Abstract
Aggressive driving behavior is one of the main reasons for traffic crashes in China. However, how cognitive interventions affect impulsive driving behavior is still unknown. In this study, a simulated drive was constructed to evaluate the influence of different cognitive interventions. In addition to speeding behavior in limit zones, speed when passing pedestrians, at intersections, and on the whole drive was adopted as a measure to evaluate aggressive driving behaviors. Forty-eight young drivers were recruited with monetary rewards. Compared to the control group, the penalty feedback intervention significantly reduced the mean speed in the 40 km/h zone, the 80 km/h zone, and when passing pedestrians. The combined feedback intervention significantly reduced the distance ratio of speeding in the 40 km/h and 80 km/h zones, as well as the mean speed at the intersection and on the whole drive. However, the interaction effects between the driving task and intervention method were not remarkably observed during aggressive driving behaviors. The findings from this study provide evidence that different cognitive interventions contribute distinctively to improving aggressive driving behaviors. These conclusive results have possible implications for the design of vehicle warning systems and traffic safety interventions.
1. Introduction
Around the world, approximately 1,350,000 people die from traffic accidents yearly, and traffic crashes are the leading cause of death for young people aged 5–29 [1]. In several developed countries, fatal crashes have increased substantially. For example, road fatality data for 2020 compared to the average of 2017–2019 rose by 5.1% in America and 4.6% in Switzerland [2]. Similarly, in China, road traffic safety is still facing a grim situation, posing a serious threat to the sustainable development of the social economy, as well as people’s lives in general. In this context, the number of fatalities and injuries in traffic accidents was about 290,000, resulting in a direct economic loss of almost 1.4 billion yuan in 2019 [3]. Moreover, overspeeding is not yielding to pedestrians, and red-light running behaviors are common traffic violations in any country [4]. Exceeding the posted speed limit or driving too fast for conditions contributes to about 70% of road crashes [2]. Furthermore, not yielding to the pedestrian red light, running, and other risky driving behaviors result in fatal crashes throughout the world. Several reasons explain these traffic rule violations, such as the driver’s aggressive nature [5], time pressure [6], and others.
Aggressive behavior is one of the major reasons behind the increase in road fatalities, particularly for young drivers, and is generally considered to be speeding, running red lights, and not yielding to pedestrians [7, 8]. This phenomenon is explained using the dual-systems theory, involving both cognitive control and socioemotional systems as distinct neurobiological subsystems [9, 10]. Cognitive control refers to the ability to coordinate thoughts and actions in accordance with short- and long-term goals and in response to changing environmental demands. The two-system model suggests that the main reason why young people tend to engage in risky behaviors is that their cognitive control ability is extremely vulnerable to control impulses, resulting in aggressive behavior on the roads.
External interventions to reduce young people’s traffic violation behaviors are considered efficient. In this respect, scholars mainly adopt two approaches to reduce aggressive driving behaviors of young drivers: the indirect intervention method through improving the inhibitory control ability and the direct intervention method through cognitive intervention. This method of reducing risky driving behaviors by improving cognitive control ability is based on the assumption that poor impulse control ability is the main factor leading to risky driving behaviors of young drivers [5, 11–14]. Multiple scholars have studied this topic, but as the studies on the neural mechanism of cognitive control are still in their infancy, little is known about whether this function is improvable through training and whether the underlying brain neural network could be remodeled [15]. Generally speaking, “remodeling” refers to the improvement of behavior and brain tissue, including the acquisition of new skills, the improvement of existing skills, and the recovery of functional defects. Hatfield et al. [16] conducted an in-depth study on improving driving performance through inhibition control training. A total of 65 young drivers aged between 16 and 24 years were selected as experimental samples to investigate whether response inhibition training is able to reduce risky driving behaviors such as speeding, dangerous crossing, and traffic violations among young drivers. The experiment was divided into two stages, each of which included response suppression training and a driving performance evaluation. However, after several days of training in an adaptive manner, it was found that the improvement of the simulated driving performance was not significant, and increasing the training time could not lead to the improvement of the simulated driving performance.
Cognitive intervention focuses on improving and mastering the cognitive skills needed for safe driving, including hazard perception, hazard prediction, attention, maintenance, risk management, and others [17]. Among them, feedback and self-explanation play a positive role in improving driving behaviors. In this setting, feedback refers to the provision of information that may lead to changes in the system or process, while effective feedback is capable of promoting the system’s positive development by providing past information. Self-explanation focuses on explaining one’s behavior. Its advantage is that one can understand the reason why an individual implements a certain behavior. In driving studies, Reagan et al. [18] tested the effectiveness of real-time feedback from alarm systems and financial incentive systems in reducing speeding behavior. Fifty subjects were tested on a real car for four weeks. The results show that the incentive system is able to reduce the speeding time significantly, while the feedback system is not as effective. Molloy et al. [19] explored the influence of different types of feedback on the speed management behavior of young novice drivers through driving simulator experiments. It was found that all four types of feedback improve the speed management behavior of young drivers significantly, compared to a control group without intervention. Molloy et al. [20] compared the effects of two cognitive interventions, feedback, and self-explanation, on the improvement of the speeding behavior of young drivers through a real vehicle experiment. Seventy-five young drivers aged 18–25 were divided into three groups: an auditory alarm feedback group, a combined feedback group, and a self-explanation group. Results show that the combined group feedback in both low-speed and high-speed zones are better in terms of average speed and speeding ratio, and the auditory alarm feedback method is not significant.
Despite the amount of literature on the effect of cognitive intervention on speed management while driving, several points should be addressed before the research evidence is applied to accident prevention. On the one hand, more indicators that represent aggressive driving behavior—not just speeding behavior—are required for verification to reveal the influencing mechanism of cognitive intervention. On the other hand, the effect of cognitive intervention on aggressive driving behavior might vary between countries, such as traffic safety awareness. Hence, the social and traffic environments could be culturally dependent [21, 22]. Most of the previous studies were conducted in developed countries, such as America [18] and Australia [19, 20], with few studies in China. Evidence from these other countries might not apply to China due to differences in cultural, social, and traffic environments.
Thus, this study examines whether cognitive interventions are able to improve aggressive driving behaviors among Chinese drivers by measuring further indicators. Performance feedback, performance and penalty feedback, performance and safety feedback, combined feedback (performance, penalty, and safety feedback), and self-explanation were adopted as intervention methods. According to previous studies, two hypotheses were suggested: the first was that the two cognitive intervention methods contributed distinctively to aggressive driving behavior, while the second was that driving tasks and intervention methods had a significant interaction effect on aggressive driving behaviors. This study provides evidence that different cognitive interventions contribute distinctively to improving aggressive driving behaviors. The results have possible implications for the design of vehicle warning systems and the intervention in aggressive driving behaviors.
The remainder of this paper is organized as follows: Section 2 presents the methods of the study, including participants, apparatus, experimental design, and data analysis. Section 3 outlines the results of the experimental study. The discussion and conclusions are presented in Section 4. Finally, Section 5 provides limitations and future research opportunities.
2. Methods
First, the requirements and statistics of the participants in different groups are presented. Second, apparatus used in this experiment is described. Thirdly, an experimental design including a simulated driving task, cognitive intervention methods, and an experimental process are provided. Finally, the experimental data were analyzed and the results are given, as seen in Figure 1.
2.1. Participants
A total of 48 drivers aged between 22 and 30 were recruited as subjects (mean = 24.98, SD = 1.804), including 30 males and 18 females. Participants were required to have a valid license and 1 year of driving experience at least (mean = 3.40, SD = 1.425) and to have a normal or corrected vision.
To compare the improvement effect of different cognitive interventions on aggressive driving indicators, all subjects were divided into 6 groups randomly, each of which including 5 male and 3 female subjects, as shown in Table 1. Note that there were no significant differences in age (F(5, 42) = 2.058, ), gender (F(5, 42) = 0.000, ) and driving experience (F(5, 42) = 1.396, ) among groups. Therefore, it is concluded that there is no statistical difference is present between the abovementioned factors among all groups. Furthermore, none of the subjects showed simulator symptoms during the experiment, and all subjects who completed the experiment as required were rewarded.
2.2. Apparatus
Due to the advantages of safety, high efficiency, repeatability, and easy data processing, the simulator was widely adopted, to evaluate driving behavior in previous studies [23, 24]. Therefore, a driving simulator that provides a more realistic virtual driving platform was used in this study. The visual display system includes 5 hosts and 3 screens with 3 1280 × 800 pixels of resolution and generates dynamic traffic scenes including roads, pedestrians, vehicles, and surrounding scenes. The sound equipment system is capable of simulating various sounds in the traffic environment.
2.3. Experimental Design
2.3.1. Simulated Driving Task
The simulated driving task was carried out on a two-lane urban road with a total length of about 6.0 km, and the speed limit was 60 km/h in most sections except for multiple speed limit zones. To increase the authenticity of driving, buildings, trees, street lights, parking vehicles, and pedestrians were set up on both sides of the road. There were 9 intersections in the whole journey, and the vehicle speed was measured at 4 of them. On the straight road, there were two-speed limit zones with a length of about 1 km. To explore whether the low and high-speed limits had an impact on the effectiveness of the intervention method, the speed limit zones of 40 km/h and 80 km/h were, respectively, set, as well as the speed limit signs. Furthermore, 4 crosswalks were set. When the subjects were about to drive on these crosswalks, pedestrians would pass through the street, as seen in Figure 2.
2.3.2. Cognitive Intervention Methods
The cognitive intervention methods are designed in Table 2.
2.3.3. Experimental Process
It took about 1 hour to finish the whole experiment. Subjects were not informed on the purpose of this study, to reduce the impact of deliberately driving behaviors. First, participants were instructed to drive as they normally would and to understand the intervention’s content. Second, participants were provided a questionnaire about personal information and were required to complete a simulated driving task after a 5-minute practice on the simulator. Third, the intervention programs were implemented for different groups. Finally, the simulated driving task that differed only in the order of events was asked to complete.
2.4. Data Analysis
2.4.1. Definition of Dependent Variables
The dependent variables in this study are defined in Table 3.
2.4.2. Statistical Analysis Methods
In this study, a mixed experimental design method combining between-subjects and within-subjects was adopted. The intervention method was a within-subject independent variable, and the driving task was a between subjects independent variable. First, descriptive statistics were performed for all dependent variables. Second, the driving performance data of the control group and experimental groups were tested using the independent samples t tests to compare the differences between before and after the intervention. Finally, a mixed repeated measures ANOVA analysis was performed on all dependent variables to check whether the 2 factors had significant main effects or interactions on different measures of aggressive behavior. Intervention methods included 6 levels: no intervention, performance feedback intervention, performance penalty feedback intervention, performance safety feedback intervention, combined feedback intervention, and self-explanation intervention. The driving task was used as a repeatable measurement variable that included preintervention and postintervention tasks. All data in this study were analyzed using SPSS 22.0 software (IBM).
3. Experimental Results
3.1. Descriptive Statistical Analysis
The mean and standard deviation of all dependent variables in the preintervention and postintervention tasks were calculated, and the differences between the control group and experimental groups were compared as shown in Figures 3–4–5–6–7–8–9.
Figure 3 shows the change of the mean speed in the 40 km/h zone before and after intervention. The decrease in mean speed is greater in the experimental group than in the control group, among which the performance penalty and feedback group decreased the most, reaching 29.84%. Other experimental groups, including the performance and feedback group, performance and safety feedback group, combined feedback group, and self-explanation group, decreased by 8.81%, 14.80%, 22.05%, and 24.94%, respectively.
Figure 4 shows the change in the distance ratio of overspeed in the 40 km/h zone before and after the intervention. Both the control group and experimental groups show a decrease, noting that the decrease in each experimental group is superior to that in the control group, among which the combined feedback group shows the largest decrease reaching 60.00%. Other experimental groups, including the performance feedback group, performance and penalty feedback group, performance and safety feedback group, combined feedback group, and self-explanation group, decreased by 32.14%, 38.36%, 52.73%, and 46.88%, respectively.
Figure 5 shows the change in the mean speed in the 80 km/h zone before and after the intervention. The differences between the control group, performance feedback group, and performance and safety feedback group are relatively minor, while the differences between other experimental groups, such as the performance and penalty feedback group, combined feedback group, and self-explanation group are 4.89%, 4.23%, and 4.31%, respectively.
Figure 6 shows the change in the distance ratio of overspeed in the 80 km/h zone before and after the intervention. The decrease in each experimental group was superior to that in the control group, among which the combined feedback group had the largest decrease reaching 100.00%. Other experimental groups, including the performance feedback group, performance and penalty feedback group, performance and safety feedback group, combined feedback group, and self-explanation group, diminished by 50.00%, 63.16%, 20.00%, and 92.00%, respectively.
Figure 7 shows that the change in the mean speed when passing pedestrians before and after the intervention, the mean speed in the control group, performance feedback group, and performance and safety feedback group showed a minor increase, while the mean speed of other experimental groups, such as the performance and penalty feedback group, combined feedback group, and self-explanation group, diminished after the intervention, with a drop of 15.21%, 6.85%, and 7.65%, respectively.
Figure 8 shows the change in the mean speed at the intersection before and after the intervention. The mean speed of the control group and the performance feedback group exhibited a minor augmentation, while the mean speed of other experimental groups, such as the performance and penalty feedback group, performance and safety feedback group, combined feedback group, and self-explanation group, decreased after the intervention. The decrease rates were 7.59%, 1.33%, 14.25%, and 9.58%, respectively.
Figure 9 exhibits the change in mean speed on the entire drive before and after the intervention. The mean speed of the control group and the performance feedback group showed a minor increase, while the mean speed of other experimental groups, such as the performance and penalty feedback group, performance and safety feedback group, combined feedback group, and self-explanation group, decreased after the intervention. The decrease rates were 10.55%, 2.28%, 11.61%, and 7.76%, respectively.
3.2. Independent Sample T Test
The independent sample T-test method was adopted to further analyze the differences between independent variables before and after the intervention. Hence, 3 statistical values, including the T value, significance level ( value), and effect size (Cohen’s d), were calculated. Cohen’s d represents the degree of statistical difference in the data. In general, a value of 0.200–0.500 indicates a minor effect, 0.500–0.800 is a medium effect, and superior to 0.800 is a large one.
As shown in Table 4, in the performance feedback group and performance and safety feedback group, the significant level was superior to 0.05 and the effect size inferior to 0.800, indicating that all dependent variables had no significant differences between before and after the intervention. This was not consistent with the other experimental groups. In the performance and penalty feedback group, there was a major effect on the mean speed in the 40 km/h zone, with the significant level of 0.023 and the effect size of 1.368, indicating that the performance and penalty feedback method contributes to improving the aggressive driving behavior in the low-speed limit zone. In the combined feedback group, the intervention method had remarkable effects on both the mean speed at the intersection and the whole drive. The important level and effect size of the former were 0.030 and 1.274, while those of the latter were 0.032 and 1.187, indicating major differences in both mean speed at the intersection and on the whole drive between before and after the intervention. Furthermore, the distance ratio of overspeed in the 40 km/h and 80 km/h zones had major changes before and after the intervention, although the significant levels were 0.062 and 0.072 (superior to 0.05), the effect sizes were 1.006 and 1.053 (superior to 0.800), indicating large effects on both dependent variables.
3.3. Analysis of Variance
The main effects and interaction effects of the two factors, the driving task and intervention type, on the dependent variables are shown in Table 5. The driving task had main effects on the mean speed in the 40 km/h zone, a distance ratio of overspeed equal to 40 km/h, a distance ratio of overspeed in the 80 km/h zone, and the mean speed in the whole drive. Intervention methods had merely a significant effect on the mean speed on the entire drive. However, the interaction effects were not remarkable on all dependent variables.
4. Discussion and Conclusions
The present study aims to experimentally examine the effect of cognitive intervention on nonaggressive driving behavior among Chinese drivers. Five speed-related indicators were assessed: mean speed in the 40 km/h zone and the 80 km/h zone, distance ratio of overspeed in the aforementioned zones, mean speed when passing pedestrians, mean speed at the intersection, and mean speed on the whole drive. The results demonstrate that (1) cognitive intervention methods contribute distinctively to improving aggressive driving behaviors. Moreover, the penalty feedback intervention significantly reduced the mean speed in the 40 km/h zone, the 80 km/h zone, and when passing pedestrians, while the combined feedback intervention remarkably reduced the distance ratio of speeding in the 40 km/h and 80 km/h zones, as well as the mean speed at the intersection and on the whole drive. (2) The interaction effects between the driving task and the intervention method are not clearly observed on aggressive driving behaviors.
Previous studies have shown that cognitive interventions are beneficial in reducing drivers’ speeding behavior in the speed limit zone [19]. The study extends this conclusion, studying the cognitive intervention effect on the improvement of aggressive driving behavior by comparing the changes of more indicators. For example, higher speed when the vehicle was approaching vulnerable road users was a principal cause of traffic accidents. In this setting, multiple drivers tended to display a more aggressive driving behavior in front of vulnerable road users [25]. This study finds that cognitive interventions such as penalty feedback improve this aggressive driving behavior, suggesting that, for drivers who do not slow down or yield to pedestrians, the advanced driving assistance system provides real-time voice warning by informing them that they would be fined due to aggressive driving behavior.
There is increasing evidence that when the feedback content includes penalty information, a preferable improvement effect is acquired. For example, the performance and penalty feedback intervention reduces the mean speed in the 40 km/h area and when passing pedestrians remarkably. This is consistent with previous studies indicating that feedback interventions that include penalty information are superiorly effective to others in the reduction of overspeed behavior [20]. In the experiment, speed measurements were conducted not only in the speed limit zone but also in the other ones. The results revealed that feedback interventions that include penalty information furtherly improve the speed management behavior throughout the whole journey. The aforementioned demonstrates that this intervention method is universally applicable for improving aggressive driving behavior for Chinese drivers.
Moreover, this study found that the self-explanation group has a better improvement effect on the distance ratio of overspeed in high and low-speed limit zones, but not as good as the improvement effect of the combined feedback intervention. Nevertheless, the self-explanation intervention method better understands the reasons why drivers have aggressive driving behaviors. Furthermore, it timely determines the psychological barriers that are not conducive to safe driving and conducts counseling [26]. Therefore, in the process of implementing the combined feedback intervention, a self-explanation intervention can be used as a supplement to achieve a better intervention effect.
5. Limitations and Future Work
In this study, it is impossible to simulate physical harm at a high speed when passing pedestrians or crossing an intersection in a driving simulator. Hence, future studies could use different feedback structures in an attempt to validate and expand various findings of this study, such as physical harm information included in the feedback content. Otherwise, a comparison between the effects of intervention methods on Chinese drivers and western drivers is necessary for future studies. Thus, findings from this study might have implications on the designation of voice warning content in the advanced driving assistance system (ADAS). In China, truck drivers often pass pedestrians or intersections at high speeds to meet a certain delivery time or financial incentive goal. Subsequently, future studies should explore how different cognitive interventions impact aggressive driving behavior under certain incentives.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The authors would like to thank all participants for their contribution in this study. This work was supported by the Key Laboratory of Transportation Industry of Automotive Transportation Safety Enhancement Technology, Chang’an University (grant no 300102222506), Natural Science Basic Research Project of Shaanxi Province (2023-JC-QN-0516), National Key R&D Program of China (grant no. 2019YFB1600500), and Fundamental Research Funds of the Central Universities, CHD (grant nos. 300102222110 and 300102410104).