Turner, JessicaVanderschantz, NicholasJones, Ben2026-10-082026-10-082026https://hdl.handle.net/10289/18695As social robots are increasingly used for sharing information, it is essential to determine whether people can trust them. Too little trust and people abandon a useful tool; too much and it may cause over-reliance. Morphology is part of communication, a channel that operates in parallel to spoken words, and it can affect perceptions before the interaction has even begun. Human-like appearances are often thought to make a robot more trustworthy, but our review of the literature shows that trust is not determined by embodiment type alone. Features such as resemblance, gaze, and expressive movement act as social heuristics, signalling what a robot is for and what it can do. Because trust arises only where a person is vulnerable and relying on the robot, information-seeking represents a meaningful context for studying trust even when the stakes are low. Within this context, trust is multidimensional, encompassing performance, relational, and moral judgements. Existing research has examined trust in information from websites, search engines, and disembodied agents, and trust in robots performing industrial or collaborative tasks, but little is known about how morphology affects trust when a robot is itself the information source. In this thesis, we address this gap by examining how a robot's physical form shapes the trust people place in it as an information source. Two studies were conducted using the same four devices: a disembodied smart speaker and zoomorphic, mechanomorphic, and humanoid robots. Each robot performed the same information-seeking task, allowing reported and enacted trust to be separated. First, in a survey, 252 participants watched a video of the robots and rated them using the Human-Robot Interaction Evaluation Scale and the Trust Perception Scale-HRI. Under passive observation, morphology had no significant effect on reported trust and only small effects on social perceptions; prior familiarity predicted perceptions instead, and the humanoid did not receive a trust benefit from anthropomorphic attributions. An observational user study then brought the scenario into the real world, where 54 participants chose to approach a robot in person and ask their own questions. Once people sought information themselves, morphology produced large effects, but the dimensions of trust moved in different, sometimes opposing, directions. The humanoid was rated as the most helpful and fit-for-purpose, yet participants were most likely to double-check what it said. The mechanomorphic robot was most relied upon because its form legibly signalled its communicative purpose. This appearance-function fit, rather than human-likeness, determined whether a robot was trusted appropriately. Cuteness and emotion encouraged people to approach, but cuteness alone did not signal credibility, as the zoomorphic robot was cute yet poorly received as an information source. Morphology's effect on trust is thus conditional on engagement, moving the facets of trust independently rather than uniformly. This thesis contributes an early empirical account of how morphology shapes trust in robots as information sources. Our findings indicate that a form that honestly signals what a robot is for, rather than one that is simply human-like, better supports users in calibrating trust. Systematically varying morphological features is a clear next step for tracing these effects to specific design choices.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.hrihcisocial roboticstrustperceptionsmorphologyinformation-seekingThesis