study not only advances our theoretical understanding but also lays the groundwork for applied innovations in medicine, technology, and beyond. The journey from motor command to felt agency is now better charted, promising exciting horizons in the neuroscience of action and self.
Subject of Research: The formation of a sense of agency during the acquisition of novel motor skills through structural internal model development.
Article Title: Sense of agency for a new motor skill emerges via the formation of a structural internal model.
Article References:
Tanaka, T., Imamizu, H. Sense of agency for a new motor skill emerges via the formation of a structural internal model. Commun Psychol 3, 70 (2025). https://doi.org/10.1038/s44271-025-00240-7
Image Credits: AI Generated
