In previous studies it was shown that both pilots and non-pilots sometimes make roll reversal errors (RREs) when the aircraft bank angle is presented on a moving horizon type attitude indicator. This incorrect input has been shown to be caused by a misinterpretation of the horizon symbol on such an attitude indicator. The figure-ground relation of the aircraft symbol and horizon symbol on the attitude indicator has been cited as being a contributing factor to this misinterpretation. A Multi-Layer Display (MLD) is tested as a possible intervention tool for these interpretation errors. With this display type, the horizon symbol is presented on a physically deeper layer, making it possibly more easily interpreted as being in the background, and the aircraft symbol is presented on a second layer on the foreground. A group of 23 non-pilots and 18 pilots were tasked to perform a recovery task on a desktop-based simulator without outside visuals and only an Attitude indicator on both a MLD as well as a conventional Single Layer Display for repeated-measures. Although roll reversal errors were observed at a comparable level to previous research, no significant difference were observed on the error rate between the two display types. A possible explanation is that the stereoscopic depth of the MLD dissipates when the subject sits still and remains exactly in front of the display. This leaves open the possibility that other types of depth cues, such as monocular cues, can be added and used to enhance the figure-ground relation in displays.

In previous studies, pilots made roll reversal errors (RREs) when responding to a ‘moving- horizon’ type attitude indicator (AI). It was argued that it was the ambiguity of this display leading to RREs. Later, using non-pilots, it was found that RREs were in many cases caused by expectation-induced misperception of the AI bank angle, which can arise from a pilot experiencing spatial disorientation. The current study evaluated the role of expectation in the control strategy of professional pilots using a hexapod simulator. First, an effective man-in-the-loop spatial disorientation scenario was developed, based on the ‘leans’ illusion. Here, the expectation of ten experienced and eight novice pilots was manipulated with the simulator motion without the AI, after which they had to turn the aircraft back level when the AI was shown again. An error was made in 16.7% of the runs where the bank angle on the AI was opposite to the earlier given physical motion cues, which is 2.9 times more compared to a baseline condition in which no motion, and thus no manipulated expectation, was present. In a third condition where a motion-induced expectation of the bank angle was present, but a level AI was shown later, no errors were made. It was also found that experienced pilots made slightly more errors than novice pilots and that their reaction time was on average half a second higher.
It was concluded that an initially induced expectation about the bank angle does not directly influence the control strategy of pilots, but that it does make them more vulnerable to misinterpretations of the AI, leading to RREs being made. This should be taken into account when developing new displays and technologies, as it should at all times guide pilots in their decision-making, minimizing the chance of misinterpretations.