The Role of Psychophysiological Measures as Implicit Communication Within Mixed-Initiative Teams

There has been considerable effort, particularly in the military, at integrating automated agents into human teams. Currently, automated agents lack the ability to intelligently adapt to a dynamic operational environment, which results in them acting as tools rather than teammates. Rapidly advancing technology is enabling the development of autonomous agents that are able to actively make team-oriented decisions meaning truly intelligent autonomous agents are on the horizon. This makes the understanding of what is important to team performance a critical goal. In human teams, mission success depends on the development of a shared mental models and situation awareness. Development of these constructs requires good intra-team communication. However, establishing effective intra-team communication in a mixed-initiative team represents a current bottleneck in achieving successful teams. There has been significant research aimed at identifying modes of communication that can be used both by human and agent teammates, but often neglects a source of communication or information for the agent teammate that has been adopted by the human robot community to increase robot acceptance. Specifically, the use of psychophysiological features supplied to the agent that can then use algorithms to infer the cognitive state of the human teammate. The utility of using psychophysiological features for communication within teams has not been widely explored yet representing a knowledge gap in developing mixed-initiative teams. We designed an experimental paradigm that created an integrated human-automation team where psychophysiological data was collected and analyzed in real-time to address this knowledge gap. We briefly present a general background to human automation teaming before presenting our research and preliminary analysis.