🧬 biology

Voluntary Action Facilitates Self-Other Discrimination but Leaves Metacognition Unchanged

This study demonstrates that while voluntary action enhances perceptual sensitivity in self-other discrimination, it does not correspondingly improve metacognitive monitoring, suggesting these two processes rely on partially independent mechanisms.

Original authors: Kazuma Takada, Wen Wen, Shunichi Kasahara, Tom Froese

Published 2026-09-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Kazuma Takada, Wen Wen, Shunichi Kasahara, Tom Froese

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Every day, we navigate a world where some things happen because we do them, and others happen despite us. When you reach for a cup and it moves toward your hand, you feel a sense of agency, a quiet certainty that you are the cause. But if that same cup slides across the table on its own, your brain must quickly decide that you are not the one moving it. This ability to tell the difference between self-made actions and those caused by others is the foundation of how we experience control. It is a skill we use constantly, from driving a car to having a conversation, and it is so fundamental that when it breaks down, as it can in certain mental health conditions, our sense of reality can become unstable. Scientists have long known that the brain uses two main ways to figure out what is happening. One way relies on a prediction: before you move, your brain sends a signal to your muscles and simultaneously sends a copy of that signal to your sensory centers to say, "I am about to do this, so expect to feel this." If the feeling matches the prediction, you know it is you. The other way is simpler; it looks for patterns. Even without a perfect prediction, if your hand moves and a light on the screen moves at the same time, your brain can learn that connection through repetition.

A team of researchers at the Okinawa Institute of Science and Technology and Rikkyo University set out to see if these two ways of knowing are also different when it comes to how confident we feel about our judgments. They wanted to know if the brain's ability to predict its own actions makes us not only better at spotting the difference between self and other, but also better at knowing that we are right. To test this, they built a device that could move a person's hand for them. In one set of trials, participants moved a handle with their own hand, and in another, the machine moved the handle for them, forcing their hand along a path they did not choose. While their hand moved, two shapes on a screen—a square and a circle—wobbled around. One shape moved in sync with the hand, while the other moved randomly. The participants had to pick which shape was connected to their hand and then rate how sure they were about their choice.

The results showed a clear split between what the participants could do and how they felt about what they could do. When people moved their own hands, they were significantly better at identifying the correct shape than when their hands were moved for them. This suggests that having the ability to predict your own movement gives your brain a sharper tool for distinguishing your actions from the world around you. However, when the researchers looked at the confidence ratings, the story changed. While voluntary movers were indeed more confident than involuntary movers at higher levels of control (60% and 70%), this advantage did not translate into better metacognitive sensitivity. The ability to judge how well one is doing did not improve just because the person was in control. Instead, metacognitive sensitivity showed no comparable group-level difference between the two conditions and exhibited substantial individual variability, with some people being very sure of their answers and others less so, regardless of whether they were moving or being moved.

This finding points to a fascinating separation in how our brains work. It appears that the mechanism that helps us detect agency—the feeling of being in control—can be sharpened by voluntary action, but the mechanism that monitors how reliable that feeling is operates differently. The researchers suggest that while the brain uses its predictive power to make the initial judgment of control more precise, the part of the brain that checks the quality of that judgment does not seem to rely on the same prediction signals. Instead, it may depend on other factors that remain constant whether we are moving or being moved. The study did not find evidence that being in control makes us better at knowing we are in control. It simply makes us better at knowing that we are in control. This distinction helps explain why we can sometimes feel certain about our actions even when our perception is flawed, or why we might doubt our agency even when we are acting freely. The work highlights that the sense of agency is not a single, unified feeling, but a complex interplay of different processes, where the ability to act and the ability to monitor that action can be enhanced independently.

Technical Summary: Voluntary Action Facilitates Self-Other Discrimination but Leaves Metacognition Unchanged

Problem Statement
The sense of agency—the subjective feeling of controlling environmental changes—is fundamental to motor learning, decision-making, and social interaction. While the "comparator model" suggests that agency arises from comparing sensory predictions (derived from motor commands) with actual feedback, alternative "regularity detection" processes may also underpin agency when predictions are unavailable or inaccurate. A critical gap in the literature is whether the perceptual processes underlying self-other discrimination (distinguishing self-caused from other-caused events) are distinct from the metacognitive processes that monitor the reliability of these judgments. Specifically, it remains unclear if voluntary actions, which provide sensory predictions via efference copies, facilitate not only the accuracy of discrimination but also the conscious monitoring (metacognition) of that discrimination compared to involuntary actions, which lack such predictions.

Methodology
The authors conducted two experiments utilizing a custom-built two-dimensional manipulator driven by DC motors. This device allowed for the precise execution of both voluntary movements (initiated by the participant) and involuntary movements (imposed by the device on the participant's hand).

Task: Participants performed a control detection task where two visual stimuli (a square and a circle) moved on a screen. The movement of the "target" stimulus was a blend of the participant's real-time hand movement and a pre-recorded trajectory, while the "distractor" stimulus moved solely based on a pre-recorded trajectory. Participants had to identify which shape was associated with their hand movement and rate their confidence (1–4 scale).
Experimental Design:
- Experiment 1: Employed a within-participant design with two movement conditions (voluntary vs. involuntary) crossed with three levels of control (50%, 60%, 70%). This design aimed to assess perceptual sensitivity ( $d'$ ) and metacognitive sensitivity ($meta-d'$) across varying difficulty levels.
- Experiment 2: Utilized a staircase procedure (2-down/1-up algorithm) to adjust the level of control dynamically. The goal was to equate perceptual performance (accuracy) between voluntary and involuntary conditions, thereby isolating differences in metacognitive sensitivity ($meta-d'$) and efficiency ( $m$ -ratio) independent of task difficulty.
Analysis: The study employed Signal Detection Theory (SDT) and hierarchical Bayesian modeling (using the hmetad package) to estimate $d'$ , $meta-d'$, and the $m$ -ratio ($meta-d'/d'$). This approach was chosen to stabilize estimates given the trial counts per condition.

Key Results

Perceptual Sensitivity ( $d'$ ): Across both experiments, voluntary movement consistently yielded higher perceptual sensitivity than involuntary movement. In Experiment 1, accuracy was significantly higher in the voluntary condition at all control levels. In Experiment 2, participants required a lower level of control (i.e., less self-motion signal) in the voluntary condition to achieve the same accuracy as in the involuntary condition, indicating that voluntary actions provide additional cues (likely sensory predictions) that enhance discrimination.
Metacognitive Sensitivity ($meta-d'$) and Efficiency ( $m$ -ratio): Despite the clear advantage in perceptual sensitivity for voluntary actions, no credible group-level differences were found between voluntary and involuntary conditions for metacognitive sensitivity or efficiency.
- In Experiment 1 (50% control level), the posterior distributions for $meta-d'$ and $m$ -ratio overlapped significantly between conditions.
- In Experiment 2, where perceptual performance was equated, $meta-d'$ and $m$ -ratio remained comparable between voluntary and involuntary conditions.
Individual Variability: While group-level metacognitive measures were unchanged, there was substantial individual variability. Some participants showed higher metacognitive sensitivity in voluntary conditions, while others showed the opposite, resulting in no net group difference.
Correlations: Individual differences in perceptual sensitivity and metacognitive sensitivity were positively correlated across conditions (participants who performed well in one condition tended to perform well in the other), suggesting consistent individual traits in both domains. However, in Experiment 2, the level of control (a proxy for perceptual difficulty) did not correlate with metacognitive measures, contrasting with some previous findings regarding prediction-error-based processes.

Significance and Claims
The paper claims to demonstrate a dissociation between perceptual processes and metacognitive monitoring in the context of self-other discrimination.

Perceptual Advantage of Volition: Voluntary action enhances the perceptual basis of agency judgments, likely by enabling the use of prediction-error-based processes in addition to regularity detection.
Independence of Metacognition: The ability to consciously monitor the reliability of agency judgments (metacognition) does not appear to depend on the presence of voluntary motor commands or the associated sensory predictions. The authors suggest that metacognitive monitoring may rely on at least partially independent mechanisms that are accessible regardless of whether the action is voluntary or involuntary.
Mechanistic Distinction: The findings support the view that while the generation of the sense of agency may differ between voluntary (prediction-based) and involuntary (regularity-based) conditions, the monitoring of these judgments may draw upon common metacognitive resources.

The authors maintain a modest tone, acknowledging that the lack of group-level metacognitive differences could reflect opposing individual tendencies rather than a complete absence of condition effects. They also note limitations regarding trial numbers and the potential variability in task difficulty introduced by the blending procedure, suggesting that future work should explore a broader range of metacognitive measures.

Technical Summary: Voluntary Action Facilitates Self-Other Discrimination but Leaves Metacognition Unchanged

More like this