Main Subcommittee Category
Human Performance Analysis and EngineeringHow would you label your submission?
GovernmentAbstract
Adaptive automation has been identified as one of the most important topics in the history of human factors. Adaptive automation can help manage operator cognitive workload to acceptable levels to facilitate optimal performance. Future systems will become more complex and will have the potential to include more scalable autonomy as part of their design. Being able to assess cognitive workload in these future systems will be critical to ensure high levels of performance while mitigating negative outcomes for an operator. To investigate the impacts of future systems using these capabilities, this study leveraged cognitive workload prediction models using the Improved Performance Research Integration Tool (IMPRINT) to model operator workload while completing an adaptive automation scenario in NASA’s Multi-Attribute Task Battery-II (MATB). After completing a task analysis, IMPRINT models were developed using the default anchors in IMPRINT and with feedback from expert users. Forty participants completed a 20-minute trial in MATB which consisted of multiple levels of workload and dynamically changing levels of automation. During completion of the task, the researcher prompted participants every 60 seconds to rate their experienced cognitive workload as a percentage of their maximum workload. This approach to subjective workload assessment is known as the Continuous Subjective Workload Assessment Graph (CSWAG) technique. CSWAG results from the study showed statistically significant differences between workload and automation conditions. Additionally, CSWAG results were correlated with the workload prediction models, serving to validate the CSWAG approach as a one method to assess the representational capacity of the IMPRINT models. This paper is important to the community as the approach of coupling IMPRINT models with surrogate measures of cognitive workload showed promise to gain insight more closely into an operator’s experience with adaptive automation systems and can further be used to forecast workload impacts in future systems yet to be developed.