Automation, assessment, and the unresolved question of cognitive resilience

The cockpit did not evolve alone

Commercial aviation is often described in technological generations. Early jets required continuous manual manipulation and mechanical interpretation. Integrated auto-flight systems shifted pilots into supervisory roles. Glass cockpits layered abstraction over raw data. Fly-by-wire introduced control laws that mediate pilot intent through software. With each transition, the cockpit became more capable and more opaque.

The pilot’s task evolved from direct control to cognitive oversight of complex, software-mediated systems.

At the same time, another generational shift occurred. Today’s pilots in training have never known a world without smartphones, algorithmic feeds, or persistent digital connectivity. Their cognitive development unfolded within environments shaped by immediacy, abstraction, and reliance on interfaces. Information is accessed rather than stored. Systems operate invisibly in the background.

The cockpit did not evolve alone. Human cognitive ecology evolved alongside it.

Performance is visible. Understanding is not.

Modern training and assessment systems are sophisticated. Competency-Based Training and Assessment (CBTA) evaluates behavioural markers across decision-making, workload management, communication, and procedural discipline. Simulator sessions test the application under structured scenarios. Checkrides confirm adherence to defined standards.

Performance is observable.

But cognition is not directly observable.

Bloom’s taxonomy, often cited but rarely interrogated in aviation contexts, distinguishes between remembering, understanding, and applying knowledge. In practice, aviation training clearly assesses remembering and applying. Pilots demonstrate that they can execute procedures and manage scenarios within expected boundaries.

Understanding, however, occupies a more ambiguous space.

It is possible to apply a procedure effectively without articulating the underlying system logic it protects. It is possible to manage a simulator event competently without fully internalising how automation modes interact under degradation.

This is not an accusation of deficiency. It is an acknowledgement of a challenge in structural assessment.

How do we verify the robustness of internal mental models rather than the fluency of procedural execution?

Stability and its limits

In routine operations, the system appears stable. Automation performs predictably. Crews manage expected variations competently. Assessment data suggests proficiency.

But resilience rarely reveals itself in stable contexts.

It emerges at the edge, where cues conflict, where automation behaves in partially degraded ways, where no checklist title neatly captures the unfolding situation. These are not necessarily failures of knowledge. They may be tests of interpretation, projection, and trust calibration.

If assessment primarily captures performance within structured scenarios, what does that tell us about cognitive flexibility under genuinely emergent conditions?

Are we measuring resilient cognition or resilient familiarity?

The digital-native intersection

The generational dimension complicates the picture further. If cognitive habits are shaped in environments where information is retrievable on demand, how does that influence performance when time pressure restricts retrieval? If digital ecosystems encourage rapid task-switching, how does that interact with sustained monitoring demands in automated cockpits?

Conversely, might digital-native cognition confer strengths such as comfort with abstraction, interface fluency, and rapid pattern recognition across layered displays, thereby enhancing adaptation?

The question is not whether one generation is stronger than another. It is whether training and assessment frameworks have sufficiently evolved to reflect the cognitive realities of those entering the system.

Not Recognised – Not Understood

Training necessarily draws on history. Case studies, known failure modes, and established procedural frameworks form the backbone of aviation safety.

Yet resilience is most severely tested in what cannot be fully anticipated.

The “Not-Recognised -Not Understood” is not an abnormal checklist. It is an emergent convergence of system behaviour, human interpretation, workload, and social dynamics. In such moments, procedural recall may provide scaffolding, but adaptive reasoning becomes decisive.

These raise searching questions:

• Does modern assessment distinguish between procedural fluency and conceptual robustness?
• Can behavioural markers reliably indicate internal model integrity?
• How do we evaluate resilience in contexts that cannot be fully scripted?
• Does explicit system understanding matter, or is tacit expertise sufficient?
• How might generational cognitive conditioning influence performance at the edge of ambiguity?

An open research frontier

Work on the Crew Cognitive Resilience Index (CCRI™) attempts to explore these tensions empirically rather than rhetorically. By examining constructs such as situation awareness, decision accuracy, adaptive performance, cognitive load, and trust calibration, the aim is not to critique existing systems, but to better understand how resilience manifests in contemporary operations.

Yet CCRI™ must remain open to revision. If resilience proves to be more distributed, more tacit, or more context-sensitive than current models assume, then measurement approaches will need to evolve accordingly.

The central issue is not whether pilots today are competent. They demonstrably are. 

The deeper issue is whether our assessment paradigms are sufficiently sensitive to the kind of cognition modern, automated, digitally mediated flight operations now demand.

As aircraft systems continue to evolve and as digitally conditioned generations assume command, resilience may no longer be inferred solely from procedural performance. It may need to be examined precisely where familiarity ends and ambiguity begins.

That terrain remains open.