Epistemic emotions concerning knowledge building or generation are important for higher education (HE) students, yet little is known about how they relate to generative artificial intelligence (AI). Guided by Pekrun’s generalized Control-Value Theory (CVT), we examined epistemic emotions associated with ChatGPT use. The sample comprised 11,427 HE students representing 127 countries (based on self-reported citizenship), who completed an online survey. Structural equation modeling showed positive relations between ChatGPT use and ChatGPT control and value. ChatGPT control was positively related to curiosity and negatively related to surprise, enjoyment, anxiety, frustration, and boredom, whereas ChatGPT value was positively related to the three positive emotions and anxiety, and negatively related to the remaining negative emotions (i.e., confusion, frustration, and boredom). The model showed broad comparability across the gender, study field, and developmental-context groups represented in the sample, providing qualified support for the cross-group stability of the relations proposed by the CVT. A linear mixed model indicated that positive emotions were more frequent than negative emotions and identified modest differences across developmental contexts. Overall, these findings provide descriptive evidence of how ChatGPT use, control, and value are associated with HE students’ epistemic emotions and may inform the future development and evaluation of educational practices, while not supporting causal or prescriptive policy conclusions.
Parkinson’s disease (PD) is a progressive neurodegenerative disorder that affects both motor and non-motor functions. These implementations can significantly interfere with daily activities, particularly those that require coordination, attention, and decision-making such as driving. For older adults, the ability to drive is closely related to autonomy and quality of life, as it provides access to essential resources, social interaction, and personal independence. Therefore, understanding how PD influences emotional behavior in driving contexts is essential. In this study, the emotional responses of people with PD and a control group of healthy older adults are analyzed while driving simulated scenarios using deep learning facial expression recognition. Emotional probabilities were extracted frame by frame and analyzed to identify group differences. The findings of these experiments suggest that although general emotional patterns may appear similar, specific high-demand situations can elicit different affective states in individuals with PD. This highlights the usefulness of facial expression recognition systems in studying human behavior during complex driving tasks and supports their potential role for the design of adaptive driver assistance technologies for neurologically vulnerable populations.
Immersive experiences in virtual reality have traditionally been associated with graphical displays, but any system that artificially stimulates any of our senses can potentially simulate experiences of the real world. Recent studies have shown that multisensory compatibility can not only enhance positive psychological experiences of illusory self-motion (vection) and feelings of ‘being there’ (presence) in the virtual environment, but also reduces the adverse symptoms and severity of cybersickness. Here, we considered whether simulated viewpoint oscillation and bone-conducted vibration (BCV) enhance experiences of vection and presence when users passively view simulations of self-motion in head-mounted display (HMD) based VR. Our findings suggest that (pure-tone) sinusoidal frequencies of 500Hz BCV are effective in improving vection (i.e., the immersive benefits) of passive self-motion simulations without significantly altering cybersickness (i.e., the primary risk factor). We present a new method for quantitatively estimating the “beneficence” of these immersive systems, which favoured the use of sinusoidal 500Hz BCV over any other configuration of non-visual stimulation we explored. Results from postural sway analysis revealed “frequency entrainment” in postural sway responses to visually simulated head acceleration, which were able to significantly predict vection and presence ratings. The findings have profound significance for guiding the ethical design and validation of novel immersive systems.
Stereotypes about older adults and technology paint older adults as slow, disengaged, and unable to effectively use technology. Yet, older adults constitute a large and burgeoning user base of both traditional and emerging technologies. How do older adult adjudicate using technology themselves with the perceptions that technology “isn’t for them?” In this mixed method study, we qualitatively examine how older adults (N = 28) address aging stereotypes and the ways in which they engage with technology and quantitatively examine differences in their communication about stereotypes based on their engagement with technology. Results revealed a two-fold perception of older adults and technology. On the one hand, older adults affirmed stereotypes through direct endorsement or by suggesting that aging stereotypes were true for other older adults, but untrue for themselves (potentially as a way of psychologically distancing themselves from older adulthood). On the other hand, older adults also resisted aging stereotypes by focusing on the technology capabilities of older adults. Both perspectives aligned with assumed differences between technology engagement (i.e., low, moderate, high use) profiles. Those in the low and moderate engagement profiles more frequently endorsed stereotypes directly, while participants in the high engagement profile attributed aging stereotypes to others but not to themselves.
Face processing is a fundamental component of social cognition, but the rapid proliferation of artificial intelligence (AI) technologies capable of generating highly realistic synthetic faces is increasingly challenging how facial information is perceived. Although AI-generated faces become increasingly common in everyday environments, little is known about individual differences in behavioral and eye-gaze responses to such synthetic social cues.In this study, we recruited 46 parents (aged 28–44 years) and collected their behavioral responses and gaze patterns during a real-versus-synthetic face perception task. We found that parents showed above-chance performance in identifying AI-generated faces, with substantial individual variability. Eye-tracking results revealed greater visual attention to real faces, reflected by longer total fixation duration and more frequent fixations.As a secondary and exploratory component, we also assessed theory of mind in their children (aged 3–8 years) to examine whether individual differences in parental processing of real and AI-generated faces were associated with children’s social cognition. Exploratory analyses revealed that a stronger parental visual preference for real faces was associated with lower cognitive theory-of-mind abilities in children. This association, however, was not significantly moderated by parent-child relationship quality or daily interactions.Our findings provide converging behavioral and eye-tracking evidence that parents can discriminate real from AI-generated faces and highlight a potential association between individual differences in parental processing of synthetic social stimuli and children’s social cognition. Future research with larger samples and longitudinal designs is needed to elucidate the directionality and mechanisms underlying these associations.
Artificial Intelligence-Generated Content (AIGC) painting tools are becoming embedded in creative work, yet less is known about how they mediate creators’ interactions with technology across workflows and professional contexts. Drawing on semi-structured interviews with 24 practitioners in visual and related creative fields, this study uses grounded theory to examine how creators experience and negotiate AIGC painting tools in practice, and interprets these findings through technological mediation theory. The results show that these tools are used mainly in the early and intermediate stages of creation, where they support ideation, reference generation, visualization, and iterative exploration. However, they also redistribute rather than remove labor by creating new work around prompting, comparing outputs, selecting alternatives, and revising results. Participants described a negotiated and asymmetrical form of human–AI creative interaction in which creators retain agency primarily through setting directions, making aesthetic judgments, and choosing among generated outputs. Evaluations of AIGC painting tools varied by professional context and self-reported proficiency, and were further shaped by concerns about copyright, platform dependence, governance, and changing labor expectations. Overall, creator–AIGC interaction can be understood as a dynamic process of contact, embedding, negotiation, judgment, and adaptation. By linking workflow reorganization, role negotiations, agency, and institutional conditions, this study clarifies how technological mediation connects tool use with broader questions concerning human–AI boundaries.
Technology-Facilitated Sexual Abuse (TFSA) has been widely examined from a victimisation perspective, yet little is known about perpetration, particularly among men who have sex with men (MSM) in Chinese contexts. This qualitative study explores the manifestations, motivations, and socio-digital conditions underlying TFSA perpetration among MSM in Hong Kong. Sixty cisgender MSM (aged 18–53 years) were recruited between October 2023 and August 2024 and participated in in-depth, semi-structured interviews. Data were analysed using reflexive thematic analysis and interpreted through Bronfenbrenner’s Socio-Ecological Model.Participants described diverse forms of perpetration, including digital sexual harassment, image-based sexual abuse, sexual coercion, and gender- and sexuality-based harassment. Perpetration was driven by validation seeking, emotional retaliation, thrill-seeking, and prior victimisation, and sustained through cognitive rationalisations such as harm minimisation, implied or transferable interpretations of consent, and perceived immunity from consequences. At interpersonal and community levels, peer reinforcement and digital sexual scripts within online MSM spaces normalised image-sharing and attenuated perceptions of harm. Structural factors, including platform affordances, anonymity, and gaps in legal awareness and sexuality education, further shaped opportunities for abuse and accountability.Findings reconceptualise TFSA perpetration as a socially learned and technologically situated practice embedded within interacting individual, relational, community, and structural systems. Effective prevention, therefore, requires multi-level strategies that address cognitive distortions, peer norms, platform governance, legal communication, and Comprehensive Sexuality Education that explicitly incorporate digital consent and ethical image-sharing.
Understanding how environmental characteristics influence human well-being is increasingly important in the face of rapid urbanization and global environmental change. Virtual reality (VR) is well suited for this, as it enables ecologically valid and controlled, immersive, and repeatable exposure to complex environments. The present VR study examines responses in simulated forest and urban environments. The study proposes an experimental procedure which includes a baseline assessment, a standardized cognitive fatigue induction, to model a more ecologically realistic pre-restoration state, followed by a six-minute VR exposure. Effects on perceptual, cognitive, and emotional restoration were measured with perceived restorativeness (PRS), cognitive performance, affective valence, and arousal, while sense of presence, cybersickness, and alexithymia were examined as VR-specific and individual-difference factors. The forest environment yielded significantly higher overall PRS scores (p < .001, d = 1.08) and higher scores across all PRS-subscales compared to the urban environment, with large effects on Being Away (d = 1.07) and Fascination (d = 1.03) scales. Regression analyses showed that sense of presence was a significant positive predictor of PRS (b = 1.66, p < .001), whereas cybersickness and alexithymia were not. In the more demanding task, a significant Phase × Condition interaction, revealed the cognitive performance improving following VR forest exposure (p = .013, ηp2 = .08). Affectively, the forest VR increased valence more than the urban VR, and valence changes were positively associated with PRS. Overall, this study contributes methodological insights into the use of immersive VR as a research proxy, multiple assessment points and standardized fatigue induction which were effective in inducing the expected effects on users cognitive and emotional state and offer a structured approach for investigating restorative processes across environmental conditions.
Vehicle teleoperation is gaining relevance not only as a fallback solution for autonomous vehicles but also as an independent mode of vehicle control. While a common design goal of teleoperation interfaces is to increase immersive qualities and the telepresence experience, it remains unclear whether this merely improves user experience or if telepresence is also related to operational outcomes, such as increased situation and spatial awareness and remote driving behaviours. This test-track study compared two immersive features - vestibular feedback via a motion platform and surround-view via video-stitching - against a baseline remote control version and in-car driving to manipulate the level of presence and examine its relation to user experience and operational outcomes. Both immersive features increased the users' presence experience. Presence experience was related to spatial awareness, marginally related to situation awareness, and strongly related to risky driving behaviours, specifically the number of harsh brakings. Anecdotal observations also supported a "video game effect," where individual differences were observed in how participants treated the remote setup: while some acted with the utmost caution, aware of real-world consequences, others treated the setup like a video game and appeared indifferent even when they crashed. Overall, the study implies that presence experience is a meaningful design target not only to enhance user experience but also to mitigate risky control behaviours. We advocate for the inclusion of risk perception in the remote environment as a dimension of presence experience.
Background Parental phubbing is an emerging family risk factor for the psychological health of children and adolescents. This study addressed the longitudinal reciprocal relations and gender differences between paternal phubbing, smartphone addiction, and emotional symptoms among adolescents. Methods A two-wave longitudinal study was conducted involving 1252 junior high school students (Mage = 13.47 years, 46.8% female) in China via online surveys. A cross-lagged design across a year (with a 12-month interval) was employed to test a theoretical model in which paternal phubbing at T1 predicted emotional symptoms at T2, with smartphone addiction (T2) as a mediator. Results The results indicated that: (a) Paternal phubbing (T1) indirectly predicted emotional symptoms (T2) via smartphone addiction (T2). (b) Female adolescents are more vulnerable to paternal phubbing and emotional symptoms. Conclusion The results highlight the potentially detrimental impact of paternal phubbing on emotional symptoms among adolescents; adolescent females are more susceptible to the negative effects of paternal phubbing and emotional symptoms. The results clarify the link between paternal phubbing and adolescent emotional symptoms mediated through smartphone addiction. Specifically, fathers should pay more attention to reducing phubbing when they stay with female adolescents.
Digital platforms, personal archives, social media profiles, and emerging AI tools are reshaping the afterlives of information: how personal data, digital remains, and memory continue to matter after death. People increasingly face decisions about what should be accessed, preserved, deleted, kept private, remembered, or mediated through AI, yet these practices have not been examined as a coherent human-technology domain. Drawing on survey data from a large sample of 1,735 adults, this study introduces death-related information behavior as a multidimensional construct that captures how people engage with information before and after loss: from managing another person’s materials and planning for one’s own posthumous information, to avoiding emotionally difficult content, using digital spaces for remembrance, and considering AI-supported assistance. Descriptively, participants reported greater engagement with information following another person’s death than planning for their own posthumous information. Exploratory analysis yielded a provisional six-component solution that has not yet been validated in an independent sample. Everyday AI use was associated with greater openness to AI-supported bereavement, digital remembrance, and posthumous information management. By connecting death, memory, personal data, and AI, the study advances research on digital death, bereavement technologies, and emotionally sensitive information systems, while offering a framework for examining how technologies mediate responsibility, avoidance, access, and continuity after loss.
Problematic short video use (PSVU) has emerged as a growing behavioral concern, yet its association with cognitive flexibility remains unclear. The present study adopted an exploratory approach to examine cognitive flexibility in individuals with PSVU using a task-switching paradigm. A total of 177 participants were recruited, including 87 individuals with PSVU and 90 control participants. Participants completed cued and fixed task-switching conditions to assess switch costs and preparation effects. Reaction time (RT) data were analyzed using linear mixed-effects models. Results showed that the PSVU group exhibited significantly longer RTs on switch trials and greater switch costs compared with the control group. An exploratory analysis further indicated that the association between fixed-condition switch cost and preparation effect was weaker in the PSVU group. Overall, the findings indicate that PSVU is associated with greater difficulty in task switching, particularly when behavior must be adapted to changing task demands, and raise the possibility of differences in the use of advance task information during task switching.
We conducted a longitudinal study with the primary aim of exploring how exposure to coworker phubbing (phone snubbing) during shared breaks relates to organizational commitment and work engagement, via the proposed mediating variables social support from colleagues and sense of community at work. Coworker phubbing refers to situations in which one or more individuals attend to their phone during face-to-face interactions with coworkers, thereby potentially disrupting communication and social engagement. The study was conducted across three waves, spaced six months apart, in a sample of the Swedish working population (n = 887) recruited using probability-based methods. A structural equation modeling approach was used in the analysis. Phubbing and the psychosocial work environment outcomes were significantly and negatively correlated at baseline. However, we found no evidence of mediational effects of phubbing on organizational commitment or work engagement in the present sample and time lag. Neither did we find any longitudinal effects of phubbing on social support or sense of community, or of social support and sense of community on organizational commitment and work engagement. Thus, our findings suggest that the degree of perceived coworker phubbing behaviors during shared breaks may serve as a concurrent indication of underlying work environment issues, while not providing evidence for longitudinal effects under the conditions of the present study.
Academic procrastination is prevalent among university students and often associated with poor grades and wellbeing outcomes. While traditionally assessed with self-report instruments, here we leverage on over 1.4 million interactions with the university’s learning management system (LMS) from 563 freshmen (277 males; 286 females) across their first semester at university to passively identify profiles of those at-risk of procrastination. Profiles were derived from LMS behaviours in the 72h period prior to 6,458 assignments, and area under the curve (AUC) values of these profiles were used to classify students into Low, Moderate and High Procrastination phenotypes based on a tertile split. These LMS-driven phenotypes were found to align with survey-based procrastination reports indicating higher levels of self-reported procrastination (F(2,443)=4.86, p=0.008), as well as with lower conscientiousness levels (F(2,433)=3.63, p=0.02) in High compared to Low Procrastinators consistent with prior literature. High Procrastinators also submitted more assignments in the last 24h (78% compared to 50% and 20% in Moderate and Low Procrastinators), and achieved lower end-of-semester academic grades (F(2,532)=19.33, p<0.001). Furthermore, integration of these profiles with longitudinal, objective sleep tracking and time use diaries revealed that High Procrastinators consistently slept shorter, later and more irregularly than Low Procrastinators (all p<0.005), and were more likely to engage in social activities close to bedtime (all p<0.05) despite reporting similar aspirational sleep schedules (all p> 0.37). Our findings support LMS as a passive and scalable early detection and intervention tool to improve student sleep and academic performance through enhanced time-management and self-regulation strategies.
This paper envisions a future where people engage as immersive life-sized, three-dimensional representations, with remote interactions feeling as natural and encompassing as physical presence. For this to become ubiquitous, it is imperative to understand the attitudes of potential users and adoption patterns. To this aim, we present the first empirical investigation where real end users experience an immersive telepresence system as a medium for real-time communication, similar to existing digital communication media. To maximize ecological validity when exploring user perceptions, we designed a realistic prototype system that allowed participants to experience a tangible and credible approximation of what future immersive systems could deliver. In our design, participants could perceive their communicating partner as a life-sized 3D photorealistic representation in real time through mixed reality headsets. This allowed perceptions, user experience, and behavioural outcomes to emerge more authentically, instead of relying on hypothetical vignettes.Our mixed-methods analyses showed that participants perceived the immersive call as more realistic, engaging and socially present, given a sense of shared space. Their willingness to adopt it increased markedly, approaching parity with video calls despite the technology’s much earlier maturity. Participants identified additional benefit for professional and clinical use, and great potential as a credible alternative to in-person consultations, provided sufficient fidelity and reliability. Concerns about headset comfort, motion sickness and technical stability were raised, but also willingness to upgrade in the future, indicating that current barriers relate to practical implementation rather than user acceptance of the underlying paradigm. The study demonstrates the promise of this innovation and highlights the importance of user-centred design for its ongoing development.
Online gambling operators collect vast amounts of consumer data to track behaviour and personalise inducements. Under the General Data Protection Regulation (GDPR), access to these data requires user consent, obtained via pop-up consent banners. However, such banners tend to exhibit “dark patterns” or designs which nudge users towards accepting data sharing. To date, little is known about the extent of such dark patterns in online gambling. Here, we investigated the format of consent banners used on UK gambling websites and assessed how they may influence behaviour. Study 1 reports an audit of all UK-licensed gambling sites (n = 624). We found that 86% of banners exhibited at least one dark pattern, while GDPR infringements were also identified: 24% of sites offered no option to reject tracking and over two-thirds (67%) processed personally identifiable data prior to obtaining consent. Only 1 in 7 or 14% of sites were GDPR compliant. In Study 2, an online randomised experiment was conducted, whereby participants (n = 615) were assigned to different consent banners within a simulated gambling platform. The most common banner identified in Study 1 significantly increased acceptance of tracking and produced significantly lower alignment between user choices and preferences. No association between banner decisions and self-reported gambling harm severity was detected. Overall, our findings reveal widespread dark patterns and non-compliance of GDPR requirements on UK online gambling sites, influencing users towards unintended data sharing. These findings highlight the need for stronger regulatory enforcement and stricter consent banner design standards to protect consumers.
Domestic humanoid robots may be evaluated through cognitive, social-contextual, and affective-relational expectations before use, but mixed-methods evidence among digitally familiar older adults remains limited. This exploratory cross-sectional mixed-methods study analyzed survey data from a selective, predominantly community-based sample of 170 relatively active, digitally familiar older adults in South Korea, with male participants and those aged 60–64 predominating, and conducted interviews with 10 participants. Acceptance Intention was treated as prospective willingness to try or use a domestic humanoid robot in anticipated contexts rather than as actual adoption. Exploratory quantitative analyses were integrated with thematic analysis to contextualize, extend, or complicate patterns.Across the domain-specific specifications, Social Influence and Intimacy showed more consistent conditional alignments with Acceptance Intention than Effort Expectancy and Empathy, respectively. In the full four-predictor specification, Social Influence, Intimacy, and Effort Expectancy were statistically distinguishable from zero, whereas Empathy was not. Empirical proximity between Social Influence and Acceptance Intention and between Intimacy and Empathy limits claims of construct separability.Emotional Reassurance contextualized the Intimacy-related pattern through predictable, low-burden support, while Substitutive Intimacy extended relational comfort through companionship-like availability and routine interaction. Cognitive Identification contextualized Social Influence through social visibility and role appropriateness and complicated the Empathy-related pattern through ambivalent interpretations of human-like responsiveness. The study offers bounded mixed-methods evidence and conceptual clarification rather than model validation, causal inference, or evidence of actual or sustained use. Functional clarity, user control, ethical safeguards, and transparent disclosure of machine limitations remain provisional priorities for prototype, field-based, and longitudinal evaluation.
Deepfakes, AI-generated media that modify images, videos, or audio, represent a rapidly expanding digital phenomenon. While prior research has emphasized deepfakes in misinformation, limited attention has been given to user interaction with non-malicious deepfakes. This study employs the Approach–Avoidance Motivation Framework to examine motivational orientations and applies Goal Framing Theory to assess normative, hedonic, and gain motivations. Findings indicate that social media news use and skepticism positively predict verification intentions, whereas government engagement does not. Perceptions of reasonable use and non-deceptiveness influence ethical acceptability, which does not significantly predict viral behavioral intention. Perceived enjoyment, gain motivation, and verification intentions positively affect viral behavioral intention toward educational and entertainment-oriented non-malicious deepfakes, with gain motivation exerting a stronger effect among highly conscientious individuals. These results advance theoretical understanding of user engagement with synthetic media and suggest that governance and digital literacy initiatives should differentiate educational and entertainment-oriented non-malicious deepfakes from higher-risk forms, balancing creative engagement with transparency and responsible use.
Research on perceptions of gaming toxicity has focused on its verbal forms, while gameplay-related forms remain relatively understudied. Furthermore, assessments of gameplay actions as toxic may depend on situational considerations. The present study used a controlled experiment to examine player interpretations of gameplay sabotage using pre-recorded gameplay scenarios from the game, League of Legends. We found that the presence of gameplay sabotage elicited stronger negative emotions and triggered greater player-reported intentions to retaliate and correct the offending player than in comparable instances of unintentional poor gameplay. Further exploratory analyses suggested that participants rated poor gameplay as less acceptable in more competitively framed gameplay settings, even without the presence of overt gameplay sabotage. Exploratory analyses also suggested that players reported a greater likelihood of correcting poor gameplay when anticipated future interaction with the offending individual was low. These findings provide novel insights into how intentional gameplay sabotage and broader gameplay contexts such as competition and anticipated future interaction may independently shape players’ interpretations of poor gameplay in competitive multiplayer online games.
The co-occurrence of smartphone addiction and depression is a growing concern among adolescents. From a network perspective, this comorbidity may arise from dynamic interactions between symptoms that bridge these two issues. However, previous studies have rarely examined these associations at the symptom level or identified central symptoms underlying the overlap between smartphone addiction and depressive symptoms among adolescents. To address this gap, this study investigated the longitudinal relationships between smartphone addiction and depressive symptoms in 936 Chinese adolescents (Mage = 15.93, SD = 0.52; 38.03% girls) using Cross-Lagged Panel Network (CLPN) analysis over a three-month interval. The findings indicated that specific depressive symptoms, namely ‘Meaninglessness’ (Out-Expected Influence [OEI] = 1.76) and the smartphone addiction symptoms such as ‘Phone tolerance’ (OEI = 0.70), predicted the subsequent development of symptoms for each mental health issue. ‘Meaninglessness’ showed the highest OEI across all symptom nodes, whereas ‘Phone tolerance’ demonstrated the highest OEI specifically within the smartphone addiction symptoms. Furthermore, ‘Meaninglessness’ (Bridge Expected Influence [BEI] = 2.06) and ‘Apathy’ (a depressive symptom; BEI = 1.47) were identified as critical bridge symptoms linking the smartphone addiction and depressive symptoms. These findings provide evidence of temporal predictive associations between smartphone addiction and depressive symptoms and identify ‘Meaninglessness’ and ‘Apathy’ as key candidate symptoms that may warrant further attention in future prevention and intervention research.
