ResearchPod Summary
This paper investigates why potential users resist adopting autonomous vehicles (AVs), moving beyond traditional technology acceptance models (like TAM or UTAUT) that focus primarily on usability. The authors apply Cognitive Appraisal Theory (CAT) to map how specific cognitive evaluations of AVs generate five distinct negative emotions—fear, anger, sadness, frustration, and dislike—and how these emotions subsequently influence behavioral intentions to reject the technology.
The researchers developed the Emotional Barrier in Autonomous Vehicle Adoption Model (EB-AVAM) and tested it using survey data from 605 participants in China. The framework links external antecedents (performance expectancy, effort expectancy, social influence, facilitating conditions, habit, hedonic motivation, and price value) to internal cognitive appraisals (outcome desirability, certainty, fairness, and perceived control). Structural equation modeling was then used to determine which of these appraisals trigger negative emotions and, ultimately, which emotions drive the rejection of AV services.
The study finds that negative emotions are not a monolithic barrier; rather, they are multidimensional and have different impacts on behavior. Fear, anger, and dislike are high-arousal emotions that significantly increase the intention to reject AVs, whereas sadness and frustration do not have a statistically significant impact on rejection. The study highlights that unfairness is a major driver of negative affect; when users feel the technology violates social norms or creates unfair outcomes, it triggers stronger resistance. Additionally, the study demonstrates that factors like effort expectancy and perceived control are critical: when users feel that operating or understanding an AV requires too much effort or leaves them without control, their emotional distress increases.
Alex: Feeling bad about driverless cars does not always mean refusing to use them. In a Chinese survey, fear, anger, and dislike were linked to rejection, but sadness and frustration were not significantly linked.
Sam: That distinction could change what a transport provider tries to fix. Whose research is this, and what evidence supports it?
Alex: It’s a survey study by Rongbin Yang and Santoso Wibowo, published in Transportation Research Part F in 2025. They analysed responses from 605 people in China. Their central claim is that negative emotion needs separating into different feelings, with different relationships to rejection.
Sam: So this matters beyond asking whether people like the technology. It could help researchers distinguish resistance from dissatisfaction.
Alex: And help providers choose what concerns to address, rather than offer generic reassurance. We’ll unpack the emotional distinctions, how the authors tested them, and what a survey of intentions can actually establish.
Sam: Start with the gap. Technology adoption research already asks about trust and risk. What was missing here?
Alex: The authors describe traditional acceptance models as emphasizing performance and ease of use, while underplaying emotion. More recent models do include emotional responses. But many studies combine emotions into one overall measure, or examine only fear or anger.
Sam: Combining them assumes that different unpleasant feelings tell us roughly the same thing. What does this study put in their place?
Alex: A framework separating fear, anger, dislike, sadness, and frustration. It also asks which evaluations are associated with each emotion. The authors call it the Emotional Barrier in Autonomous Vehicle Adoption Model. The contribution is the linked structure, not simply a longer emotion checklist.
Sam: Before we follow those links, which vehicles are we talking about? Driver assistance and having no driver are quite different experiences.
Alex: Here, driverless means systems that perform the entire driving task without human input. The questionnaire gave respondents that definition. So the study concerns fully driverless services, not just features that help someone who is still driving.
For policymakers and practitioners, the findings emphasize that overcoming resistance to AVs requires more than just improving technical performance. Because different negative emotions are driven by different underlying appraisals, providers must employ targeted strategies: addressing fear through better safety education, mitigating anger through transparent and fair service policies, and reducing dislike through more engaging and user-friendly vehicle designs.
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.
Sam: What makes the framework work? Give us the idea underneath all those relationships.
Alex: It uses cognitive appraisal theory: how people’s evaluations of a situation shape emotions and behavior. The same technology can mean different things to different users. Those evaluations help explain why one person feels threatened while another feels disappointed.
Sam: Can you make that concrete using an example the authors give?
Alex: They describe a traveler encountering an unreliable driverless vehicle. The traveler may see it as relevant to personal safety and at odds with efficient travel. That combination can produce fear. The framework asks about the perceived outcome, not just the presence of unfamiliar technology.
Sam: So the assessment includes whether it helps me, but also whether I can manage what happens.
Alex: The authors separate desirability, certainty, and fairness: whether outcomes serve your goals, seem predictable, and feel deserved or equitable. They also examine perceived control, meaning your sense of being able to manage or cope with the situation.
Sam: Are those assessments supposed to happen in sequence? First I judge the outcome, then decide whether I can cope?
Alex: Not necessarily. That’s another distinction the authors make from earlier models. Evaluating an outcome and evaluating your coping resources can happen together. Their framework does not require a fixed sequence between those assessments.
Sam: How did they turn that framework into evidence, rather than just a plausible account?
Alex: They recruited through a Chinese online survey platform. From 810 collected responses, they retained the valid sample after excluding incomplete, disqualified, disengaged, and outlying responses. Participants rated statements about technology expectations, appraisals, emotions, and rejection intentions.
Sam: Were these passengers describing actual driverless journeys, or people considering a service they might use?
Alex: Eligibility required recent travel by motor vehicle, not experience with driverless vehicles. The authors did not categorize respondents by prior driverless experience or familiarity. That matters: these are reported perceptions and intentions, not observations of passengers choosing or refusing rides.
Sam: With so many connected concepts, how did they check that the survey was measuring distinct things?
Alex: They tested the measures for consistency and whether the constructs could be distinguished. One poorly performing control item was removed. They then used partial least squares structural equation modeling, a method for estimating linked relationships among measured concepts. They also assessed potential shared-method bias and model fit.
Sam: A model can fit survey responses without proving the proposed causal story. How should we read those links?
Alex: As associations consistent with the framework, not evidence that changing an appraisal will necessarily change behavior. The authors describe appraisals leading to emotions and intentions. But this survey does not establish that causal sequence through an intervention.
Sam: Then let’s separate the results. Which emotions were most closely linked to rejection?
Alex: Fear had the largest estimated association with rejection intention, with a standardized coefficient of about point three three. Anger’s coefficient was about point one eight. Dislike also had a significant positive association, while sadness and frustration did not.
Sam: Does that mean frustration is harmless, or just that it wasn’t a clear predictor of this particular outcome?
Alex: The second. Rejection intention was the outcome here, not every possible service response. The authors argue that frustration may prompt attempts to resolve problems rather than outright refusal. They also caution that sadness and frustration still matter for satisfaction and customer well-being.
Sam: That makes the distinction useful: unpleasant feelings can accompany continued engagement. What evaluations were linked to fewer negative feelings?
Alex: Greater desirability, fairness, and perceived control were associated with less of every emotion studied. Certainty had a narrower pattern. It was associated with less frustration, dislike, and fear, but not significantly with sadness or anger.
Sam: So making a service predictable might address fear without addressing anger. What explains that difference in the authors’ account?
Alex: They interpret anger as closely connected to perceived mistreatment or injustice. An outcome can be predictable and still feel unfair. Fairness had especially prominent associations with anger and dislike. Reliability alone is therefore not the whole emotional story in this model.
Sam: What sits upstream of those evaluations? Are the authors only measuring beliefs about safety?
Alex: No. They include expected benefits, required effort, social influence, available support, habits with artificial intelligence, enjoyment, and value for money. Enjoyment and price value were associated with desirability and fairness. Available support had a prominent association with perceived control.
Sam: Familiarity sounds useful there. Did habits with other artificial intelligence tools also make driverless outcomes seem more predictable?
Alex: Not significantly. Habit was associated with desirability and control, but not certainty. The authors suggest familiarity with other automated tools may not help people anticipate this new application. That’s an interpretation of the pattern, rather than a separately tested explanation.
Sam: How far should we carry these findings beyond the sample?
Alex: Cautiously. It was an online survey in one country, with potential self-selection and social-desirability biases. Most participants were young adults. The lack of experience-based comparisons also leaves open whether actual users would show the same emotional pattern.
Sam: The paper recommends things like clearer communication and emergency features. Were those approaches tested?
Alex: No. Those are proposed implications, not demonstrated remedies. Examples include travel updates to reduce confusion, tutorials and accessible emergency features to address fear, and inclusive pricing to support fairness. The model gives reasons to investigate those approaches, not proof they work.
Sam: Who should read the full document, and where would you send them first?
Alex: Researchers studying transport adoption or emotional responses to artificial intelligence should start with Hypothesis development, then Hypothesis testing. Look at the contrasts between certainty, fairness, and the different emotions. Read the Conclusion’s limitations before generalizing. Practitioners can use Implications for policy and practice as a set of proposals to test.
Sam: And what should everyone else remember when someone describes users as simply negative about automation?
Alex: Ask which emotion, and which response. In this sample, fear, anger, and dislike tracked rejection; sadness and frustration did not. Negative emotion is not a single vote against a technology.
Sam: Thanks for listening.