Loss in Human-Robot Interaction: An Examination of Attachment, Persona, and Embodiment

Lydia Melles

Advisor: Elizabeth Phillips, PhD, Department of Psychology

Committee Members: Eileen Roesler, Tyler Shaw

David J. King Hall, #2007b
November 23, 2026, 12:00 PM to 02:00 PM

Abstract:

Humans are social by nature and tend to develop emotional attachments, strong affectionate bonds, with others. Attachments are typically formed towards other humans but may also occur towards other entities like pets and objects, such as robots. In human-robot interaction, attachment can be beneficial, but it can also be problematic. One of the problems with attachment to robots occurs when robots break. Losing an attachment figure of any kind can cause negative emotions and even grief. In cases of attachment to robots, people have reported being resistant to replacing damaged robots, experiencing intense negative emotions when their robots are no longer functional, and even performing funerals for their robots. As robots and other artificial agents are becoming increasingly prevalent, it is important to understand the potential downsides these technologies may bring with them and what can be done about those downsides. The purpose of this dissertation is to investigate what factors influence individuals to become attached to robots and a method to reduce the negative consequences of attachment when a robot is lost. Also discussed is previous research on deconstructed trustee theory which inspired this work on attachment.  

To study robot loss, an experiment was conducted in which participants (N = 80) were encouraged to form attachments to a robot, and then the robot appeared to break by falling off a table. The robot was then replaced in one of four ways. In the full replacement condition, the robot was fully replaced. In the re-embodiment condition, the original robot's persona, its programmed personality, was preserved by transferring, or re-embodying, it into a new physical form. In the re-personation condition, the original robot's physical embodiment, its hardware, was preserved, but a new persona was transferred, or re-personated, into it. In the control condition, the robot fell but was not damaged or replaced.

The results indicate that when replacing a robot that has broken, if the persona of the robot is preserved, negative affect associated with the robot breaking is significantly reduced compared to if the persona is not preserved. This finding suggests that what people are upset about losing is not the body of the robot but the persona of the robot. In fact, there was no significant difference found between the control condition, where no part of the robot is replaced, and the re-embodiment condition, where the body was replaced but the persona was preserved. Preserving the persona of a robot may be a method for lessening the negative emotions experienced when robots break. This finding could inform the design of robots in the future. It was also found that attitudes towards robots, attachment style, and mind perception of robots were all significant predictors of attachment formation. This finding builds on the limited research that currently exists on robot attachment and may help identify which people may be the most likely to form attachments to robots.