Effect of External Force on Agency in Physical Human-Machine Interaction

Satoshi Endo, Jakob Fröhner, Selma Musić, Sandra Hirche, Philipp BeckerleFrontiers in Human Neuroscience 2020 · DOI

Abstract

As intelligent robotic tools increasingly assist humans in physical tasks, the intuitiveness of control designs has become a paramount feature. However, subjective user experience is difficult to quantify using traditional performance metrics. This study investigates the Sense of Agency (SoA)—the feeling of control over one's actions—in the context of physical human-machine interaction (pHMI). By adapting the "intentional binding" paradigm to a virtual reaching task, researchers evaluated how assistive forces from a robotic device modulate a user's sense of control. The findings reveal that appropriate robotic assistance can actually enhance a user's sense of agency, even when their physical contribution decreases. This suggests that agency is a viable objective metric for evaluating and optimizing human-in-the-loop control systems.

Introduction: The Challenge of Intuitiveness

In the development of powered exoskeletons and robotic rehabilitation tools, engineers face a critical hurdle: making the interaction feel natural. While it is straightforward to measure a robot's speed or accuracy, measuring the user's internal experience is far more complex.

Central to this experience is the Sense of Agency (SoA). In pHMI, a robot often applies forces directly to a user's body. If these forces do not align with the user's intent, the sense of agency is disrupted, often leading to user frustration. Historically, SoA has been measured through subjective questionnaires, which are prone to bias and difficult to integrate into real-time control optimization. This study explores an objective alternative: the Intentional Binding Effect.

Methods and Experimental Design

The researchers employed the Intentional Binding Effect (IBE) paradigm, adapted for a continuous physical task. The study involved 22 healthy young adults performing a virtual reaching task using a 2D robotic manipulandum.

Experimental setup and task paradigm: (A) Physical setup featuring a 2D linear stage and mirror; (B) Sequence of a trial; (C) Velocity profiles showing the temporal delay between the manipulandum and the cursor; (D) Response screens for delay estimation and agency questionnaires.

The Experimental Setup

Participants moved a virtual cursor toward targets under varying conditions in a $3 \times 3$ within-subject design:

  • Guiding Force: Correct assistance (toward the target), Incorrect guidance (toward a distractor), or No Force.
  • Visual Delay: The cursor's movement was delayed by 300, 500, or 700 ms.

The robot utilized an admittance controller to simulate a virtual mass-damper system, ensuring smooth and compliant interaction between the human and the machine.

Measuring Agency via Intentional Binding

IBE is a psychological phenomenon where the perceived time between a voluntary action and its outcome is compressed when the individual feels in control. Participants were asked to estimate the visual delay in milliseconds. These objective estimates were then compared against standard SoA questionnaires and physical metrics, such as "manipulation share" (the ratio of force contributed by the human versus the robot).

Key Insights: The Assistance Paradox

A significant finding of this study is the "Assistance Paradox." While one might expect a human to feel less in control as a machine takes over more of the work, the results showed that correct assistance maintained or even boosted the sense of agency. This occurred even when the human's physical contribution dropped to approximately 77%.

This indicates that agency is not solely a product of physical effort, but rather how well the machine's force aligns with the user's intent. When the robot helps the user achieve their goal, the brain effectively "claims" the robot's force as its own. This alignment was validated by a significant correlation ($r = -0.41$) between the objective timing bias (IBE) and subjective questionnaire scores.

Key Results and Statistical Findings

  • Perceptual Binding: Participants perceived delays to be 60.2 ms shorter on average than they actually were, confirming the presence of IBE in pHMI tasks.
  • Impact of Force: Incorrect guidance significantly degraded the sense of agency. In these trials, participants were more accurate at detecting actual delays because they did not feel responsible for the movement, resulting in less temporal binding.
  • Experience vs. Performance: Although incorrect forces made the task more difficult, they did not significantly alter total trial duration. This highlights that SoA is a distinct measure of user experience rather than simple task performance.
  • Statistical Significance: Analysis (ANOVA) confirmed a main effect of Delay ($p < 0.0005$) and a significant interaction between Guiding Force and Delay ($p < 0.05$) on perceived timing.

Limitations and Future Directions

The primary limitation of the IBE paradigm is its reliance on introduced delays, which are impractical for safety-critical real-world applications like surgery. Furthermore, the simple reaching task used here may not fully represent complex, unpredictable environments.

Future research aims to develop ambient, implicit measures of agency that can allow robotic systems to tune their control parameters dynamically. By understanding the psychological state of the user in real-time, assistive technologies can become truly intuitive partners in human movement.

Summary updated

Abstract

In the advent of intelligent robotic tools for physically assisting humans, user experience, and intuitiveness in particular have become important features for control designs. However, existing works predominantly focus on performance-related measures for evaluating control systems as the subjective experience of a user by large cannot be directly observed. In this study, we therefore focus on agency-related interactions between control and embodiment in the context of physical human-machine interaction. By applying an intentional binding paradigm in a virtual, machine-assisted reaching task, we evaluate how the sense of agency of able-bodied humans is modulated by assistive force characteristics of a physically coupled device. In addition to measuring how assistive force profiles influence the sense of agency with intentional binding, we analyzed the sense of agency using a questionnaire. Remarkably, our participants reported to experience stronger agency when being appropriately assisted, although they contributed less to the control task. This is substantiated by the overall consistency of intentional binding results and the self-reported sense of agency. Our results confirm the fundamental feasibility of the sense of agency to objectively evaluate the quality of human-in-the-loop control for assistive technologies. While the underlying mechanisms causing the perceptual bias observed in the intentional binding paradigm are still to be understood, we believe that this study distinctly contributes to demonstrating how the sense of agency characterizes intuitiveness of assistance in physical human-machine interaction.
BibTeX
@inproceedings{endo2020effect,
  author = {Satoshi Endo and Jakob Fröhner and Selma Musić and Sandra Hirche and Philipp Beckerle},
  title = {{Effect of External Force on Agency in Physical Human-Machine Interaction}},
  booktitle = {Frontiers in Human Neuroscience},
  year = {2020},
  doi = {10.3389/fnhum.2020.00114},
}
Figure