Interactions of Training and Decision Support: Measuring the Benefits in Trust, Situation Awareness, and Workload

Inappropriate trust of automation can lead to suboptimal performance, in which the respective capabilities of humans and computers are not effectively exploited. In some cases, over- or under-reliance can lead to catastrophe. The primary objectives of the proposed Phase I work are (1) to develop a systematic general framework for training users of decision aids, with particular attention to the role of facilitating accurate and useful assessments of trust, and (2) to develop methods for measuring the benefits of such training, and of decision aid use in general, for workload and situation assessment. CTI's proposed work falls into four tasks. We will: (1) develop a framework for identifying the training requirements imposed by decision aid use, as a function of the aided function, the automation style, decision aiding features, and user experience: (2) identify specific training techniques that will address the requirements of decision aid use; (3) develop robust and valid measures of situation awareness and workload suitable for testing the effects of training and decision aid use; and (4) identify a specific decision aiding system and application domain in which the framework, training methods, and measures can be tested. The proposed training requirements framework, training methods, and measures will find promising markets in a variety of military, civilian, and non-governmental domains where decision aids are being introduced. These include Army aviation, Naval surface warfare, air traffic control, and commercial air transport.