What Robots Do Matters More Than What They Look Like: Task Context Shapes Trust in Educational HRI

Synthesis: This Discobot project study (2026) examines how robot appearance and task type jointly influence trust in socially assistive robots (SARs) in educational and information-sharing contexts. Using a within-subjects video-based experiment (N = 81), participants evaluated three robots with distinct appearances while performing three educationally relevant tasks: teaching, procedural instruction, and personal-information discussion. Repeated-measures analyses showed a strong main effect of task on trust, with participants reporting the highest trust during instructional tasks — indicating that what a robot does matters more than what it looks like.

Key Findings

• Socially assistive robots (SARs) are increasingly deployed in educational and information-sharing contexts, supported by Large Language Models (LLMs) advances enabling fluent real-time interaction.
• Despite the growing diversity of robot embodiments, it remained unclear whether a single robot appearance is appropriate across different interaction tasks or whether trust depends primarily on contextual factors.
• A within-subjects video-based experiment (N = 81) had participants evaluate three robots with distinct appearances across three tasks: teaching, procedural instruction, and personal-information discussion.
• Repeated-measures analyses showed a strong main effect of task on trust, with the highest trust during instructional tasks — task context shapes trust in educational HRI more than appearance.

Study Design & Method

This is a within-subjects experimental study. The researchers conducted a video-based experiment with 81 participants who evaluated three socially assistive robots with distinct appearances while the robots performed three educationally relevant tasks (teaching, procedural instruction, and personal-information discussion). Repeated-measures analyses examined the effects of robot appearance and task type on participants' trust in the robots, testing whether trust is determined primarily by appearance or by contextual task factors.

What this means for practice

• Designers. Choose the robot's role and behavior for the task rather than optimizing embodiment, since trust was highest during procedural instruction and teaching and lowest during personal-information discussion across all three robots, with appearance exerting only limited influence.
• Designers. Match deployment to that trust profile — assign procedural and instructional roles first and treat personal-information discussion as the task needing the most safeguards and explanation.
• Educators. Tell learners what the robot has been tasked to do before deployment, because their trust tracks the perceived risk and expected competence of the task rather than the robot's looks.
• Educators. Revisit trust after real classroom use: this study captured first impressions under strictly scripted robot behavior, and trust in human-robot interaction shifts through repeated encounters.

Limitations

The study used a within-subjects video-based paradigm rather than physical interaction, which may not fully capture real-world trust dynamics. The three tasks and robot appearances are a sample of a broader design space. The sample (N = 81) and specific task framing may limit generalizability across educational contexts and populations.

Citation

Velentza, A.-M., Nikou, K., Bosser, A.-G., & Fachantidis, N. (2026). What robots do matters more than what they look like: Task context shapes trust in educational HRI.