How do Individual Differences in Attention Control and Working Memory Capacity Relate to Monitoring Automation?

Abstract

With the proliferation of automated tasks, software, and systems, humans are moving from an active participant in the function of a task to a passive monitor of an automated system that is completing that task. Unfortunately, humans are not well-suited for monitoring roles and there is a need to better understand the factors involved when humans successfully identify when an automated system fails. The goal for this research was to determine whether individual differences in attention control (as measured by the anti-saccade task) and working memory capacity (as measured by the shortened operation span) related to an individual’s ability to detect automation failures. In experiment 1, there was a significant positive relationship ( r = .31) between scores on the anti-saccade task and the number of automation failures that participants detected. In experiment 2, there was a significant positive relationship ( r = .32) between scores on the shortened operation span and the number of automation failures that participants’ detected. The results suggest that certain individuals are better suited for detecting automation failures. Selecting for these individuals may be a fruitful endeavor as automated systems continue to grow across society.

Attentional Control and Performance Across Increasing Degrees of Unreliable Automation

“…These findings support those of Foroughi et al (2019) and expand the literature by introducing high-level automation. Our results also allow for a more granular understanding of the relationship between AC and the use of unreliable automation.…”

“…In precisely the situations where automation is typically implemented (high workload, multitasking situations), the ability to control attention seems crucially important. Indeed, in the first empirical demonstration of a link between AC and automation performance, Foroughi et al (2019) observed a positive relationship between AC and failure detection using a supervisory control task. In their study, participants completed a task where they had to monitor a trisected screen which had various shapes flowing through each section.…”

“…There is little work investigating the connection between other high-level cognitive processes such as attentional control and performance. Foroughi et al (2019) found a positive correlation between attentional control and task performance. However, they only included a low-level form of automation.…”

Conceptualizing Team Working Memory: Implications for Human-Automation Collaboration

“…Furthermore, positive correlational relationships have been found between WM scores as measured by the shortened Ospan (Foster et al, 2015) and the ability to notice automation failures (Foroughi et al, 2019). High spans detected more automation failures.…”

“…SA was measured with a change detection task. High spans are better at noticing (Foroughi et al, 2019) and managing automation failure (Jipp, 2016), which can be characterized as departures from normal operation. The other 3 automation levels did not provide this flexibility.…”