An Immersive Investment Game to Study Human-Robot Trust

Sebastian Zörner, Emy Arts, Brenda Vasiljevic, Ankit Srivastava, Florian Schmalzl, Glareh Mir, Kavish Bhatia, Erik Strahl, Annika Peters, Tayfun Alpay, Stefan WermterFrontiers Robotics AI 2021 · DOI · OpenReview

TL;DR

This paper introduces an immersive, science-fiction-themed investment game designed to more realistically measure human trust in robots. The study found that this novel scenario effectively measures trust and that non-verbal communication from robots significantly enhances human trust and positive perceptions of the robots.

What Problem Does This Paper Address?

Developing trust is crucial for effective human-robot interaction (HRI), especially as robots become more integrated into our daily lives. However, measuring this trust in a way that reflects real-world complexity is challenging. Traditional methods, like the economic "investment game," often fall short when applied to HRI because they lack the social nuances and immersive environments that influence human trust. This paper addresses the need for a more sophisticated and ecologically valid method to study human-robot trust, particularly by incorporating non-verbal communication (NVC) and a more engaging scenario.

Methods

The researchers adapted the classic investment game into an immersive, science-fiction HRI scenario. Participants took on the role of a spaceship commander tasked with delivering cargo, advised by two "Neuro-Inspired Companion" (NICO) robots. Instead of money, participants allocated "energy cells" between the robots and themselves to solve simulated emergencies. This setup aimed to:

  • Create a plausible currency: Energy cells were tied to mission progress, making them valuable to both humans and robots.
  • Foster intrinsic motivation: The sci-fi narrative and cooperative goal encouraged engagement beyond simple monetary incentives.
  • Control for prior knowledge: The scenario was designed so participants couldn't rely on pre-existing domain expertise, forcing them to depend on the robots' advice.
  • Investigate NVC: Two conditions were used: one robot exhibited elaborate NVC (gaze, facial expressions, gestures), while the other had minimal NVC.

Trust was measured quantitatively through energy cell allocation (allocation metric) and qualitatively via post-experiment questionnaires (relative trust metric). The study also used the Godspeed questionnaire to assess robot perception (anthropomorphism, animacy, likeability, intelligence).

Datasets

Not discussed in the paper. The study utilized a custom-built experimental setup and collected data from participants.

Key Insights and Ideas

  • Immersive scenarios are key for HRI trust research: Simply adapting economic games isn't enough; creating an engaging, realistic context is vital for eliciting genuine trust behaviors.
  • Non-verbal cues matter, a lot: Even simple NVC from robots significantly boosted human trust and made them seem more human-like and animated. This suggests that robots don't need to be perfectly human to benefit from human-like communication.
  • Trust is a multi-faceted construct: The study found a strong correlation between cognitive and affective trust, reinforcing the idea that a robot's competence and its perceived motives are intertwined in building trust.
  • Behavioral patterns reveal insights: The identification of "alternating-minimum investment" behavior highlighted a subset of participants who didn't engage with the trust game in the same way, suggesting potential avenues for future research into participant motivations and engagement.

Key Results

  • Validating the scenario: A statistically significant correlation (r = 0.43, p = 0.02) was found between the measured trust (allocation metric) and self-assessed trust (relative trust metric) for the "main group" of participants, validating the immersive investment game as a measure of HRI trust.
  • Impact of NVC: Participants allocated significantly more energy cells to the robot exhibiting NVC compared to the one with minimal NVC (p = 0.0008).
  • Perception enhancement: NVC significantly increased perceived anthropomorphism (p = 0.008) and animacy (p = 0.00001) of the robots, as measured by the Godspeed questionnaire. Likeability and intelligence were not significantly affected.
  • Personality correlations: Participants scored lower in conscientiousness and higher in openness than the general German population. Neuroticism was positively correlated with anthropomorphic perception of the robot.

Limitations and Future Work

  • Generalizability of Sci-Fi: While the sci-fi setting enhanced immersion, it might limit direct generalizability to real-world, non-fictional HRI contexts. Future work could explore other settings.
  • Role-playing effects: High immersion might lead some participants to engage in role-playing, potentially influencing their behavior beyond typical responses.
  • Limited NVC comparison: The study compared "elaborate NVC" to "minimal NVC." Future research should explore the boundaries of minimal NVC and the impact of specific gestures and cues.
  • Robot limitations: The study used NICO robots, which have certain limitations in expressiveness. Future work could explore how different robot appearances and capabilities affect trust.
  • Participant pool: The study was conducted with participants from a computer science department, potentially biasing their familiarity with robots and technology.
  • Experiment length: The authors suggest that a more prolonged experiment with more rounds might be beneficial for building trust, especially for participants who initially showed less engagement.
  • Autonomous system issues: Occasional malfunctions in the autonomous system could have broken immersion. Future studies might consider a "wizard-of-oz" approach for more seamless interaction.

Summary updated

Abstract

As robots become more advanced and capable, developing trust is an important factor of human-robot interaction and cooperation. However, as multiple environmental and social factors can influence trust, it is important to develop more elaborate scenarios and methods to measure human-robot trust. A widely used measurement of trust in social science is the investment game. In this study, we propose a scaled-up, immersive, science fiction Human-Robot Interaction (HRI) scenario for intrinsic motivation on human-robot collaboration, built upon the investment game and aimed at adapting the investment game for human-robot trust. For this purpose, we utilize two Neuro-Inspired COmpanion (NICO) - robots and a projected scenery. We investigate the applicability of our space mission experiment design to measure trust and the impact of non-verbal communication. We observe a correlation of 0.43 (p=0.02) between self-assessed trust and trust measured from the game, and a positive impact of non-verbal communication on trust (p=0.0008) and robot perception for anthropomorphism (p=0.007) and animacy (p=0.00002). We conclude that our scenario is an appropriate method to measure trust in human-robot interaction and also to study how non-verbal communication influences a human’s trust in robots.
BibTeX
@article{zorner2021immersive,
  author = {Sebastian Zörner and Emy Arts and Brenda Vasiljevic and Ankit Srivastava and Florian Schmalzl and Glareh Mir and Kavish Bhatia and Erik Strahl and Annika Peters and Tayfun Alpay and Stefan Wermter},
  title = {{An Immersive Investment Game to Study Human-Robot Trust}},
  journal = {Frontiers Robotics AI},
  year = {2021},
  doi = {10.3389/frobt.2021.644529},
}

Loading related content...

Figure