Sensitivity of the spatial distribution of fixations to variations in the type of task demand and its effectiveness as a trigger for adaptive automation

High-risk environments such as healthcare, transport and air traffic control are characterised by highly dynamic, unpredictable, and uncertain events. A human operator's presence is necessary to monitor and control the system when critical events occur in these contexts. At the same time, the system should monitor the operators' functional status and support him when necessary. The proposed research activity investigates the spatial distribution of eye fixations as a real-time measure of mental workload. Ocular activity is known to be sensitive to variations in mental workload. Many attempts have been made to derive a stable measure of the cognitive resources allocated to a task using eye-trackers' information. Recent studies have successfully related the distribution of eye fixations to the mental load. The scope of this research project is to devise a set of experiments for separating the contribution of three types of tasks demands (i.e., temporal, mental, and physical) and, to determine which of these (and when) should be considered for using an index of spatial distribution as a trigger in ocular-based adaptive systems. The project has three different objectives: 1) assessing the sensitivity of the proposed measure to different types of tasks demands with a large sample and a within-subject design; 2) evaluate the effectiveness of the proposed measure as a trigger for adaptive automation and 3) using more complex algorithms to provide a more stable measurement over time and investigate variations in the frequency domain. The first chapter provides a review of the theoretical models proposed in the literature about automation, highlighting the relationship between machine and operator, and the cognitive processes involved. The second chapter describes the physiological indicators of mental workload present in the literature, focuses on measures derived from ocular parameters such as pupillary diameter, saccades, fixations and scanpath analysis. In the last two chapters, four experimental studies are described and discussed. The aim was to evaluate how the visual exploration strategy changes with different mental workload levels and task demands. The index used to analyse the visual strategy, the nearest neighbour index, was then investigated as a trigger in an adaptive automation system. The results indicated the high diagnostic power of the measure and provided the background for future applications.