Model Based Analysis of Subjective Mental Workload During Multiple Remote Tower Human-In-The-Loop Simulations

We report on the analysis of subjective mental workload (WL) and objective task load (TL) measurements of a Multiple Remote Tower Operation (MRTO) simulation experiment with 12 air traffic control officers (ATCos). The experiment was performed as part of a project for the development of a remote tower center (RTC) for the centralized control of several airports (APs) from afar (Fürstenau, Virtual and remote control tower. Springer, Switzerland, 2016). Specifically, we were interested in the question if being responsible for two or more traffic systems at the same time, causes workload independently from actual traffic load. Subjective WL was measured by means of the one-dimensional quasi real time Instantaneous Self Assessment method (five level ISA scale) whereas objective TL data were obtained online by monitoring ATCo’s communication with pilots (radio calls frequency RC and duration RD), both dependent on the environmental traffic load n. In addition to variance analysis (ANOVA) for quantifying linear correlations (WL/TL~n) a new cognitive resource limitation model for nonlinear (logistic) regression-based parameter estimates was applied to the data (Fürstenau et al., Theor Issues Ergon Sci, 2020). ANOVA results supported initially stated hypotheses on significant increase of subjective and objective WL/TL measures with increasing traffic flow n, as well as a WL increase under transition from one controller per airport (baseline) to two-airport control by a single ATCo (Lange et al., Analyse des Zusammenhangs zwischen dem Workload von Towerlotsen und objektiven Arbeitsparametern, 2011). Furthermore, a hypothesized mediator effect of communication TL was determined, mediating the dependency of ISA-WL on traffic load n. The extension of the of the (linear) ANOVA by the (nonlinear) logistic model-based analysis of ISA(n) and RC(n) data allowed for the quantification of theoretically founded WL/TL sensitivity (ν/ρ) and bias parameters, the latter characterizing the difference between work conditions. The validity of the regression-based parameter estimates was supported by the theoretical prediction of model parameters based on prior information (e.g. scale limits). Estimates of the nonlinear model parameters quantified the dissociation between the subjective WL and objective communication load measures. Derived from the assumption of cognitive resource limitation the logistic model provides a theoretical foundation for the discussion of the initially stated hypotheses regarding WL/TL characteristics. Specifically, a stimulus (RC)—response (ISA) power law analysis according to (Fürstenau and Radüntz, Power law model for subjective mental workload and validation through air-traffic control human-in-the-loop simulation, 2021) allowed via the Stevens exponent γ(=ρ/ν) to formalize and quantify the assumed mediator role of the objective communication TL between traffic flow and the subjective ISA-WL response.