Impact of Autonomous System's Reliability and Explainability on Human Operator's Attitudinal Trust and Behavioral Reliance

The rapid advancement of autonomous systems (AS) highlights the importance of understanding the factors that influence their adoption, particularly trust. System reliability is consistently recognized as a strong determinant of trust, whereas the influence of explainability remains less clear. Prior research has produced mixed findings, partly due to inconsistent definitions and measures of trust used in literature - some studies use behavioral reliance, while others argue that attitudinal measures are better. In this study, we investigated the relationship between attitudinal trust and three indicators of behavioral reliance in a troop monitoring task using four AS with varying reliability and explainability levels. Results showed that trust was significantly influenced by reliability (p < 0.001) and explainability (p = 0.03). In contrast, reliance was unaffected by explainability, though high reliability increased appropriate reliance (p < 0.001), reduced overreliance (p < 0.001), and increased agreeableness with the system (p < 0.001). Trust and reliance were moderately correlated (r = 0.5, p < 0.001), confirming their related yet distinct nature. These findings emphasize reliability as the dominant driver of both attitudinal and behavioral outcomes while revealing more nuanced effects of explainability. Our results underscore the importance of disentangling these constructs to better design AS that promote calibrated trust and appropriate reliance.