Dissecting Driver Behaviors Under Cognitive, Emotional, Sensorimotor, and Mixed Stressors

From the time of Henry Ford, the car has been powering the world economy and defining the modern way of life1. Every morning millions of cars take to the highways transporting commuters from their suburban homes to metropolitan business centers; late in the afternoon the same cars transport the same people back to their places. In these daily commutes, thousands of car accidents incur a significant material and human cost while reducing productivity. It is telling that rush hour traffic reports on radio are a staple of metropolitan areas. The focus of these short segments is on the accidents that took place and how these accidents are affecting traffic flow. Highway driving is a challenging task. Human physiology responds to challenges through sympathetic arousal. To cope successfully with the task at hand, a measured sympathetic response is required - too little or too much is an invitation to failure2,3. In some instances elevated arousal is precipitated by a secondary task, which deprives resources from the main task; these cases are suspected to be the most prevalent cause of commuter traffic accidents4. Texting is a well-known example of a secondary task antagonistic to driving; it is a sensorimotor stressor, where the driver needs to move her/his eyes and one hand between the car’s controls and the smartphone all the time. Most other types of antagonistic stressors are cognitive or emotional in nature5. There has been little work about the distracting effect of each stressor category. Typically, the studies reported in the literature refer indirectly either to a sensorimotor stressor or to an unspecified mix of stressors6. The problem stems from poor abstraction, with most experimental designs centered on devices rather than stressor types. A driver may employ no device, appearing to be concentrated on the driving task, and still be under a hidden stressor (i.e., cognitive or emotional) that is potentially as distracting as an apparent stressor (e.g., texting). The lack of clearly abstracted studies on driving distractions has as a result partially informed regulations; there are traffic laws that ban texting while driving but no consideration has been given to driving under cognitive or emotional distress. Should cognitive and emotional distractions prove damaging to driving behaviors, the problem is how do authorities regulate this? It is difficult to prove in a court of law somebody’s inner thoughts or feelings, the intensity of which many times the subject herself/himself underestimates. This brings to the fore the need for methods to sense the sympathetic and driving effect of various types of stressors, enabling timely orthotic actions. Here we report results on a driving simulator experiment where subjects operated under normal and stressful conditions; the stressful conditions featured four types of stressors - cognitive, emotional, sensorimotor, and mixed. We delivered the cognitive and emotional stressors through appropriate oral questionnaires. We delivered the sensorimotor stressor in the form of texting while driving. For mixed stressors or absence thereof, subject behaviors were measured at two time scales - sustained engagement without exogenous surprises versus reactions to startling events. We used instantaneous perspiration at the perinasal area as proxy of the subject’s sympathetic state7, thus forming the study’s explanatory variable. We used instantaneous steering angle and maximum right-side/left-side lane departures as proxies of the subject’s steering and driving performance, respectively, thus forming the study’s response variables. The perinasal perspiration signal was extracted through facial thermal imaging according to the method reported by Shastri et al.8. The steering angle and lane departure signals were recorded by the driving simulator. The response variables were abstracted per the model introduced by Pavlidis et al.3. The purpose of this model is to establish causality links between explanatory and response variables by differentiating between reactionary response, driven purely by sympathetic arousal, and error response, that is, the performance outcome as adjusted by a number of other factors. In the present study, instantaneous steering angle indicates raw reaction, while maximum right-side/left-side lane departures indicate driving performance. The former is precipitated by sympathetic arousal, while the latter is the result of higher-order modulation over the initial steering responses. We hereby studied a comprehensive set of stressors through a three-level causal decomposition (arousal → reaction → error) in long/uneventful versus short/eventful time scales, arriving at intriguing conclusions about driving behaviors. These conclusions contribute to open questions in neuroscience regarding conflict resolution9, while they stand to inform human-car interfaces, tying them to traffic regulations of the future.