Revisiting Trust in Machines: Examining Human–Machine Trust Using a Reprogrammed Pasteurizer Task

Abstract

Automated technologies have brought a number of benefits to professional domains, expanding the area in which humans can perform optimally in complex work environments. Human–automation trust has become an important aspect when designing acceptable automated systems considering general users who have no comprehensive knowledge of the systems. Muir and Moray (1996) proposed a model of human–machine trust incorporating predictability, dependability, and faith as predictors of overall trust in machines. Though Muir and Moray (1996) predicted that trust in machines grows from predictability, then dependability, and finally faith, their results suggested the opposite. This study will reexamine their theoretical framework and test which of the three dimensions governs initial trust in automation. Participants will be trained to operate a simulated pasteurization plant, as in Muir and Moray (1996), and they will be asked to maximize system performance in the pasteurizing task. We hypothesized that faith governs overall trust early in the interaction with the automated system, then dependability, and finally predictability as lay automation users become more familiar with the system. We attempt to replicate the results of Muir and Moray (1996) and argue that their model should be revised for trust development for general automation users.