Between Anthropomorphism, Trust, and the Uncanny Valley: a Dual-Processing Perspective on Perceived Trustworthiness and Its Mediating Effects on Use Intentions of Social Robots

Designing social robots with the aim to increase their acceptance is crucial for the success of their implementation. However, even though increasing anthropomorphism is often seen as a promising way to achieve this goal, the uncanny valley effect proposes that anthropomorphism can be detrimental to acceptance unless robots are almost indistinguishable from humans. Against this background, we use a dual processing theory approach to investigate whether an uncanny valley of perceived trustworthiness (PT) can be observed for social robots and how this effect differs between the intuitive and deliberate reasoning system. The results of an experiment with four conditions and 227 participants provide support for the uncanny valley effect. Furthermore, mediation analyses suggested that use intention decreases through both reduced intuitive and deliberate PT for medium levels of anthropomorphism. However, for high levels of anthropomorphism (indistinguishable from real human), only intuitive PT determined use intention. Consequently, our results indicate both advantages and pitfalls of anthropomorphic design