AI and Gaze Tracking in Aviation Psychology

Human factors aviation psychologists are combining cockpit gaze-entropy tracking with cognitive-workload artificial intelligence (AI) to prevent deadly pilot errors during instrument flight emergencies. By analyzing a pilot’s eye movements and biometric data in real time, these systems detect when a pilot is experiencing the startle effect or mode confusion—two leading causes of loss-of-control accidents. When the AI identifies a breakdown in situational awareness or an impending cognitive overload, it dynamically adapts the cockpit displays and alerts to guide the pilot’s attention back to critical flight parameters, ensuring safe and decisive recovery actions.

During instrument flight conditions, pilots rely entirely on cockpit displays rather than outside visual references. When an unexpected emergency occurs, the sudden spike in stress can trigger the startle effect. This physiological response impairs the prefrontal cortex, temporarily freezing decision-making abilities and inducing tunnel vision. Simultaneously, modern aircraft rely on complex automated systems. Under high stress, pilots frequently suffer from mode confusion, a dangerous state where they misunderstand what the aircraft’s automation is currently doing or failing to do.

To monitor for these vulnerabilities, aviation psychologists implement non-intrusive eye-tracking cameras inside the cockpit to measure gaze entropy. Gaze entropy is a mathematical calculation of the randomness or predictability of a pilot's visual scan pattern. Under normal instrument flying conditions, a well-trained pilot exhibits a structured, predictable scan across primary flight displays, navigation, and engine instruments.

When an emergency strikes, gaze entropy shifts dramatically to one of two extremes:

• Extremely Low Entropy: The pilot’s gaze becomes hyper-fixated on a single instrument or warning light, indicating cognitive tunneling and the startle effect.
• Extremely High Entropy: The pilot’s eyes dart erratically across the flight deck without processing information, signaling panic, high stress, and severe mode confusion.

Gaze data alone is not enough to fully understand a pilot's mental state, which is where cognitive-workload AI steps in. Machine learning algorithms are trained by aviation psychologists to fuse real-time gaze-entropy metrics with flight deck interactions and physiological data, such as heart rate variability, pupil dilation, and respiration rates. By comparing this live biometric stream against the current phase of flight and active aircraft malfunctions, the AI accurately calculates the pilot's instantaneous cognitive workload.

When the AI detects a dangerous combination of abnormal gaze entropy and overwhelming cognitive workload, it initiates adaptive automation to break the error chain. Instead of bombarding a startled pilot with more auditory alarms and flashing lights—which only worsen cognitive overload—the system simplifies the environment.

To counter the startle effect, the system dims non-essential screens and uses clear, visual cues on the primary flight display to pull the pilot out of tunnel vision. It directs their eyes straight to the most critical flight data, such as attitude and airspeed.

To eliminate mode confusion, the AI predicts which automated mode the pilot is misinterpreting based on what instruments they are ignoring. The system then highlights the active automation state in plain language or provides a direct, unambiguous voice prompt explaining what the aircraft is doing and what manual input is required. By closing the loop between human psychology and aircraft automation, these systems ensure pilots maintain control during the most chaotic flight scenarios.