Data underlying the article: "I See What You Did There": Understanding People's Social Perception of a Robot and its Predictability

This data collection is associated with the article ``I See What You Did There'': Understanding People's Social Perception of a Robot and its Predictability, published in ACM Transactions in Human-Robot Interaction in 2021.<br>Abstract:Unpredictability in robot behaviour can cause difficulties in interacting with robots. However, for social interactions with robots, a degree of unpredictability in robot behaviour may be desirable for facilitating engagement and increasing the attribution of mental states to the robot. To generate a better conceptual understanding of predictability, we looked at two facets of predictability, namely the ability to predict robot actions and the association of predictability as an attribute of the robot. We carried out a video human-robot interaction study where we manipulated whether participants could either see the cause of a robot's responsive action, or could not see this, because there was no cause, or because we obstructed the visual cues. Our results indicate that when the cause of the robot's responsive actions was not visible, participants rated the robot as more unpredictable and less competent, compared to when it was visible. The relationship between seeing the cause of the responsive actions and the attribution of competence was partially mediated by the attribution of unpredictability to the robot. We argue that the effects of unpredictability may be mitigated when the robot identifies when a person may not be aware of what the robot wants to respond to, and uses additional actions to make its response predictable.

本数据集关联2021年发表于《ACM人机交互汇刊（ACM Transactions on Human-Robot Interaction）》的论文《“我知晓你所为”：理解人类对机器人的社会感知及其可预测性》。
摘要：机器人行为的不可预测性可能会给人机交互带来障碍。然而，在机器人社交交互场景中，适度的行为不可预测性或有助于提升互动参与度，并增强人类对机器人的心理状态归因。为了更深入地理解可预测性的概念内涵，本研究聚焦可预测性的两个维度：一是预测机器人动作的能力，二是将可预测性作为机器人属性的关联认知。本研究开展了一项视频人机交互实验，通过两种方式操控参与者能否获知机器人响应动作的成因：一是无成因存在，二是遮挡视觉线索，以此设置参与者可见/不可见成因的实验条件。实验结果显示，相较于机器人响应动作成因可见的情况，当成因不可见时，参与者会认为该机器人的不可预测性更强、能力水平更低。“获知机器人响应动作成因”与“对机器人能力的归因”二者间的关联，部分由“对机器人不可预测性的归因”所中介。本研究提出，若机器人能够识别出人类可能未知晓其响应意图的场景，并通过额外动作使自身响应具备可预测性，则可缓解不可预测性带来的负面影响。