Investigating explanations in conditional and highly automated driving: The effects of situation awareness and modality

With the level of automation increases in vehicles, such as conditional and highly automated vehicles (AVs), drivers are becoming increasingly out of the control loop, especially in unexpected driving scenarios. Although it might be not necessary to require the drivers to intervene on most occasions, it is still important to improve drivers’ situation awareness (SA) in unexpected driving scenarios to improve their trust in and acceptance of AVs. In this study, we conceptualized SA at the levels of perception (SA L1), comprehension (SA L2), and projection (SA L3), and proposed an SA level-based explanation framework based on explainable AI. Then, we examined the effects of these explanations and their modalities on drivers’ situational trust, cognitive workload, as well as explanation satisfaction. A three (SA levels: SA L1, SA L2 and SA L3) by two (explanation modalities: visual, visual + audio) between-subjects experiment was conducted with 340 participants recruited from Amazon Mechanical Turk. The results indicated that by designing the explanations using the proposed SA-based framework, participants could redirect their attention to the important objects in the traffic and understand their meaning for the AV system. This improved their SA and filled the gap of understanding the correspondence of AV’s behavior in the particular situations which also increased their situational trust in AV. The results showed that participants reported the highest trust with SA L2 explanations, although the mental workload was assessed higher in this level. The results also provided insights into the relationship between the amount of information in explanations and modalities, showing that participants were more satisfied with visual-only explanations in the SA L1 and SA L2 conditions and were more satisfied with visual and auditory explanations in the SA L3 condition. Finally, we found that the cognitive workload was also higher in SA L2, possibly because the participants were actively interpreting the results, consistent with a higher level of situational trust. These findings demonstrated that properly designed explanations, based on our proposed SA-based framework, had significant implications for explaining AV behavior in conditional and highly automated driving.     

Modelling perceived risk and trust in driving automation reacting to merging and braking vehicles

Perceived risk and trust are crucial for user acceptance of driving automation. In this study, we identify important predictors of perceived risk and trust in a driving simulator experiment and develop models through stepwise regression to predict event-based changes in perceived risk and trust. 25 participants were tasked to monitor SAE Level 2 driving automation (ACC + LC) while experiencing merging and hard braking events with varying criticality on a motorway. Perceived risk and trust were rated verbally after each event, and continuous perceived risk, pupil diameter and ECG signals were explored as possible indictors for perceived risk and trust.The regression models show that relative motion with neighbouring road users accounts for most perceived risk and trust variations, and no difference was found between hard braking with merging and hard braking without merging. Drivers trust the automation more in the second exposure to events. Our models show modest effects of personal characteristics: experienced drivers are less sensitive to risk and trust the automation more, while female participants perceive more risk than males. Perceived risk and trust highly correlate and have similar determinants. Continuous perceived risk accurately reflects participants’ verbal post-event rating of perceived risk; the use of brakes is an effective indicator of high perceived risk and low trust, and pupil diameter correlates to perceived risk in the most critical events. The events increased heart rate, but we found no correlation with event criticality. The prediction models and the findings on physiological measures shed light on the event-based dynamics of perceived risk and trust and can guide human-centred automation design to reduce perceived risk and enhance trust.     

The role of causal attribution in hurt feelings and related social emotions elicited in reaction to other's feedback about failure

The present research addressed the question of what it is that makes certain types of feedback on the reasons for failure hurtful. The results of two studies demonstrated that the causal structure implied by an explanation for failure explains the degree to which the explanation is perceived as hurtful and likely to elicit anger, shame, and guilt. In contrast, the perceived validity of the explanations is of relatively less importance for the elicitation of hurt feelings and anger than the content of the explanation. Overall, these results provide further evidence for the importance of attributional information for social emotions, whereas the validity of the information had a relatively lesser effect.     

Type a behavior pattern and attributions of responsibility

This study set out to examine the relationship between the Type A/B behavior pattern and reaction to negative outcomes. It was predicted that whereas the B types would show clear differences in their reactions to controllable versus uncontrollable situations, A types would show significantly less differentiation. Over 160 subjects completed two A-type questionnaires and an attributional control questionnaire with either six controllable or six uncontrollable situations. Results revealed, as predicted, that A and B types differed mainly in their reactions to uncontrollable situations, whereas A types perceived more causal responsibility and more moral responsibility, and reported more anger with self. These results are discussed within the context of the research on attributional style, depression and Type A behavior.     

The Relation Between Perceived Risk and Preventive Action: A Within-Subject Analysis of Perceived Driving Risk and Intentions to Wear Seatbelts

Intentions to wear seatbelts in 12 different driving situations were predicted from attitudes toward wearing seatbelts, subjective norms concerning seatbelt uses, and perceived driving risk. In a given driving situation, appropriate measures of attitudes and subjective norms both had significant effects on intentions to wear a seatbelt, whereas there was little relation between risk and intentions. Intentions across the 12 driving situations were significantly related to perceived driving risk, both for aggregate data and for a substantial portion of individual subjects. However, further analyses indicated that risk seemed to affect intentions indirectly through subjective norms and attitudes associated with seat-belt use. The results suggest that attempts to increase seatbelt use should target all relevant beliefs important in determining people's attitudes toward and subjective norms concerning seatbelt use, rather than just focusing upon making people aware of the risks associated with driving.     

Predicting pedestrians’ intention to cross the road in front of automated vehicles in risky situations

Due to the absence of a human driver, the introduction of fully automated vehicles (FAVs) may bring new safety challenges to the traffic system, especially when FAVs interact with vulnerable road users such as pedestrians. To ensure safer interactions between pedestrians and FAVs, this questionnaire-based study aims to understand Australian pedestrians’ intention to engage in risky road-crossing behaviors when they interact with FAVs vs. human-driven vehicles (HDVs). A 2 × 2 between-subject design was utilized, in which two risky road-crossing scenarios were designed and took into account the vehicle type (FAV vs. HDV) and vehicle speed (30 km/h vs. 50 km/h). A total of 493 participants (aged 18–77) were randomly assigned to one of the four experimental conditions and completed an online questionnaire based on the extended Theory of Planned Behavior (TPB). This questionnaire measured pedestrians’ intentions to cross the road in the assigned scenarios as well as the motivational factors behind these intentions in terms of attitude, subjective norm, perceived behavioral control, perceived risk and trust in the vehicle. The results show that pedestrians had significantly higher intentions to cross the road in front of approaching FAVs than HDVs. Participants also reported a lower risk perception of crossing in front of FAVs and greater trust in this type of vehicle. Attitude, subjective norm, and perceived behavioral control were significant predictors of intentions to engage in risky road-crossing behavior. Findings of this study provide important implications for the development and implementation of FAVs in the future road transport system.     

The Social Goals of Excuses: Self‐Serving Attributions or Politeness Strategies1

Participants provided examples of explanations given as excuses, or withheld in favor of a false excuse, both for failing to keep a social contract and for rejecting a social invitation. Results show that the likelihood of an explanation being given in social‐contract situations was best predicted by intentionality and controllability; while in social‐rejection situations, the most important factor was to avoid personal reasons (related to the recipient). Excuse‐givers were even willing to blame themselves to do this. These results are discussed in terms of the need to extend attributional categories beyond traditional self‐serving functions to include social goals. such as the recipient saving face, to provide an adequate account of how excuse‐making varies across different types of social predicaments.     

Icy roads and urban environments. Passenger experiences in autonomous vehicles in Finland

This study focused on identifying customerś real-life experiences, perceptions and feelings about travelling in different autonomous vehicles and in various operating conditions in Finland in 2018. Quantitative convenience sample (n = 141) were collected from passengers travelling on an autonomous shuttle bus in Helsinki. Qualitative data (n = 70) were gathered by interviewing passengers of a driverless shuttle bus in Helsinki and passengers of an autonomous car in winter conditions in Lapland. This research was first one which included passengers’ real-life experiences after using autonomous vehicles in winter conditions. We applied the Theory of Planned Behaviour. The research questions were (a) What beliefs about outcomes and evaluation of outcomes do passengers have and carry out when they travel in an autonomous vehicle, irrespective of vehicle type or operating conditions? (b) What key factors influence people’s positive or negative attitudes towards autonomous vehicles? (c) What key factors could induce people to use autonomous vehicles? The quantitative data were analysed by nonparametric Kruskal-Wallis H test and qualitative data by inductive content analysis. According to the results, trust, safety and security were the main factors influencing people’s positive attitudes towards using autonomous vehicles. Results from passengers travelling in heavy winter conditions indicate that winter conditions do not significantly influence passengers’ attitudes towards using autonomous vehicles. There were no significant differences between gender regarding passengers’ perceptions of traffic safety, personal security and emergency management. However, younger passengers felt their personal security on board to be significantly better than older and students their possibilities to act in a case of emergency significantly better than employed people.     

Developing human-machine trust: Impacts of prior instruction and automation failure on driver trust in partially automated vehicles

To prompt the use of driving automation in an appropriate and safe manner, system designers require knowledge about the dynamics of driver trust. To enhance this knowledge, this study manipulated prior information of a partial driving automation into two types (detailed and less) and investigated the effects of the information on the development of trust with respect to three trust attributions proposed by Muir (1994): predictability, dependability, and faith. Furthermore, a driving simulator generated two types of automation failures (limitation and malfunction), and at six instances during the study, 56 drivers completed questionnaires about their levels of trust in the automation. Statistical analysis found that trust ratings of automation steadily increased with the experience of simulation regardless of the drivers’ levels of knowledge. Automation failure led to a temporary decrease in trust ratings; however, the trust was rebuilt by a subsequent experience of flawless automation. Results showed that dependability was the most dominant belief of drivers’ trust throughout the whole experiment, regardless of their knowledge level. Interestingly, detailed analysis indicated that trust can be accounted by different attributions depending on the drivers’ circumstances: the subsequent experience of error-free automation after the exposure to automation failure led predictability to be a secondary predictive attribution of drivers’ trust in the detailed group whilst faith was consistently the secondary contributor to shaping trust in the less group throughout the experiment. These findings have implications for system design regarding transparency and for training methods and instruction aimed at improving driving safety in traffic environments with automated vehicles.     

On-road trust and perceived risk in Level 2 automation

To encourage appropriate use of driving automation, we need to understand and monitor driver’s trust and risk perception. We examined (1) how trust and perceived risk are affected by automation, driving conditions and experience and (2) how well perceived risk can be inferred from behaviour and physiology at three levels: over traffic conditions, aggregated risk events, and individual risk events.30 users with and without automation experience drove a Toyota Corolla with driving support. Safety attitude, subjective ratings, behaviour and physiology were examined.Driving support encouraged a positive safety attitude and active driver involvement. It reduced latent hazards while maintaining saliently perceived risks. Drivers frequently overruled lane centring (3.1 times/minute) and kept their feet on or above the pedals using ACC (65.8% of time). They comfortably used support on curvy motorways and monotonic and congested highways but less in unstable traffic and on roundabouts. They trusted the automation 65.4%, perceived 36.0% risk, acknowledged the need to monitor and would not engage in more secondary tasks than during manual driving.Trust-in situation reduced 2.0% when using automation. It was 8.2% higher than trust-in-automation, presumably due to driver self-confidence. Driving conditions or conflicts between driver and automation did not affect trust-in-automation.At the traffic condition level, physiology showed weak and partially counter-intuitive effects. For aggregated risk events, skin conductance had the clearest response but was discernible from baseline in  < 50%. Pupil dilation and heart rate only increased with strong braking and active lane departure assist. For individual risk events, a CNN classifier could not identify risk events from physiology. We conclude that GSR, heart rate and pupil dilation respond to perceived risk, but lack specificity to monitor it on individual events.     

自动驾驶车中的人机信任

自动驾驶是当前智能汽车发展的重要方向。在实现完全自动化驾驶前, 驾驶员和自动驾驶系统共享车辆控制权, 协同完成驾驶任务。在该人-机共驾阶段, 人对自动驾驶系统的信任是影响自动驾驶中人机协同效率与驾驶安全的关键要素; 驾驶员对自动驾驶车辆保持适当的信任水平对驾驶安全至关重要。本研究结合信任的发展阶段与影响因素提出了动态信任框架。该框架将信任发展分为倾向性信任、初始信任、实时信任和事后信任四个发展阶段, 并结合操作者特征(人)、系统特征(自动驾驶车系统)、情境特征(环境)三个关键因素分析不同阶段的核心影响因素以及彼此间的内在关联。根据该框架, 信任校准可从监测矫正、驾驶员训练、优化HMI设计三类途径展开。未来研究应更多关注驾驶员和人机系统设计特征对信任的影响, 考察信任的实时测量和功能特异性, 探讨驾驶员和系统的相互信任机制, 以及提升信任研究的外部效度。     

Religious Attributions and Proximity of Influence: An Investigation of Direct Interventions and Distal Explanations

Investigations into religious attributions have focused on attributer's immediate, proximal causes of events, paying little attention to underlying, distal explanations. In an effort to explain the relatively low incidence of religious attributions and further a new model of proximal-distal attributions, we present two experiments investigating the proximal and distal use of religious and nonreligious supernatural attributions. Participants in both studies were presented a series of sixteen vignettes that varied on several attribution-relevant dimensions. After reading each vignette, subjects gave an initial explanation for the event, and were then probed for any underlying explanations. Experiment 1 used an interview format that allowed participants maximum latitude when explaining the event's outcome. Consistent with our predictions, participants perceived God as having a greater distal than proximal influence, though this difference was not evident for attributions to Satan or nonreligious supernatural agents. Experiment 2 was performed via a microcomputer, with a participant's initial attributional response branching to a set of appropriate distal explanations. Overall, the results suggest that attributer's perceive God working through indirect influences rather than direct intervention, with this effect being moderated by the attributer's level of religiosity. This pattern was evident in the use of God as a distal explanation as well as in the distal explanations to proximal attributions to God.     

Does betrayal aversion really guide trust decisions towards strangers?

Trust is a double-edged sword. When warranted, it leads to positive and rewarding interactions. When not, it leads to disappointment and anger. Therefore, it has been argued that people will display “betrayal aversion” in trust situations (i.e., avoid trusting to avoid betrayal). Yet, people also feel tense and uneasy when they signal distrust to another person and thus show signs of “principled trustfulness” (i.e., choosing to trust others although being skeptical of their trustworthiness). These two theoretical orientations imply directly opposite influences on trust behavior. Thus, we conducted two laboratory studies (with a total of 841 participants) with binary trust games (implying a risk of being betrayed) and extended lottery games (implying no such risk). In both studies, we varied the payoff structures of both games. Further, we made sure that the average perceived likelihood of winning or losing money when choosing the risky option was identical in both games, as was the distribution of these likelihoods. Neither study showed any sign of betrayal aversion. Rather, participants were more willing to risk their money in the trust game than they were to invest their money in a lottery, supporting the principled trustfulness view. We discuss possible explanations why, unlike previous studies, we did not find any indication of betrayal aversion.     

Sharing the road with autonomous vehicles: A qualitative analysis of the perceptions of pedestrians and bicyclists

Public perception assessment is important for gaining a better understanding of the acceptance of autonomous vehicles (AVs) and identifying potential ways to resolve public concerns. This study investigated how pedestrians and bicyclists perceived AVs based on their knowledge and road sharing experiences, applying a combined inductive and deductive data analysis approach. Survey responses of pedestrians and bicyclists in Pittsburgh, Pennsylvania, USA collected by Bike Pittsburgh (BikePGH) in 2019, were analyzed in this research. AVs following traffic rules appropriately and AVs driving safer than the human drivers were the most notable positive perceptions towards AVs. Pedestrians and bicyclists showed comparatively fewer negative perceptions towards AVs than positive perceptions. Negative perceptions mostly included a lack of perceived safety and comfort around AVs and trust in the AV technology. Respondents also concerned about AV technology issues (e.g., slow and defensive driving, disruptive maneuver), while sharing the road with AVs. Perceptions of the respondents were significantly influenced by their views on AV safety, familiarity with the technology, the extent respondents followed AV on the news, and household automobile ownership. Regulating AV movement on roadways, developing safety assessment guidelines, and controlling oversights of improper practices by AV companies were the major suggestions from the survey participants. Findings of this study might help AV companies to identify potential improvement needed in AV technology to increase pedestrians and bicyclists acceptance, and policymakers to develop policy guidelines to ensure safe road sharing among pedestrians, bicyclists, and AVs.     

A matter of style? Driver attributional ‘style’ in accounting for the driving of others as protective or as predisposing drivers towards retaliatory aggressive driving

Driver cognitions about aggressive driving of others are potentially important to the development of evidence-based interventions. Previous research has suggested that perceptions that other drivers are intentionally aggressive may influence recipient driver anger and subsequent aggressive responses. Accordingly, recent research on aggressive driving has attempted to distinguish between intentional and unintentional motives in relation to problem driving behaviours. This study assessed driver cognitive responses to common potentially provocative hypothetical driving scenarios to explore the role of attributions in driver aggression. A convenience sample of 315 general drivers 16–64 yrs (M = 34) completed a survey measuring trait aggression (Aggression Questionnaire AQ), driving anger (Driving Anger Scale, DAS), and a proxy measure of aggressive driving behaviour (Australian Propensity for Angry Driving AusPADS). Purpose designed items asked for drivers’ ‘most likely’ thought in response to AusPADS scenarios. Response options were equivalent to causal attributions about the other driver. Patterns in endorsements of attribution responses to the scenarios suggested that drivers tended to adopt a particular perception of the driving of others regardless of the depicted circumstances: a driving attributional style. No gender or age differences were found for attributional style. Significant differences were detected between attributional styles for driving anger and endorsement of aggressive responses to driving situations. Drivers who attributed the on-road event to the other being an incompetent or dangerous driver had significantly higher driving anger scores and endorsed significantly more aggressive driving responses than those drivers who attributed other driver’s behaviour to mistakes. In contrast, drivers who gave others the ‘benefit of the doubt’ endorsed significantly less aggressive driving responses than either of these other two groups, suggesting that this style is protective.     

Leader Attributions for Subordinate Performance: Consequences for Subsequent Leader Interactive Behaviors and Ratings1

An important application of attribution theory deals with leader explanations for subordinate performance and their effects on future leader–member interactions and performance. In the present study, subjects worked on a 2-trial task in which there was a leader and 2 members. Leaders received performance feedback and an attributional explanation for subordinate performance after Trial 1, and subsequent Trial 2 behavior was videotaped. Results showed that attributions significantly affected the amount of time spent by the leader talking to the group during the second trial, as well as the number of negative leader comments. Level of performance was a significant determinant of subordinate ratings and reward/punishment recommendations. Attributions differentially affected the punishment advocated by leaders, with failure due to internal causes more likely to be punished than failure due to external causes. Implications for an attributional theory of leadership are discussed.     

Behavioral adaptations of human drivers interacting with automated vehicles

Advancements in technology are bringing automated vehicles (AVs) closer to wider deployment. However, in the early phases of their deployment, AVs will coexist and frequently interact with human-driven vehicles (HDVs). These interactions might lead to changes in the driving behavior of HDVs. A field test was conducted in the Netherlands with 18 participants focusing on gap acceptance, car-following, and overtaking behaviors to understand such behavioral adaptations. The participants were asked to drive their vehicles in a controlled environment, interacting with an HDV and a Wizard of Oz AV. The effects of positive and negative information regarding AV behavior on the participants’ driving behavior and their trust in AVs were also studied. The results show that human drivers adopted significantly smaller critical gaps when interacting with the approaching AV as compared to when interacting with the approaching HDV. Drivers also maintained a significantly shorter headway after overtaking the AV in comparison to overtaking the HDV. Positive information about the behavior of the AV led to closer interactions in comparison to HDVs. Additionally, drivers showed higher trust in the interacting AV when they were provided with positive information regarding the AV in comparison to scenarios where no information was provided. These findings suggest the potential exploitation of AV technology by HDV drivers.     

Comments on ‘Hempel's ambiguity’ by J. Alberto Coffa

Using Coffa's paper as a point of departure, this brief note is designed to show that Hempel's inductive-statistical model of explanation implicitly construes explanations of that type as defective deductive-nomological explanations, with the consequence that there is no such thing as genuine inductive-statistical explanation according to Hempel's account. This result suggests a possible implicit commitment to determinism behind Hempel's theory of scientific explanation.