Human Factors and Cognitive Load in the Control Room

Designing the fusion control room around the limits of human attention and working memory so operators stay effective during the worst minutes, not just the calm ones.

Design for the worst minute

A control room that is comfortable during steady operation can still fail during a disruption countdown or a plug-density excursion, because that is when information rate, stakes, and stress all peak. Kronos L6 is designed against that worst minute. The governing principle is that human working memory and attention are hard limits, and every display either respects them or degrades operator performance exactly when it matters most.

Managing intrinsic and extraneous load

Cognitive load splits into intrinsic (the irreducible difficulty of controlling a fusion plasma) and extraneous (load added by poor presentation). L6 cannot lower the intrinsic difficulty of holding the breeder's negative-triangularity shape or the burner's ambipolar potential, so it attacks extraneous load ruthlessly: consistent color grammar, stable layouts, geometry-anchored data, and a single rationalized alarm stream instead of scattered indicators.

• Stable layouts: elements do not move between modes, so muscle memory holds
• One color grammar: a color never means two things across screens
• Geometry anchoring: data appears where it physically lives, not in a table to decode
• Progressive disclosure: summary first, detail on demand, never all at once
• Alarm rationalization: one ranked stream, not a wall of equal reds

Vigilance and automation complacency

Two failure modes are designed against explicitly. Vigilance decrement: humans miss rare events during long calm periods, so the interface uses the predictive shadow and precursor lead times to re-engage attention before an event, not after. Automation complacency: humans over-trust automation, so uncertainty and provenance are always visible and the interface actively flags when the AI is out of its depth.

Load is measured, not assumed. In the training simulator, operator load is estimated (task rate, error rate, response latency, optionally physiological proxies) and layouts that spike load during scripted upsets are revised. See control-room layout, attention and notification design, and training-simulator mode.