社会与工程心理学研究室知识库

危险驾驶行为是事故的最直接因素，严重威肋、交通安全。己有研究表明，愤怒(状态及特质)会导致危险驾驶行为增加，影响交通事故的发生和严重程度。但是，对愤怒影响驾驶行为的心理机制及消除不良后果的干预方法缺乏深入系统的研究。本研究从认知加工的角度，针对愤怒对驾驶行为具体过程的影响展开系统研究，揭示愤怒影响下驾驶过程中的认知和行为的变化规律，并在此基础上探索愤怒状态的识别和干预方法。本论文共包括以下四个研究:

研究一探究愤怒与自我报告的驾驶行为之间的关系。依据一般攻击性行为理论，通过问卷调查，考察了愤怒特质、驾驶愤怒、驾驶愤怒想法和危险驾驶行为产生之间的关系。研究结果显示愤怒特质对危险驾驶行为的影响受到驾驶愤怒的不完全中介;驾驶愤怒通过影响驾驶员对情境的认知从而影响危险驾驶行为。该结果说明，驾驶员感知环境信息后并产生一定的情绪，并对该情境形成一定的认知评价，最终产生相应的驾驶行为，整个过程均受到人格特质的影响。

研究二探究愤怒情绪对驾驶员在两类典型场景一一跟车(车与车交互)、行人过马路(车与人交互)一一中的相关能力及表现的影响。采用3(愤怒、快乐、中性)X2(高、低特质愤怒)的混合实验设计。研究二(A)要求驾驶员回忆情绪性或中性事件后，完成跟车驾驶任务。结果表明:相比于中性条件，愤怒和快乐情绪均使驾驶员刹车反应时增加、最小碰撞时间减小;此外，驾驶员在快乐情绪下感知到的风险显著低于中性和愤怒条件。研究二(B)要求驾驶员观看情绪绪或中性电影片段后，完成行人避让任务。结果表明:相比于中性和快乐情绪愤怒情绪增加了驾驶员从行人前方绕行的概率，并且速度更快、遇到行人时的最小速度更大、与行人的横向距离更小。

研究三探究驾驶过程中的情绪性状态及危险驾驶行为识别。通过采集并提取研究二中驾驶员在模拟驾驶任务过程中的脑电信号频域特征，采用支持向量机分类器及粒子群寻优算法对愤怒状态的危险驾驶行为进行识别。结果显示:驾驶过程中愤怒情绪的识别正确率为87.26% ， ROC曲线下面积为0. 94;愤怒情绪下危险跟车行为的识别正确率为83.08% ， ROC曲线下面积为0.86 。

研究四探讨如何通过放松音乐减少愤怒情绪对驾驶表现的负面影响。两组被试均要完成中性唤起后模拟驾驶和愤怒唤起后模拟驾驶两个阶段，中性驾驶条件下两组被试均不听音乐，愤怒驾驶条件下干预组被试听预先选取的放松音乐而控制组被试不听音乐。结果显示:跟车驾驶任务中，干预组比控制组的刹车反应时更短、最小碰撞时间更大;行人过马路任务中，干预组比控制组的驾驶平均速度更小、人车横向距离更大，感知到的体力需求更小。

综上，本研究通过分阶段和交互式的研究范式，系统地探讨了愤怒对驾驶行为影响的认知一行为机制，即愤怒不仅会减慢驾驶员在紧急事件中的反应能力，而且会促使其选择更加风险的行为模式。这些危险的行为变化能够基于愤怒情绪下脑电活动特点进行有效地识别，并且可以通过放松音乐进行缓解。本研究建立了一套基于愤怒后驾驶行为的认知一行为一识别一干预的研究模式，相关的研究成果不仅可以深化愤怒对驾驶行为影响的理论认识，同时为驾驶安全辅助系统设计提供理论和思路。

Dangerous driving behavior is the direct factor of accidents and seriously threatens traffic safety. Studies had shown that anger (state and trait) could lead to risky driving behavior, which affected the occurrence and severity of traffic accidents. However, the mechanism of anger affecting driving behavior and the intervention methods to eliminate adverse consequences should be explored systematically. From the perspective of cognitive processing, this study systematically throw light on the influence of anger on the specific process of driving behavior, revealed the changes of cognition and performance of driving under the influence of anger, and explored the identification and intervention of anger state on this basis. This paper includes the following four studies:

Study 1 explored the relationship between anger and self-reported driving behavior. Based on the theory of general aggressive behavior, a questionnaire survey was conducted to examine the relationship between the anger trait, driving anger, driver's angry thoughts, and dangerous driving behavior. The results showed that the impact of anger trait on dangerous driving behavior was incompletely mediated by driving anger; driving anger affected dangerous driving behavior by affecting the driver's angry thoughts. It indicated that drivers generated emotions and a certain cognitive evaluation after perceiving the environmental information, which provided basis for the final driving behavior. Furthermore, the whole process was affected by personalities.

Study 2 explored the impact of anger on the driver's capacities and performance in two typical scenarios, car following task (vehicle-to-vehicle interaction) and pedestrian crossing task (vehicle-to-human interaction), using 3 (emotions: angry, happy, neutral) X 2 (trait anger: high, low) mixed experimental design. Experiment 1 required participants to complete the car-following task after recalling the emotional or neutral event. The results showed that both anger and happiness increased drivers' brake response time and the decreased minimum time to collision being compared with the neutral condition. In addition, the driver's perceived accident risk under happy state is significantly lower than the other conditions. Experiment 2 required participants to complete the pedestrian一crossing task after viewing the emotional or neutral film clips.

The results showed that the probability of the driver passing in front of the pedestrian was higher, the speed was faster, the minimum speed when encountering a pedestrian was greater, and the lateral distance from the pedestrian was smaller in anger condition than neutral and happy conditions.

Study 3 explored identification of the angry state and prediction of dangerous driving behavior during driving. By collecting and extracting the frequency domain features of electroencephalogram signal during simulation driving task of Study 2, The results the Support Vector Machine (SVM) classifier and Particle Swarm Optimization (PSO) algorithm showed that the accuracy of anger recognition during driving was 87.26%, and the area under the ROC curve was 0.94; the accuracy of risky car-following behavior recognition was 83.08%, and the area under the ROC curve was 0.86.

Study 4 explored how to reduce the negative impact of anger on driving performance by relaxing music. Both groups were required to complete the simulated driving after the neutral and angry arousal. Under the neutral condition, the two groups did not listen to the music while driving. Under the angry condition, the intervention group listened to the relaxing music while driving, however the control group did not listen to music. The results showed that, in the car-following task, the intervention group had shorter braking response and longer time to collision than the control group; in the pedestrian-crossing task, the intervention group had smaller average driving speed and lateral distance to the pedestrian and perceived lower physical load than the control group.

In summary, this study established a phased and interactive experimental paradigm of anger impact on driving behavior, and systematically revealed the behavior-cognitive processing mechanisms of anger on driving behavior, that is, anger not only impairs driver's response capacity in an emergency, but also leads to more risky behavioral pattern. It can be effectively identified based on the features of EEG activity under rage, and can be alleviated by relaxing music. This study establishes a cognitive-behavior-recognition-intervention research paradigm aiming at driving behavior under rage.Relevant results could deepen the theoretical understanding of the anger influence on driving behavior, and provided theory and thinking for the design of driving safety assistance system.