Post-roll effects on attitude perception: "the Gillingham Illusion"

Several aircraft each year are lost because of an unexplained collision with the ground. The attitude of most of these aircraft prior to impact was nose-low and with excessive bank, i.e., greater than 90 degrees . Prior to these accidents, each aircraft was noted as either changing heading or making an abrupt roll.

Could there be some underlying tendency for the pilot to make unnoticed stick inputs after completing a roll from one bank angle to another?

Since ground-based flight simulators cannot create the true sensation of rolling an aircraft from one side to the other, the instrumented CALSPAN NT-33 aircraft was used for this study. Six pilots were given a series of three roll rates and two head positions while the aircraft automatically changed bank from 45 degrees of bank in one direction to 45 degrees of bank in the opposite direction. The subject's view of the external visual scene was restricted with a blue-amber vision restricting transparency combination. All attitude-indicating instruments were blanked, requiring the subjects to make stick inputs based on their vestibular (somatosensory) feedback.

Subjects experienced a consistent tendency to increase bank angle after given control of the aircraft immediately following the roll maneuver, while thinking they were maintaining a constant bank angle. In some cases, the pilots rolled the aircraft completely inverted.

When pilots rely on their perception of bank, following a roll, they will inadvertently increase their bank in the direction of the previous roll.