Abstract

This thesis investigated susceptibility of the human brain to auditory signals under different (driving) conditions and the ability of humans to act on alerts in specific contexts In Part I, three experiments were conducted to answer the main question: “How susceptible are humans to auditory stimuli?” To answer this question, we measured electroencephalogram (EEG) Event-Related Potential (ERP) responses in an oddball paradigm in (automated) driving tasks and cognitive load inducing tasks. The main finding was that automated driving and cognitive load inducing tasks can both reduce susceptibility to auditory alerts. Given this observation that auditory susceptibility is sometimes reduced and, more generally, to explore ways to support accurate human behavior while driving, in Part II, we explored interventions that might alleviate the associated potential problems. In the first experiment, we introduced early warnings, or pre-alerts, to allow drivers to anticipate a later acute imperative alert for an upcoming transition of control from the car to the human driver. The results showed that pre-alerts can support safer transitions of control. In the second experiment, we tested an intervention where visual in-car warnings were presented to indicate an upcoming, necessary lane change. Results showed that, on average, drivers were able to change lanes in time. However, at the same time, there are occasionally some drivers that did not respond in a timely manner.

Abstract

International audience; The acoustic environment is critical in Air Traffic Control (ATC), as operators exchange information with planes and must also be aware of the occurrence of auditory alarms. In such situations, observing inattentional deafness is likely. In this study, we aimed to identify the physiological indicators of inattentional deafness through the analysis of the P300 evoked potential, known to be an indicator of attention allocation, an important step to a stimulus reaching consciousness. Based on the assumption that the high mental load generated by an ATC task may reduce the alarm detection rate, we wished to test whether this effect would be reflected in the alarm-evoked P300 amplitude. Participants had to perform simulated ATC tasks within the LABY microworld while electroencephalographic (EEG) activity was recorded. Simultaneously to the LABY tasks, participants were asked to respond to target tones (the “alarm”) and to ignore standard tones. Behavioral results showed that 4.6% of alarms were not reported. For these alarm detection failures, the EEG analysis showed a diminution of the P300 amplitude in comparison to a control condition in which participants only focused on the tones. These results suggest that the P300 amplitude seems to be a valid physiological indicator of vulnerability to inattentional deafness in complex environments. Relevant applications include the prevention of alarm omission and the assessment of warning designs.