Repeated low-priority alerts create a conditioned reflex: alert fires, pilot resets, no consequence follows. Over 4-6 exposures, the reset action migrates from prefrontal cortex (deliberate) to basal ganglia (automatic).

Each interrupt costs attentional resource. Vigilance decrement research (Mackworth, 1948) shows sustained attention degrading after 20-30 min; repeated false alarms accelerate the curve to under 10 min.

Dixon & Wickens (2006): when alarm false-alarm rate exceeds 40%, operators begin treating all alarms as false. Reaction time decreases BUT appraisal drops to near-zero - speed is not safety.

The reset sequence (see alert, reach, press) becomes a learned motor program. Under high workload or fatigue, the motor program fires before the frontal lobe appraises the stimulus content.

Short final (< 500 ft AGL) is the highest-workload segment. Attentional tunneling is already active. Any novel alert must break through a heavily loaded attention bottleneck.

0.17s reset is physiologically impossible as a deliberate response (minimum deliberate RT: ~250ms). The action was purely procedural memory. The pilot did not read "ENG 1 FIRE" before pressing reset.

Fire handle not pulled. Engine continues to burn. At < 300 ft AGL, insufficient altitude for ejection. Outcome: loss of aircraft and crew. Root cause: alert system design, not pilot error alone.

Countermeasures: (1) tiered alarm architecture with forced-read for Warnings. (2) Distinct audio encoding by criticality. (3) Copilot challenge protocol for all red-level events. (4) MFOQA alarm logging.