[Paper Review] If you trust me, I will trust you: the role of reciprocity in human-robot trust.

Joshua Zonca, Anna Folsø+1 more|arXiv (Cornell University)|Jun 28, 2021

Social Robot Interaction and HRI80 references4 citations

TL;DR

This study investigates how reciprocity shapes human-robot trust through a joint perceptual task where participants and a humanoid robot signaled trust to each other. Results show that participants withheld overt distrust when expecting future interactions, indicating emergent reciprocity in trust expression—a dynamic absent in human-computer interactions.

ABSTRACT

Trust is fundamental in human-human and human-robot interaction. Extensive evidence has shown that trust among humans is sustained by reciprocity, whereas knowledge on reciprocal dynamics of trust in human-robot interaction is very limited. To investigate reciprocity in human-robot trust, we designed a joint task in which a human participant and a humanoid robot made perceptual judgments while signaling their trust in the partner. The robot's trust was dynamically manipulated along the experiment. Results show that participants were less willing to learn from a robot that was showing high trust in them. However, participants were unwilling to overtly disclose their distrust to the robot if they expected future interactions with it, suggesting the emergence of reciprocity in their overt expression of trust. Importantly, these effects disappeared when participants interacted with a computer. Our findings may have an impact on the design of robots that could effectively collaborate with humans.

Motivation & Objective

• To examine whether reciprocity influences trust expression in human-robot interaction.
• To investigate how expectations of future interaction affect human willingness to express distrust toward robots.
• To compare trust dynamics between human-robot and human-computer interactions.
• To assess how dynamic manipulation of robot trust influences human trust behavior.

Proposed method

• Conducted a joint perceptual judgment task between human participants and a humanoid robot.
• Dynamically manipulated the robot's displayed trust level throughout the experiment.
• Collected participants' trust expressions through overt behavioral signals during interaction.
• Used a control condition with a computer instead of a robot to isolate social dynamics.
• Measured participants' reliance on robot judgments and their willingness to disclose distrust.
• Analyzed behavioral data to identify patterns of trust reciprocity and future-interaction effects.

Experimental results

Research questions

• RQ1 How does a robot's displayed trust influence a human's willingness to learn from it?
• RQ2 Does the expectation of future interaction alter a human's expression of distrust toward a robot?
• RQ3 In what ways does human-robot trust differ from human-computer trust in terms of reciprocity?
• RQ4 How do humans regulate their trust expression when interacting with socially responsive robots versus non-social agents?

Key findings

• Participants were less willing to learn from a robot that displayed high levels of trust in them.
• Participants refrained from overtly expressing distrust when they expected future interactions with the robot, indicating reciprocal trust behavior.
• This pattern of withholding distrust did not occur when participants interacted with a computer, highlighting the role of social cues.
• The emergence of reciprocity in trust expression was specific to human-robot interaction, not human-computer interaction.
• Dynamic trust signaling by the robot significantly influenced human trust behavior.
• The findings suggest that social responsiveness in robots can shape human trust dynamics through reciprocal expectations.