Manual Control. As a result, mandated AA subjects were more likely to observe "unacceptable" pointer deviations in the secondary task than non-mandated AA subjects, which indicated reduced workload for the former group. Poorer non-mandated AA group performance in gauge monitoring can be attributed to devotion of greater resources to the primary task. Decrease in non-mandated AA subject performance in the secondary task indicated increased primary task workload.
For mandated AA subjects, manual control in the primary task served to significantly increase secondary task workload, as compared to the workload associated with automation. This can be attributed to higher attentional demands being placed on the operator under Manual Control in terms of implementing a target processing strategy, as compared to Blended Decision-Making. Under Manual Control subjects were required to track and collapse targets using the mouse controller; whereas Blended Decision-Making provided for automatic computer target elimination. The result may also be attributed to the observed lack of mandated AA subject concern with allocations to automation from Manual Control and, consequently, decreased operator attention to the gauge display for computer assistant mandates. Under Multitask© Manual Control, the mandated AA group was less likely to observe pointer errors, decreasing secondary task performance and indicating increased subject workload.
Human workload measurement through comparison of operational secondary task performance against baseline levels established under optimal conditions can be used as a trigger mechanism for AA to cause changes in operator primary task workload. This workload assessment method appears to be sensitive to both the degree of human involvement in a primary cognitive task and human responsibility in decision- making concerning control allocations. Whether completely human manual processing or blended human- computer strategizing accompanied by computer task implementation is employed, objective workload is significantly affected when operators are mandated to use either LOA, as compared to non-mandated use. The responsibility of allocation decisions in a cognitively complex task appears to have an effect on task performance and operator workload; however, this is dependent upon whether the allocation is to manual or automated control. Humans who are mandated in allocation decisions by a computer controller tend to be more vigilant of Manual Control mandates than those for automation. This may occur because operators view the manual processing mandates as directives to work. Lack of vigilance to automation mandates can be attributed to concentration on Manual Control performance.
On the contrary, humans who are not mandated in implementing AA, appear to be more vigilant of suggestions by a computer assistant for automated control allocations, as compared to suggestions for manual functioning. This can be attributed to operators interpreting computer suggestions to use automation as a form of validation of their decision to do so. Lack of non-mandated AA operator vigilance to manual control suggestions is likely due to a lack of a perceived need to use Manual Control. This is a particularly troubling inference, as it suggests that operators may not be aware of out-of-the-loop performance problems or appreciate their associated consequences.
Endsley M. R. & Kaber D. B. (in press). Level of automation effects on performance, situation awareness and workload in a dynamic control task. Ergonomics.
Hilburn B. & Jorna P. ( 1997). The effect of adaptive air traffic control (ATC) decision aiding on controller mental workload. In M. Mouloua & J. M. Koonce (Eds.), Human-Automation Interaction: Research and Practice (pp. 84-91). Mahwah, NJ: Lawrence Erlbaum Associates.