All automations must, at some point in their lifecycle, interface with one or more humans. Whether operators, end-users, or bystanders, human responses can determine the perceived utility and acceptance of an automation. It has been long believed that human trust is a primary determinant of human-automation interactions and further presumed that calibrating trust can lead to appropriate choices regarding automation use. However, attempts to improve joint system performance by calibrating trust have not yet provided a generalizable solution. To address this, we identified several factors limiting the direct integration of trust, or metrics thereof, into an active mitigation strategy. The present paper outlines our approach to addressing this important issue, its conceptual underpinnings, and practical challenges encountered in execution. Among the most critical outcomes has been a shift in focus from trust to basic interaction behaviors and their antecedent decisions. This change in focus inspired the development of a testbed and paradigm that was deployed in two experiments of human interactions with driving automation that were executed in an immersive, full-motion simulation environment. Moreover, by integrating a behavior and physiology-based predictor within a novel consequence-based control system, we demonstrated that it is possible to anticipate particular interaction behaviors and influence humans towards more optimal choices about automation use in real time. Importantly, this research provides a fertile foundation for the development and integration of advanced, wearable technologies for sensing and inferring critical state variables for better integration of human elements into otherwise fully autonomous systems.