Air Traffic Management (ATM) has long focused on the challenge of air traffic controller overload. Increasing traffic and staffing shortages all contribute to high workload. Yet there is another cognitive state that receives less attention, despite potentially carrying its own safety implications.
Global Airspace Radar’s Editor Katarzyna Żmudzińska spoke with Professor Mark Young, Professor of Human Factors in Transport at the University of Southampton, to explore the topic of mental underload in Air Traffic Control (ATC) and its connection to automated environments.
Evidence from other domains characterised by continuous tasks shows that prolonged mental underload, caused by low workload or highly automated environments, can reduce people’s ability to quickly re-engage when faced with an emergency or another unexpected situation. There is not enough research regarding this phenomenon in the ATC domain, however, assuming controlling air traffic is a continuous activity, we might expect similar results.
Measuring performance
Measuring Air Traffic Controller (ATCO) performance remains one of the major challenges in human factors research, and it is an area currently being investigated at the University of Southampton. Before performance can be measured, researchers first need to answer a fundamental question: what actually constitutes good performance in ATC? The answer is not straightforward; safe operations are clearly essential but performance is also influenced by efficiency, workload management and other factors, making it difficult to define using a single measure.
As a result, researchers rely on a combination of potential indicators rather than one definitive metric. For instance, when measuring mental workload these indicators include primary task performance or assessing how well controllers carry out their operational tasks; secondary task performance, which is used to estimate spare cognitive capacity; subjective measures, such as self-assessment rating scales for mental demand, effort and perceived performance; and physiological indicators including eye movements, blink rate and heart rate.
Ultimately, researchers are aiming to identify reliable indicators that signal when an ATCO is approaching the limits of effective performance, just before performance begins to deteriorate. Detecting this transition could help identify both overload and underload, allowing future systems to provide support before safety or performance are affected.
The automation paradox
Good practice in human factors states that automation should aim to support, rather than replace, the human operator. Research suggests that operators perform best when they remain actively involved in meaningful tasks rather than acting solely as passive monitors.
Automation, however, presents a paradox. The more capable automated systems become, the more important it is to invest in refresher training to ensure operators maintain the skills needed to intervene when required. While automation reduces routine workload, it also reduces opportunities to practise those skills.
Researchers believe that people perform best within an optimal range of cognitive engagement, somewhere between overload and underload. Exactly where this optimum lies varies between individuals and depends on factors such as experience, skills and even mood. Although the existence of this sweet spot is well established in theory, translating it into practical recommendations for task and system design remains challenging.
The human-automation partnership
Trust plays a fundamental role in how people interact with automated systems. Interestingly, some research indicates that trust in automation develops much like trust between people. Operators tend to trust a system when they believe it is at least as capable as they are of performing a task. That trust is built gradually over time but can be lost quickly in cases of unexpected behaviour.
Once trust has been established, operators naturally begin to rely on the system. This behaviour is often described as overtrust or complacency. However, Prof. Young cited research that argues that this behaviour can be completely rational: If an automated system has consistently demonstrated that it performs reliably, there is little reason for an operator to suddenly question it if their attention is demanded elsewhere. Particularly during periods of high workload, relying on a trusted system is a rational allocation of attention.
The challenge, therefore, is not to prevent operators from trusting automation, but to design systems that support appropriate trust. Well-designed automation should communicate clearly, provide feedback on its actions, and help operators understand what the system is doing and why. In many ways, effective automation should follow the same principles as good human collaboration – being a good team member.
How the role of the air traffic controller will evolve over the medium and long term remains an open question. “If we are still expecting ATCOs to supervise and monitor the system,” Prof. Young said, “then I would argue that we need to support, rather than replace, the human and let them continue some meaningful involvement in the task. If, on the other hand, automation becomes reliable enough not to need supervision, then ATCOs can transition to a more strategic role.”
