Sensitivity of Self-Report Measures of Trust in Human-Autonomy Teaming

As autonomous agents are integrated into human teams, measuring trust in Human–Autonomy Teams (HATs) becomes essential. Existing trust surveys, developed for human or automation contexts, may not capture HAT-specific dynamics. This study evaluated the sensitivity of three trust measures to trust manipulations in a simulated HAT environment. Thirty-six participants completed five missions with confederates via a Wizard-of-Oz approach, while trust was manipulated through team composition and communicative versus behavioral trust/distrust spread. Surveys assessed trust in teammates and systems after each mission. Results showed varying sensitivity across measures: McAllister’s scales captured communicative cues, Jian’s measures reflected mission and communication effects, and only McAllister’s cognitive scale detected behavioral trust spread. These findings suggest that trust instruments vary in responsiveness across trust types and manipulation sources. Combining instruments based on research goals is critical for capturing trust dynamics in human-autonomy teams.