Giving A Hand To Pilots With Animated Alarms Based On Mirror System Functioning

• ¹ Institut Supérieur de l'Aéronautique et de l'Espace (ISAE-SUPAERO), Department of Aerospace vehicles design and control - DCAS, France

INTRODUCTION Controlled Flight Into Terrain (CFIT) accidents are among the most frequent and deadly accidents in aviation (IATA, 2016). Pilots have sometimes only a few seconds to react to the ‘Pull-Up’ alarm that indicates an imminent collision with the ground. The auditory modality of the alarm can generate stress, which can disrupt cognitive processes (Porcelli et al., 2008) and negatively impact the ability to take timely and appropriate actions (Scholz et al., 2009). While in general humans are less sensitive to auditory stimuli than to visual stimuli (Mrugalska et al. 2016), the auditory component of the ‘Pull-Up’ alarm is loud and prominent whereas its visual component is more discreet (i.e., small text). Consequently, a first step to enhance the efficiency of this alarm would be to increase the conspicuity of its visual component by enlarging the ‘Pull-Up’ text size and to increase its salience by displacing it from the artificial horizon to the bottom of the Primary Flight Director (PFD). However, enlarging the inscription may not be sufficient to cope with the loss of cognitive performance due to stress (BEA, 2009). Previous researched found that watching an action activates premotor cortex and pre-initiates the gesture imitation through the mirror neuron system (Ocampo & Kritikos, 2011). Therefore, presenting a video of the gesture to perform (i.e., a hand pulling the sidestick), in case of imminent impact with the ground, should pre-activate the motor reaction to the alarm and then improve the pilots’ performance in terms of reaction time. The present experiment aimed at assessing whether these ‘Pull-Up’ video better performance than more conventional ‘Pull-Up’ alarms. Both behavioral and electroencephalographic (EEG) results were used to assess the best alarm design. MATERIAL & METHODS Twenty-five pilots participated in the present study which was composed of two parts. The first part aimed at comparing the reaction times of pilots to three different ‘Pull-Up’ alarms: an usual ‘Pull-Up’, an enlarged ‘Pull-Up’ inscription, and the ‘Pull-Up’ video (see respectively Figure 1.A., 1.B. & 1.C.) during a 30 minutes scenario in a flight simulator. Each alarm type was presented three times. Volunteers were instructed to pull the stick as fast as possible each time one of the three ‘Pull-Up’ alarms was displayed on the PFD. To ensure that pilots were not only focused on the PFD, they were repeatedly distracted with several events such as autopilot or auto-thrust disengagements. Moreover, control videos (i.e. fillers) were also randomly displayed on the PFD to avoid automatic reactions to the mere occurrence of a stimulus. The second part of the experiment aimed at measuring electrophysiological responses to the two visual alarms with the fastest reactions in order to identify the more efficient in terms of neurophysiological arousal. The participants were equipped with a 64-channel EEG to examine mirror neuron system activity and were asked to passively (to avoid muscular artefacts) observe the enlarged ‘Pull-Up’ inscription and the ‘Pull-Up’ video 30 times each. The desynchronization of both the mu (i.e., motor alpha, 8-12Hz) and the beta (13-25Hz; Babiloni et al., 2002) rhythms were investigated at usual central sites (i.e., C3, C4; Stancák Jr. & Pfurtscheller, 1996). Visual inspection of the data revealed alpha desynchronizations at parieto-occipital sites that were statistically assessed. RESULTS Simulator Experimental Part. A 3 (Alarm [Usual ‘Pull-Up’; enlarged ‘Pull-Up’; ‘Pull-Up’ video]) one-way ANOVA was conducted on the mean reaction times. Pilots were longer to react to the usual ‘Pull-Up’ [F (1, 24) = 16.97, p < .001, ηp² = .52; M = 2.96s ± 1.5] than to both the ‘Pull-Up’ video (M = 1.60s ± .30) and enlarged ‘Pull-Up’ (M = 1.47s ± .31; Figure 2.A.). Electrophysiological Part. Two 2 x 2 (Electrode [C3; C4]) x Video [enlarged ‘Pull-Up’; ‘Pull-Up’ video]) ANOVAs were conducted on both mu and beta rhythm powers respectively on 0-3000ms and 300-1300ms time windows on electrodes C3 and C4. The results showed a greater desynchronization of both mu [F (1, 24) = 13.38, p < .001, ηp² = .36; Figure 2.B.] and beta [F (1, 24) = 5.56, p < .05, ηp² = .19; Figure 2.C.] rhythms in response to the ‘Pull-Up’ video alarm (mu: M = - 30.79dB ± 2.69; beta: M = - 13.52dB ± 3.04) compared to the enlarged ‘Pull-Up’ alarm (mu: M = - 28.31dB ± 3.91; beta: M = - 11.49dB ± 3.70). A 3 x 2 (Cluster [right cluster: PO7/PO3/O1; central cluster: Pz/POz/Oz; left cluster: PO8/PO4/O2] x Videos [enlarged ‘Pull-Up’; ‘Pull-Up’ video]) ANOVA was also performed on alpha rhythm measurements. Greater desynchronization was found in response to the ‘Pull-Up’ video (M = - 30.27 ± 2.97) compared to the enlarged ‘Pull-Up’ [M = - 27.04 ± 3.56; F (1, 24) = 24.81, p < .001, ηp² = .51; Figure 2.D.] DISCUSSION The aim of the present study was twofold: 1) testing if the enhancement of the visual component of the ‘Pull-Up’ alarm improved the reaction of pilots in a situation of emergency and 2) ascertaining which visually improved alarm (i.e., enlarged ‘Pull-Up’; ‘Pull-Up’ video) triggered the greater neurophysiological arousal. The results disclosed that the enlarged ‘Pull-Up’ and the ‘Pull-Up’ video led to a drastic reduction of reaction times in comparison to the usual alarm. This result is encouraging as in both cases, the reaction time dropped below the critical threshold of 2s (Causse et al., 2012). While no reaction time difference was found between the enlarged ‘Pull-Up’ and the ‘Pull-Up’ video, the results showed significant differences in the EEG responses to the two alarms. Greater neurophysiological arousal was found for the ‘Pull-Up’ video compared to the enlarged ‘Pull-Up’ with increased motor preparation (i.e., greater mu and beta desynchronizations at central sites) and more attentional resources allocated (Klimesch, 2012) to the video compared to the enlarged inscription (i.e., greater alpha desynchronization at occipital sites). Taken together these results demonstrate that gesture alarms could significantly increase the pilots’ performance in emergency situations and in that way improve flight safety.