Brain-Behavior Mappings of Trust in Shared Space Human-Robot Collaboration

Abstract

The future of collaborative manufacturing robotics must be designed to augment and support human capabilities in tasks that are difficult to automate. In such tasks, augmenting human cognition, situation awareness, and dexterity can be achieved through collaboration with an intelligent assistive robot. To optimally leverage human strengths with robot strengths, consideration for emergent human factors must be addressed. Particularly, trust in technology has repeatedly driven technology adoption strategies, operator safety-related behaviors, and operator utilization of assistive technology. While the importance of trust cannot be understated, our knowledge of trust and trusting behaviors are limited by the metrics employed to study trust. Historically, trust has been captured through subjective measures with minimal effort to identify deployable continuous measures or to provide contextual insight into how trust connects to important downstream behaviors (e.g., adoption, utilization). This has limited our understanding to considerations that are quantifiable by the users, preventing more granular understanding, such as the specific cognitive processes and neural strategies that are employed to experience and respond to trust changes. This level of insight is necessary to understand why a change in robot factor corresponds with a change in trust and why trust itself connects to changes in human behavior. Because of this gap, it is widely unknown why specific populations are more predisposed to specific responses to trust and how we can communicate this to the intelligent robot. This dissertation predominantly discusses and utilizes a neuroergonomics approach, i.e., the study of the brain and behavior in context, to understand and monitor trust in real-time. Neuroimaging methods are increasing in availability and deployability allowing for a realistic tool to study and provide feedback directly to robots. This dissertation provides the foundation for the use of brain imaging metrics in trust through providing insight into neural correlates of trust and their connection to downstream behavioral metrics. This is achieved through 1) a review of prior literature on neural correlates of trust in collaborative technologies and considerations for generalizability between domains, 2) a human-subjects study in a shared space HRC, where trust is manipulated and corresponding neural activity monitored alongside performance changes, and 3) a second human-subjects study in a shared space HRC, where trust is manipulated and corresponding brain activity, eye tracking, and performance behaviors monitored to identify brain-behavior relationships. This understanding can allow for human-centric designs of collaborative robots that consider trust related behaviors and provide direction for future research studying trust in human-technology interactions.