ABSTRACT. Social media is one of the leading forms of communication among young people today, with over five billion people globally engaging daily with platforms such as Instagram, Facebook, TikTok, and Snapchat. A wealth of research has explored its (potential) impacts on individuals’ psychosocial outcomes, such as their psychological wellbeing. However, most empirical research relies predominantly on subjective, retrospective self-report measures of social media use (SMU; e.g., “hours spent online”), ignoring how individuals behave on SMU platforms and potential inter-individual variability. Moreover, previous research typically utilises correlational, cross-sectional designs that are unable to determine cause and effect. To address these methodological gaps, we developed the Social Networking Site Behaviour Task (SNSBT) – a novel, experimental task designed to objectively measure individuals’ behavioural styles of social media use (SMU). Modelled closely on the interfaces of Instagram and Facebook, the SNSBT simulates a realistic social networking environment while enabling precise measurement of engagement patterns that reveal distinct styles of SMU. This approach allows us to move beyond retrospective self-report measures to directly observe naturalistic digital behaviour and investigate whether manipulations of the social media environment (e.g., content) have a causal influence on a range of psychosocial variables.

In a first preregistered study, we examined whether participants (N = 509; 261 females; Mage = 31.00) displayed distinct SMU styles on the SNSBT and how these related to psychosocial wellbeing. We further assessed whether such relationships were underpinned by individual differences in social competencies (“mediators”). Findings indicated three distinct styles of SMU – interactive, reactive and passive – that were differentially related to various indices of psychosocial wellbeing (e.g., social bonding) and mediated by interpersonal competencies (e.g., social motivation). This first study validated the SNSBT, revealing individual differences in SMU styles that may explain variability in psychological wellbeing and the importance of considering individual differences that underpin such styles.

A second preregistered study (N = 233; Mage = 22.60) extended this work using a longitudinal design, data collected over 3 days, to examine whether momentary positive and negative affect states relate to distinct SMU style on the SNSBT. Findings indicated that participants experiencing higher positive affect demonstrated a greater tendency toward interactive engagement, whereas negative affect did not significantly predict SMU style. These findings suggest that when positive affective state varies, so does interactivity on social media, underscoring the need for longitudinal designs in social media effects research.

Finally, a third preregistered study (in progress) examines whether experimentally manipulating the emotional valence of content shown to participants during the SNSBT impacts positive and negative affect. Participants will be randomly assigned into one of three experimental conditions that vary the proportion of positive, negative, and neural content shown. They will complete pre- and post-measures of positive and negative affect. This study will be the first to use an objective, experimental measure of SMU styles to assess how exposure to different forms of emotion content influences immediate mood and social media engagement. This research will allow us to determine whether the content on social media, independent of user prior affective state causally shapes online behaviour.

Together, these three studies fundamentally advance methodological approaches to understanding the impacts of social media on psychosocial outcomes: they highlight the need for research to utilise objective measurement and experimental, longitudinal designs.

Note: Elements of this programme of research have been presented at internal postgraduate events, and the first study is published in “Computers in Human Behavior Reports”. However, Study 2 and 3 present novel work that has not yet been presented or published. This presentation will therefore report three empirical studies forming the core of my PhD research.

ABSTRACT. This abstract presents a developing programme of research examining the potential developmental, behavioural, and psychosocial implications of Australia’s ban on social media use for children under the age of 16. The policy is designed to protect young people from online harms. This raises questions about how limiting access to major communication platforms may affect social connectedness, identity formation, and wellbeing during critical periods of development. While existing research has focused heavily on the risks of social media (Orben & Przybylski, 2019; Sala, Porcaro & Gómez, 2024), research is yet to establish how restricted access may alter normative developmental processes, peer relationships, and opportunities for social participation. This work contributes to cyberpsychology by examining the ban through a lifespan perspective that considers both risks and relational affordances of digital communication. Developmental frameworks underscore that adolescence (approximately 10–17 years) represent periods of heightened sensitivity to peer belonging, identity exploration, and autonomy (Crone & Dahl, 2012). Social media plays an increasingly central role in supporting these processes, facilitating connection and participation in peer cultures (Avci, Baams & Kretschmer, 2025). Restricting access during this period may therefore alter pathways of social capital development, impacting bonding, bridging, and linking relationships (Putnam, 2000). This has the potential to reshape how young people navigate friendships and group belonging. Evidence suggests that digital communication can reduce barriers for adolescents experiencing geographical isolation (Knowles & Danzi, 2025), minority stress (Berger et. al., 2022), disability (Shelton et. al., 2024), or mental health challenges (Yuan, Davidson & Best, 2025), offering flexible modes of participation not always available offline. Given this, the ban may have experience different impacts across cohorts, with some groups potentially more disadvantaged by reduced opportunities for mediated connection. This research programme applies a human development lens to explore how the ban may interact with adolescents’ evolving cognitive capacities, socio-emotional needs, and behavioural patterns. From a wellbeing perspective, the ban may lessen exposure to harmful algorithmic content or online harassment, but it may also inadvertently constrain avenues for support-seeking, identity exploration, and relational maintenance. Drawing on ecological models of development, the policy is understood not merely as a restriction on technology use but as a structural intervention that reconfigures adolescents’ communicative environments across their home, school, and peer environments. Understanding these dynamics requires balancing ethical tech design, developmental appropriateness, and adolescents’ rights to participation and expression. The project also considers broader socio-technical implications. Platforms currently used by young people, including messaging services, gaming-based communication, and social networking tools, serve as infrastructures for community-building and informal learning. Removing access may shift social interaction into less regulated or harder-to-monitor spaces, potentially amplifying inequities in digital literacy and access. As governments increasingly intervene in youth digital environments, evidence is urgently needed to clarify how such measures shape developmental trajectories. Through this orientation, this work contributes to ongoing debates in cyberpsychology, digital health, and ethical technology governance by examining how protective policies intersect with developmental needs for connection. By situating the Australian social media ban within a human development framework, the work aims to inform evidence-based decision-making that supports adolescent wellbeing while recognising the relational and communicative importance of digital participation in contemporary youth development.

ABSTRACT. The increasing use of social media into everyday life particularly, parenting has reshaped how care, involvement, and responsibility are portrayed and evaluated. While overparenting has been associated with parental anxiety in offline contexts, less is known about how these behaviours are expressed through social media or how they relate to parental mental health. This research examines overparenting behaviours on social media and their association with parental anxiety using an exploratory sequential mixed methods design. Study One used a qualitative approach to explore parents’ perceptions and observations of parenting behaviours both offline and online. Two focus groups were conducted with 12 parents who actively use social media and have children aged 18 to 20 years. Participants were asked to reflect on their views, opinions, and observations of parenting behaviours within their networks both online and offline. Discussions focused on perceived norms, motivations for posting and monitoring, social expectations, and emotional responses to engaging with parenting related content. Data were analysed using reflexive thematic analysis. The analysis identified two overarching themes, protection and reputation, with subthemes including going to extremes, over involvement, control, snooping behaviours, social comparison, social conformity, curating idealised images, and sharenting. Participants described social media as intensifying parental involvement and perceived responsibility, with constant monitoring and sharing frequently framed as acts of protection for both their children and their parental reputation. These perceptions were associated with feelings of worry, pressure, and emotional strain, suggesting parents’ motivation for their behaviours on social media. Findings from Study One informed Study Two, which focused on scale development. Based on the qualitative themes, an initial pool of 52 items was generated to capture social media related overparenting behaviours as perceived and portrayed within online parenting contexts. Items underwent experts’ review to assess clarity, relevance, and content validity. Following this process, a 30-item survey was administered to a broader sample of parents who use social media and have children aged between 18 and 20 years. Exploratory factor analysis was conducted to examine the underlying structure of the scale and refine item composition. Results from Study Two supported a reduced set of items reflecting distinct dimensions of overparenting on social media. These findings informed Study Three, which aimed to establish the psychometric properties of the new measure. Based on analyses from Study Two, the scale was reduced to 19 items. The final survey comprised 52 items in total, including the newly developed overparenting on social media scale, an established overparenting measure, and the Depression, Anxiety, and Stress Scales. Confirmatory factor analysis in Study Three indicated acceptable model fit and internal consistency for the overparenting on social media scale. This research advances psychological and cyberpsychology literature by clarifying how social media environments shape perceptions and portrayals of overparenting and are associated with parental anxiety. The findings demonstrate that specific social media related parenting behaviours, rather than overall social media use, are linked to parental anxiety. The newly developed scale provides a validated tool for examining overparenting as it is perceived and portrayed within digital environments and may support future research into the psychological mechanisms connecting social media engagement and parental mental health, as well as educational and preventative approaches aimed at supporting healthier parental engagement online.

ABSTRACT. This paper examines user awareness of TikTok’s data collection practices and analyzes how these practices influence users’ perceptions of privacy, trust, and platform transparency. As TikTok continues to grow as a dominant social media platform driven by algorithmic personalization, concerns regarding the scope, storage, and use of personal data have become increasingly salient. To investigate these concerns, the study adopts a mixed-method research design that combines a fifteen-question quantitative survey with qualitative analysis of documented privacy and data-misuse incidents involving TikTok. The survey captures users’ awareness of data collection, background activity, algorithmic influence, and privacy controls, while the qualitative component contextualizes these perceptions through real-world cases, including regulatory actions, legal settlements, and incidents of unauthorized data access.

The findings reveal a pronounced gap between TikTok’s data-handling practices and users’ understanding of how their personal information is collected, processed, and shared. Although many participants continue to use the platform, the results indicate persistent uncertainty and diminished trust, particularly concerning background data collection, algorithmic manipulation, and potential access to user data by external or foreign entities. This disconnect highlights broader challenges in social media privacy governance, where complex policies and opaque data practices limit meaningful user consent. The study underscores the need for greater transparency, clearer communication of privacy policies, and enhanced user-facing privacy protections to rebuild trust. Artificial intelligence tools were used solely to assist with clarity and wording; all research design, analysis, and interpretations were conducted by the authors.

ABSTRACT. Affordance theory offers a powerful framework for understanding how users perceive and enact possibilities for action in digital environments. Rather than treating platform features as fixed technological properties, affordances represent the perceived or imagined opportunities that either enable or constrain behaviour (Evans et al., 2017; Ronzhyn et al., 2022). This perspective is increasingly relevant for examining the relationship between social media use and social withdrawal, including digital hikikomori—a form of extreme isolation characterised by remaining at home, experiencing psychological distress or functional impairment, and spending extended periods online (Kato et al., 2019; Tateno et al., 2019). While social media can both intensify and alleviate isolation, emerging evidence suggests that passive use (e.g., scrolling, watching without interacting) may increase withdrawal risk, whereas active use (e.g., messaging, commenting) may help rebuild social connection (Gavin et al., 2025; Shah & Junaid ul Shafi, 2024).

To explain these mixed effects, Park and Yap (2024) proposed a seven affordance model organised across individual, community, and societal levels of social reintegration. Their framework emphasises how platforms support specific reintegration relevant activities—such as sharing personal stories, building relationships, and transitioning to offline engagement—rather than simply how much platforms are used. However, no empirical research has examined how these affordances are distributed across platforms that are predominantly used actively versus passively.

This study provides the first analysis of perceived affordances across popular social media platforms differing in typical use patterns. A sample of 838 UK based young adults (18–24 years) recruited via Prolific identified the platform they used most actively and most passively over the past month. Participants then rated the perceived usefulness of that platform for each of Park and Yap’s seven affordances. Analyses focused on eight widely used platforms previously categorised as supporting mainly active (WhatsApp, Instagram, Snapchat, Discord) or mainly passive (YouTube, TikTok, Pinterest, Twitch) use. One sample t tests assessed significant deviation from the scale midpoint (3) with Bonferroni correction (α < .006).

Results revealed distinct and uneven affordance profiles across platforms. Among active use platforms, WhatsApp and Snapchat strongly supported relationship building, sharing emotions, and meeting offline, while constraining threaded interaction and skills development. Discord and Instagram supported most interactional and relational affordances and showed no constrained affordances. Among passive use platforms, TikTok and YouTube strongly supported skills development and access to support resources but constrained sharing emotions, building close relationships, and meeting offline. Pinterest showed the most consistently constraining profile, and Twitch had insufficient nominations for clear interpretation.

Aligned with Park and Yap’s layered model, no platform supported all three individual level affordances, and no platform supported both societal level affordances (meeting offline and accessing support). Active platforms supported offline meeting, whereas TikTok and YouTube supported access to educational/employment resources. Overall, Instagram supported the most affordances (five), followed by Discord (four, none constrained), WhatsApp, Snapchat, and TikTok (each supporting four but with multiple constraints). These findings suggest that the affordances that contribute to social reintegration are distributed across platforms rather than concentrated within any single environment.

The study advances affordance based approaches by demonstrating that platforms simultaneously enable and constrain different reintegration relevant behaviours, complicating a simple active–passive distinction. Practically, these findings highlight opportunities for designing platform specific interventions: Instagram and Discord may be effective for multi component reintegration programmes, while TikTok and YouTube may provide low demand, observation based entry points for highly withdrawn individuals. Future research should examine these affordances directly among digital hikikomori populations and incorporate behavioural data to assess how affordances are enacted over time.

ABSTRACT. Background As social interactions increasingly migrate to digital environments, the manifestation of prosocial behavior (i.e., voluntary actions intended to benefit others) has evolved beyond traditional face-to-face dynamics. While established instruments like the Prosocial Tendencies Measure (PTM) effectively assess offline prosocial behaviors, they lack the specificity to address how helping behaviors adapt to the digital realm. Furthermore, existing digital specific scales often disproportionately focus on visible, low-cost, and performative actions, such as “slacktivism” or “like” on social media. An important gap is that they rarely reflect the full breadth of online prosociality, including how offline behaviors are transformed online (e.g., donation to online donation/crowdfunding) and how digital affordances (e.g., anonymity, immediacy, and boundary-free reach) enable new forms of helping. To address this methodological gap, this research aimed to develop and validate the Online Prosocial Behavior Scale (OPBS), a multidimensional, psychometrically sound instrument designed to capture the full spectrum of online prosociality among young adults. Method The research employed a mixed-methods design across two sequential studies. Study I (Qualitative) explored the phenomenology of online helping through focus group interviews with 20 university students (mean age = 21.5, female = 60%) recruited through purposive sampling to ensure demographic variety. Thematic analysis was utilized to identify the typology, underlying motivations, and psychological outcomes of online prosocial behavior, which informed item generation. Study II (Quantitative) focused on scale development and validation using a sample of 295 university students (mean age = 23.4, female = 87.5%) recruited on a voluntary basis. A dual-component approach was used to construct the scale: one component involved adapting items from the PTM (measures of offline prosocial behavior) to the online context, while the second involved generating new items based on Study I findings. Analytical procedures included a split-validation strategy, utilizing Confirmatory Factor Analysis (CFA) to verify the structure of adapted PTM items and the whole scale, and Exploratory Factor Analysis (EFA) to elucidate the latent structure of newly developed items. Results Findings from Study I revealed a complex spectrum of online prosocial behaviors, confirming the persistence of four established offline typologies (Emotional, Altruism, Dire, and Anonymous) while identifying three novel typologies (Civic Engagement, Online Justice, and Value-Motivated Action) unique to the digital context. In Study II, psychometric analyses resulted in a refined 24-item instrument. The CFA of the adapted component supported a 4-factor structure retaining 13 items (Emotional, Altruism, Dire, and Anonymous). The EFA of the newly developed component identified a robust 3-factor structure retaining 11 items: Civic Engagement (participating in digital community causes), Online Justice (intervening in cyberbullying or unfairness), and Value-Motivated Action (adhering to personal moral codes online). The final 24-item integrated model demonstrated excellent model fit (Chi-square/df = 1.90; RMSEA = 0.05, 90% CI [0.04, 0.06]; CFI = 0.95) and high internal consistency (α = 0.94). Conclusion We developed and validated the 7-factor, 24-item OPBS to measure online prosocial behavior among young people by integrating qualitative evidence with rigorous psychometric validation. The OPBS enables researchers to operationalize online prosocial behavior more comprehensively and test theoretically meaningful distinctions (e.g., visible vs. less visible helping; trait-like tendencies vs. context-specific digital actions), supporting stronger inference and more comparable findings across studies. For educators, youth practitioners, and policy stakeholders, the scale provides a practical tool to assess digital citizenship outcomes and evaluate interventions; for platforms and designers, it offers a framework for identifying which affordances and governance features may facilitate protective, justice-oriented, and value-driven helping online.

ABSTRACT. Engineering decisions embedded in AI systems shape how users engage with information, regulate their behaviour, and experience a sense of agency. Yet these decisions are rarely treated as sites of ethical responsibility. This paper examines how AI agent design choices function as behavioural influence mechanisms in everyday digital environments, and argues that such choices constitute an ethical practice that has been systematically under-examined in both AI ethics and cyberpsychology.

A structured interdisciplinary review was conducted using PRISMA 2020-informed procedures. Searches across ACM Digital Library, IEEE Xplore, PsycINFO, PubMed, and Scopus, conducted between January and April 2025, identified 43 sources spanning human–computer interaction, behavioural psychology, digital health, and AI ethics. Sources were included if they analysed a deployed or described AI system, documented measurable or theoretically grounded influence on user cognition, behaviour, or well-being, and were published in English between 2014 and 2025. A comparative thematic analysis was applied across four design categories: interface and interaction design, personalisation and recommendation logic, engagement and reinforcement strategies, and system-level feedback loops. Themes were derived inductively and compared across disciplines to identify convergent patterns and their underlying psychological mechanisms.

Three recurring patterns emerged, each interpreted as consistent associations rather than definitive causal relationships, as the available evidence remains predominantly correlational. First, personalisation and recommendation systems are consistently linked to informational narrowing, where algorithmically curated content reinforces prior beliefs and inflates user confidence in partial or contested knowledge. The engineering decisions most directly implicated include collaborative filtering weighted toward prior engagement and feedback architectures in which interaction data progressively refines future recommendations. Second, engagement-oriented features, particularly those employing variable reinforcement schedules and intermittent notifications, are associated with compulsive or habitual use, attentional fragmentation, and diminished perceived self-regulation. These dynamics reflect reinforcement mechanisms well established in behavioural psychology and are understood through Kahneman's dual-process framework, whereby persuasive design targets fast, automatic cognition and progressively erodes the reflective processing on which autonomous choice depends. Third, in digital health and wellness contexts, adaptive feedback systems often prioritise engagement metrics, such as daily active users and streak retention, over therapeutic outcomes, potentially reinforcing continued platform use rather than supporting durable behavioural change. Evidence in this domain varies considerably by application type, and overgeneralisation is cautioned against.

Synthesising these findings, the paper introduces the concept of a "responsibility displacement dynamic" to describe a recurring structural tendency in AI-mediated environments, whereby behavioural outcomes produced by identifiable system-level design choices are routinely attributed to individual users rather than to the systems themselves. In personalisation contexts, narrowed informational environments are interpreted as expressions of user preference. In engagement-driven systems, habitual use is framed as a failure of individual self-control. In digital health, limited therapeutic progress is attributed to user motivation. The reviewed literature consistently shows these patterns are predictable consequences of specific engineering decisions, not random expressions of individual vulnerability. This challenges the assumption that engagement optimisation is a value-neutral technical objective and raises substantive questions about algorithmic accountability and the conditions for genuine informed consent.

Situated within the Human 5.0 framework, which holds that technology should actively support human flourishing rather than simply optimise engagement, this paper reframes AI agent design as an ethical act carrying concrete obligations to user agency, dignity, and well-being. Three structural responses are proposed: integrating behavioural impact assessments into design review processes, supplementing engagement metrics with measures of user-defined goal attainment and well-being outcomes, and embedding interdisciplinary review involving cyberpsychologists, behavioural scientists, and ethicists into product development from the outset. These responses are essential if intelligent systems are to be designed in ways that genuinely serve the people who use them.

ABSTRACT. This paper examines how three psychological personality frameworks which are the Myers-Briggs Type Indicator (MBTI), Cattell's 16 Personality Factors (16PF), and the Minnesota Multiphasic Personality Inventory (MMPI) shape interaction style, perceived intent, and ethical risk when applied to AI agent design. Drawing on 48 sources (2015–2025) identified across five databases (ACM Digital Library, IEEE Xplore, PubMed, PsycINFO, and Scopus), a comparative thematic analysis was conducted across four dimensions: theoretical intent, psychometric grounding, behavioural expressiveness, and ethical suitability within a Human 5.0 frame. Sources were included on the basis that they explicitly applied at least one target framework to AI or conversational agent design; sources referencing personality only in passing were excluded. Two independent coders reviewed a subsample of sources, yielding an inter-rater agreement of κ = .81. The Big Five was excluded on scope grounds: within the sample, it appears primarily in relation to emergent personality in large language models rather than designer-configured agent personality, which is the specific problem this review addresses.

The central finding is a 4:1 ratio of proposal to evidence in hybrid personality design: of 31 sources proposing combined MBTI–16PF architectures, only eight validated them empirically. This is a gap large enough to constitute a structural problem, not a literature lag. The per-framework analysis explains why this matters. MBTI dominated applied systems and offers communicative legibility through its sixteen discrete types, but produced rigid, context-insensitive behaviour that fails to accommodate emotional complexity, cultural variation, and individual difference. Its psychometric limitations such as poor test-retest reliability and weak discriminant validity are not merely academic concerns; an agent governing thousands of interactions across unanticipated demographics risks reproducing these limitations at scale. The 16PF offered greater adaptability through its sixteen continuous, orthogonal dimensions, allowing agents to modulate traits in response to user signals while sustaining coherent personality identities. The limitation is practical: implementing this flexibility requires significantly more design effort, and communicating the resulting behaviour to users is harder still, creating transparency deficits that are most consequential in precisely the high-stakes interactions where adaptability is most needed. The MMPI, included as a limiting case, was rarely implemented directly but was referenced as a psychometric benchmark; its clinical scales risk pathologising ordinary communicative variation and repositioning the agent from assistant to assessor without user awareness or consent, rendering it ethically unsuitable for general-purpose deployment under current design assumptions.

Viewed through a Human 5.0 lens, these are not merely technical trade-offs. Each framework embeds distinct assumptions about personality - its stability, dimensionality, and what deviation from norms means - and those assumptions shape user trust, expectation formation, and the conditions under which parasocial attachment and reliance develop. This has direct implications for vulnerable populations, including adolescents in educational platforms, adults using mental health-adjacent applications, and elderly users in companionship contexts, where the consequences of misattribution are most acute. For practitioners, the findings suggest that MBTI is most suitable where interpretability and design simplicity are priorities and user contexts are relatively homogeneous; the 16PF where adaptability and contextual responsiveness are essential and transparency mechanisms can be built in; and that MMPI-adjacent approaches should not be pursued for general-purpose deployment without clinical safeguards that current agent architectures cannot support.

The paper argues that personality framework selection is an ethical decision with direct consequences for user trust, attachment, and long-term reliance and that the field cannot responsibly advance hybrid approaches until it agrees on what empirical validation of AI personality actually means. Future work should prioritise pre-registered empirical validation of hybrid architectures, standardised psychometric benchmarks for deployed agents, and longitudinal studies examining framework effects in vulnerable user populations.

ABSTRACT. Introduction Early psychotherapy sessions are clinically dense: therapists must build rapport while rapidly gathering and integrating information to decide how to proceed, with whom, and with what level of urgency. Yet, the question of what clinicians most want to know before the first session, and how Artificial Intelligence (AI) could support that decision-making, remains underexplored. This study addresses that gap by eliciting clinicians’ priorities for pre-therapy information and translating them into actionable requirements for an AI-enabled pre-session mobile screening, designed to collect open-ended patient responses and support therapists’ initial formulation and triage. Methods A cross-sectional survey was conducted with 73 licensed mental-health professionals (80.8% women; majority psychologists; 50.7% CBT-oriented) from three countries (Spain, Portugal, Italy). Participants completed an online questionnaire consisting of: (a) closed-ended Likert ratings (1–5) covering domains relevant to pre-session clinical decision-making: Patient Information (e.g., symptomatology, history), Symptom Clusters (DSM-oriented categories), Motive/Expectations/Motivation, Other Assessments (e.g., personality, anxiety, attachment), and Red-Flag Risks (e.g., suicidal thoughts); (b) their expectation on the AI efficacy for pre-therapy decision making. To see which factors are relevant for the most in pre-therapy decision making according to professionals, data were analyzed with repeated-measures ANOVAs using Greenhouse–Geisser corrections when sphericity was violated, and Bonferroni-adjusted post-hoc comparisons. Results Findings revealed a clear hierarchy in clinicians’ intake priorities: current symptomatology and clients’ self mental-health history were rated most important, while contextual background (e.g., education, work, family medical history) received substantially lower weight, with large effects indicating non-trivial differences across domains. Across DSM-related symptom domains, psychotic and severe mood presentations were prioritised over somatic/sexual and elimination-related concerns; similarly, motive and motivation for therapy were valued more than expectations. Ratings for additional intake measures showed smaller but reliable differences, with anxiety and personality functioning marginally prioritised and attachment receiving the lowest emphasis. Moreover, suicidal and homicidal ideation emerged as the top red-flag safety targets despite all risk domains being rated high overall. Finally, AI-items reported that around 30% of professionals potentially envision an AI digital tool as effective for pre-therapy decision making. Conclusion Clinical professionals converge on a clear set of pre-therapy decision-making essentials: (1) motive for seeking help, (2) motivation/readiness, (3) current symptom profile with emphasis on severe mood/psychotic presentations, (4) anxiety levels and personality, and (5) immediate safety risks. Lower-priority information, as contextual background and some dispositional factors, can be deferred without undermining first-session effectiveness. These results support a future-facing workflow where AI augments (not replaces) clinical judgment by capturing high-yield pre-session data in a structured way, enabling therapists to spend more of the first session doing what humans still do best: building alliance, occasionally rescuing the conversation from forms, checklists, and time pressure.

ABSTRACT. Introduction The widespread adoption of Generative Artificial Intelligence (GAI) in education is reshaping collaborative learning. AI-based tools can enhance engagement, self-regulated learning, and feedback access (Chiu et al., 2022; Goslen et al., 2024; Lee et al., 2024), yet AI assistance may reduce mental effort, depth of processing, and critical engagement, fostering an illusion of learning (Stadler et al., 2024; Kosmyna et al., 2025). This is particularly relevant for collaborative learning, where effectiveness rests on social interaction, intentionality, and cooperation (Dillenbourg, 1999). Productive collaboration emerges when learners are emotionally and cognitively aligned, supporting explanation, regulation, and knowledge convergence (Yang, 2023). Brief emotional priming tasks can rapidly increase closeness, trust, and cooperation even among strangers (Aron et al., 1997), enhancing interpersonal synchrony at behavioral and neural levels (Balters et al., 2023). Within Two-Person Neuroscience, cooperation and learning emerge from inter-brain synchronization (Czeszumski et al., 2020); EEG hyperscanning shows that neural coupling in theta, alpha, and beta bands reflects alignment of attention and shared representations during cooperative learning (Pan et al., 2023) and predicts better outcomes (Bevilacqua et al., 2019). The present study examines (a) how AI affects cooperation, learning outcomes, and inter-brain synchronization, and (b) whether emotional connection enhances neural alignment and learning effectiveness in AI-mediated collaboration.

Methodology The study adopted a 2×2 between-subjects factorial design manipulating AI support (present vs. absent) and emotional priming (present vs. absent). Eighty-eight unacquainted university students were randomly assigned to 44 dyads. Priming dyads completed a brief validated emotional interaction task before collaboration. All dyads then engaged in a creative collaborative learning task (~20 min) — a text comprehension activity followed by two open-ended questions — completed with or without ChatGPT. EEG hyperscanning recorded simultaneous neural activity throughout. Inter-brain synchronization was assessed across delta, theta, alpha, and beta bands using Phase Lag Index measures (Lachaux et al., 1999) and graph-theoretical indices (clustering coefficient, characteristic path length, global/local efficiency, modularity, strength). Subjective measures captured perceived learning (Feeling of Learning), product identification (modified Inclusion of Other in the Self), and team cooperation, analyzed at the dyad level via within-dyad difference scores. Objective learning was assessed through a 10-item quiz.

Results No significant effects of AI, priming, or their interaction emerged on any inter-brain network metric across frequency bands. At the subjective level, while no main effects were observed, a significant AI×Priming interaction was found both for perceived learning (F(1,40)=11.86, p=.001, ηp²=.22) and product identification (F(1,40)=6.75, p=.013, ηp²=.13). Each manipulation alone reduced within-dyad discrepancy, indicating that AI support or emotional priming independently helped partners converge on similar subjective experiences; their combination, however, neutralized this convergence. Post-hoc comparisons confirmed that priming significantly aligned perceived learning scores in the absence of AI, while AI alone amplified within-dyad discrepancy in product identification when priming was absent. No effects emerged on team cooperation, and objective learning was uniformly high (M=8.52, SD=1.36), suggesting a ceiling effect.

Conclusions Findings should be interpreted given methodological constraints: limited statistical power (~10 dyads per cell), heterogeneous ChatGPT use, and a ceiling effect on objective learning. The absence of neural effects suggests that AI support does not necessarily impair inter-brain coupling, contrasting with recent accounts of AI-induced cognitive offloading. The subjective pattern indicates that AI and emotional priming operate on overlapping rather than additive mechanisms: both can foster intersubjective alignment within a dyad, but combining them yields no further benefit. Attending to the relational and emotional context of AI-augmented learning is therefore paramount: it shapes learners' subjective experience even when objective performance is unchanged. By integrating collaborative learning theory with EEG hyperscanning, the study contributes to the understanding of neurocognitive mechanisms of social alignment in AI-supported learning.

ABSTRACT. Interest in AI-based practice tools has been increasing in counselor education; however, empirical studies directly comparing the effectiveness of chatbot-based role-play and traditional face-to-face role-play in counseling skills training remain limited. In particular, few studies have examined core counseling competencies using a multidimensional framework. Accordingly, this study adopted deliberate practice as a theoretical framework emphasizing repeated and structured training and compared the effects of chatbot-based and face-to-face role-play to explore the educational potential and limitations of chatbot use. A quasi-experimental pretest–posttest design was conducted with master’s-level counseling students who participated in weekly role-play–based training sessions over five weeks. Both groups engaged with the same simulated counseling case (a 14-year-old client presenting anger regulation difficulties in peer conflict situations). To ensure a standardized training protocol, each session was designed based on a deliberate practice model involving a sequence of initial counseling practice, immediate feedback, and targeted re-practice. The two groups differed only in the mode of client role enactment. In the chatbot-based condition, a customized Large Language Model (LLM)-based client simulation system was used, enabling repeated practice of exploratory counseling skills. Training outcomes were assessed using validated self-report measures of counselor activity self-efficacy, empathic understanding, and therapeutic presence, and were analyzed using Quade’s rank analysis of covariance with pretest scores treated as covariates. The finding that therapeutic presence was significantly higher in the face-to-face role-play group than in the chatbot-based role-play group suggests thatthe immediacy of interaction and relational context in face-to-face settings may have contributed to greater immersion and realism in counseling situations. In contrast, no significant pre–post changes or between-group differences were found in counseling self-efficacy or empathic understanding. This lack of significant change may be attributed to the characteristics of novice trainees, who may have difficulty accurately recognizing their own performance. In addition, the deliberate practice (DP) process, involving repeated practice and feedback, may have led participants to become more aware of their limitations, resulting in more conservative self-evaluations. Furthermore, the findings suggest that, in domains requiring immediacy of interaction and relational context, such as therapeutic presence, face-to-face environments may offer relative advantages. They also suggest that, in the early stages of counselor training, learner perception and the nature of self-report measures may exert a greater influence on outcomes than differences in training modality. Future research should incorporate performance-based measures and adopt more comprehensive evaluation designs to more precisely examine training effects.

ABSTRACT. In recent years, conversational artificial intelligence has become increasingly embedded in adolescents’ everyday lives, not only for instrumental purposes but also as a relational and emotional resource, particularly by young people experiencing social vulnerability or psychological distress. Evidence suggests that adolescents may disclose personal experiences to chatbots that they would hesitate to share with adults (Ada Lovelace Institute, 2023; Drexel University, 2023; Vincent, 2023). This raises important questions regarding the capacity of AI systems to provide empathic responses during a sensitive phase of identity formation. “Reflective artificial intelligence,” is defined as the ability of AI systems to reflect users' cognitive and emotional content in a structured and meaningful way (Lewis & Sarkadi, 2024), contributing to experiences of artificial intimacy and relational engagement (Vincent, 2023). From a developmental perspective, emotional mirroring plays a central role in affect regulation and self-construction, supporting emotional awareness and psychological containment. Chatbots simulate empathic dialogue through reflective reformulation, affective labelling, and supportive language - features commonly identified as central in mental health–oriented chatbots (Abd-Alrazaq et al., 2021; McStay, 2018). The present study explores whether adolescents perceive these simulated processes as emotionally supportive, and whether different chatbot designs vary in their capacity to mirror emotional distress. Using a qualitative comparative design, three conversational AI systems (ChatGPT, Replika, and Wysa) were examined. Each system was exposed to structured case scenarios of adolescent experiences (peer exclusion, social humiliation, bullying). The analysis used a 1–5 Likert scale grid of 11 indicators of empathy and emotional mirroring, such as recognition of affective states, coherence of reflective reformulation, validation of subjective experience, and perceived emotional attunement. The replicated and anonymized dialogues were independently evaluated by three experienced psychotherapists, ensuring data reliability and comparability. The findings reveal meaningful differences among the systems. ChatGPT demonstrated a relatively strong capacity for emotional mirroring, frequently recognizing users’ affective states and reformulating them in a coherent and empathic manner. However, this empathic tone was sometimes accompanied by excessive verbosity and a tendency to over-interpret emotional intensity, potentially amplifying rather than clarifying the user’s experience. Replika emphasized relational warmth and emotional closeness, adopting a consistently supportive and non-judgmental tone. While this facilitated openness and engagement, emotional mirroring appeared less precise and less differentiated across scenarios, with a tendency to prioritize reassurance and relational continuity (Ada Lovelace Institute, 2023; Vincent, 2023). Wysa showed a markedly different profile. In our analysis, it failed to provide emotional mirroring or empathic resonance, tending instead to interpret most inputs as high-risk situations and rapidly escalating responses toward crisis-management protocols. This safety-oriented and risk-averse design significantly limited its capacity to function as a supportive or reflective interlocutor in the scenarios examined. Across systems, empathic responses were experienced as supportive despite being algorithmically generated. Nevertheless, these findings highlight both the potential and the limits of simulated empathy in contexts of adolescent vulnerability. Conversational AI may facilitate recognition and articulation of distress, but it cannot replace human relationships or provide sustained psychological containment.

ABSTRACT. Introduction: The overall vision of the VR8 project was to develop, evaluate and implement a virtual reality exposure intervention with real-time anxiety tracking providing a flexible and adjustable treatment for adults with social anxiety disorder (SAD). Here we will present the results from the randomized control trial being the central study in the VR8 project. SAD is a common anxiety disorder characterized by excessive fear of being scrutinized or criticized by others, often leading to avoidance of social situations. With a lifetime prevalence of 5.5% in high-income countries, SAD is one of the most common anxiety disorders. Cognitive behavioral therapy (CBT) is an effective treatment for SAD with exposure as a central component of treatment. However, only 30-50% of individuals with SAD seek treatment. Providing flexible and adjustable treatment solutions might be more effective and help lower the threshold for seeking treatment. In general, VR-exposure has been shown to be more acceptable to patients, with lower drop-out rates than ordinary CBT. In the randomized control trial we compared CBT including exposure therapy using individually tailored VR-content and a system to track anxiety levels (CBT-ExpVR) with CBT with exposure in vivo (CBT-Exp). The primary hypothesis was that CBT-ExpVR would result in lower levels of social anxiety post treatment than CBT-Exp.

Methods: The design of the study was a parallel group, assessor-blind, randomized controlled trial. Participants self-referred online to a research clinic at the Center for Digital Psychiatry, which is a part of the Mental Health Services in the Region of Southern Denmark. The study included 80 participants diagnosed with SAD. Participants were randomly assigned (1:1) to the two interventions using computer generated tables (block randomization). The treatment was individual and manual-based, offering a 10-week program with weekly one-hour sessions. VR-scenarios included six 360˚ videos: Taking a seat on a bench in a public park, being introduced as a new employee, performing a presentation at a meeting, shopping in a grocery store, visiting a café, and using public transportation. While the participants were in the VR-scenario, heart rate and galvanic skin response were measured and a machine learning model was used to estimate anxiety levels (low, medium, or high). The therapist watched the content of the VR-scenario on a monitor, while the participants were in VR. On the monitor, the participant’s anxiety level was displayed in real time, and the therapist could adjust the VR-scenario accordingly. The outcome on social anxiety was measured using the total score on Social Interaction Anxiety Scale (SIAS). Ranges between 0 and 80. Inclusion criteria: Adults between 18 and 75 years who fulfilled the diagnostic criteria for social anxiety disorder and had sufficient knowledge of the Danish language. Exclusion criteria: Autism spectrum disorder, psychotic symptoms, severe depression (MDI >29), dependence syndrome, intellectual disability, or epilepsy.

Results: At post-test, intention to treat analysis showed no statistically significant difference on SIAS between CBT-ExpVR, M=33.65 (SD=11.98), and CBT-Exp, M=37.66 (SD=13.63). The time × group interaction was β= -1.16, 95% CI (-7.33, 5.02), p = 0.71, d = 0.10, 95% CI (-0.38, 0.56). No harms were registered during the trial. The overall completion rate was 86% with no group difference.

Conclusion: The results of the randomized controlled trial suggest that CBT-ExpVR with 360˚ video is an effective alternative to CBT-Exp. CBT-ExpVR has the potential to address some of the obstacles hindering cognitive behavioral therapists from conducting in-session exposure. VR-based exposure could play a key role in the dissemination of exposure therapy in the treatment of SAD and other anxiety disorders.

ABSTRACT. Context. The use of virtual reality (VR) exposure for the treatment of anxiety disorders has repeatedly been shown to be effective in randomized control trials comparing conditions in which exposure was conducted either in virtual reality or in vivo. However, many clinicians ask whether it would be beneficial to use VR in a more flexible manner, for example by combining both forms of exposure during treatment.

Research question. What are the advantages in terms of efficacy and cost advantages combining both forms of exposure versus using only one or the other? Because this is an exploratory study based on clinical impressions, no hypothesis was formulated a priori.

Methods. This study examines data from the waiting list control group that had not been analyzed in a previously published SAD trial, in which participants were randomly assigned to one of three groups: (a) a waitlist condition (passive control, n = 20), (b) standard individual CBT with in vivo exposure (gold standard active control, n = 22), or (c) individual CBT with VR exposure (experimental condition, n = 17). After 14 weeks, participants on the waiting list condition were _not_ assigned to one of the two experimental conditions. Instead, participants were offered a similar treatment protocol, with the exception that both forms of exposure were “combined at will” during therapy sessions, based on clinicians’ impressions and patients’ desires. In the current study, pre- and post results of these participants are being compared, as well as the costs associated with using only one exposure modality or combining them.

Results. Results revealed that clinicians and patients naturally agreed to begin with exposure in VR, then to progressively transition to exposure in vivo. Repeated measure ANOVAs revealed no significant advantage of combining exposure modalities compared to conducting exposure solely in VR, as assessed using the Liebowitz Social Anxiety Scale, the Social Phobia Scale, the Social Interaction Anxiety Scale, the Fear of Negative Evaluation, and the Beck Depression Inventory. Similar results were found for measures of treatment processes, such as the Appraisal of Social Concerns scale (likelihood and consequences), perceived self-efficacy, and working alliance. However, the costs of combining exposure modalities were statistically significantly higher than using VR alone (p < .001).

Conclusion. In the CBT of SAD, it is probably better to conduct exposure mostly, if not essentially, in VR. The addition of exposure in vivo should be considered for between sessions homework and for the very last therapy sessions, as most of the gains can be made using VR alone at a lower cost to clinicians.

ABSTRACT. Medical professionals are frequently required to deliver difficult news to patients and their families, such as informing them that a planned medical procedure has been cancelled. These conversations can be emotionally challenging and require strong communication and empathy skills. Training opportunities for such scenarios are limited, and safe environments for reflective practice are particularly valuable. Virtual Reality (VR) offers a promising platform for simulating emotionally complex clinical encounters and supporting reflective learning.

In our initial study, we developed a VR training platform that simulated a consultation with a virtual parent in which participants, embodying a doctor, were required to deliver the news that their child’s PICC line surgery had been cancelled due to an emergency. After completing the consultation, participants were able to review their performance in VR from either the parent’s first-person perspective or from a disembodied third-person perspective. We examined how perspective during the review influenced participants’ self-efficacy and self-evaluation. While no significant differences were observed between conditions, exploratory findings indicated that nurses showed a greater shift in self-evaluation compared to doctors after reviewing their performance. Nurses also reported higher levels of nervousness when speaking to the parent.

Building on these findings, the present follow-up study further investigates how perspective during VR-based reflection may influence clinicians’ emotional and evaluative responses. Specifically, we examine whether reviewing the consultation from a first-person parental perspective or a disembodied third-person perspective affects participants’ reported empathy and willingness to change their communication approach. Additionally, we assess the accuracy of participants’ self-assessment by comparing their self-evaluation scores with expert ratings of the consultations derived from the recorded VR interactions.

This ongoing study aims to better understand how perspective-taking in immersive environments may support clinical communication training, improve VR self-evaluation, and encourage empathetic engagement during difficult medical conversations.

ABSTRACT. Anxiety disorders are frequent mental health conditions in youth. Social anxiety (SA) is one of the most predominant, with estimated prevalent rates of 4.7% in children, 8.3% in adolescents and 17% in young adults. Symptoms include an excessive fear of social interactions or of being judged negatively by others causing significant distress and alterations to daily life. Public speaking anxiety (PSA) is also prevalent in younger populations, particularly in school-age children. It is characterized as induced by anxiety in public speaking settings likely resulting in poor performance, avoidance and/or great discomfort. Selective mutism (SM), on the other hand, is an anxiety-driven inability to speak in certain social conditions despite speaking normally in other conditions. It is primarily observer in young children but also found in children of primary school age and, in rare cases, in adolescence. Exposure-based interventions are typically prioritized as a treatment tool for anxiety sourced mental health conditions. Virtual reality (VR) has been shown to be an alternative treatment option able to counteract certain difficulties presented in traditional exposure treatments. The effectiveness and efficacy of conducting exposure in VR have been well documented in adults, but there is much less empirical literature for younger populations. This scoping review aims to document the available literature on the use of VR for the treatment of SA, PSA and SM in youths. Databases from Scopus and Google Scholar were searched using the keywords: ("Social Anxiety" OR "Social Phobia" OR SA OR SAD AND "Virtual Reality" OR VR AND children OR adolescents OR youth) AND ("Public Speaking Anxiety" OR PSA AND "Virtual Reality" OR VR AND children OR adolescents OR youth) AND ("Selective Mutism" OR SM AND "Virtual Reality" OR VR AND children OR adolescents OR youth) (NOT meta-analysis, adults, autism, and others). Only studies that used VR as a treatment aiming to reduce AS, PSA and SM symptoms with measures specific to these conditions in participants aged from 7 to 18 were included. With the application of our criteria, a considerable number of studies were found. After screening, a total of nine studies remained. For SA, four studies met inclusion criteria, three with children and one with adolescents. With children, two studies illustrated promising results, with RV significantly decreasing SA symptoms. Another study comparing VR to a control group of 120 children found no significant differences between both groups in the post-test. Beele et al. (2024) found promising results when using VR with ten adolescents in a school setting with symptoms significantly reduced in post-test. For PSA, considerably more research was available in adolescents. A pilot study, an RCT and an experimental between-subject study found that VR effectively decreased PSA symptoms in adolescents. Two of these studies illustrated stable results in the follow-up period. VR was also more effective than the waitlist condition in this population. Another study aimed to decrease PSA symptoms with VR involving 89 children. The results of this study revealed a significant reduction in PSA symptoms in the experimental group compared to control. Lastly, very little research was available on the use of VR for SM with only one study meeting our criteria. A feasibility study with 20 children found a significant improvement in overall functioning. However, results were conflicting, as there were no significant changes in parent-rated anxiety measures. This scoping review outlines the use of VR with younger populations to significantly decrease symptoms of SA, PSA and SM. Although the existing findings are promising and the use of this treatment tool is encouraging, the current evidence remains limited. The presentation will conclude by presenting a list of recommendations to guide future studies in youth.

ABSTRACT. Selective Mutism (SM) is an early-onset anxiety disorder characterized by a persistent failure to speak in specific social contexts despite preserved verbal ability in others. Conceptually linked to social anxiety and maintained by avoidance processes, SM poses significant challenges for in vivo exposure, particularly within school settings where communicative demands are high. Virtual Reality (VR) offers a potentially valuable intermediate step in exposure hierarchies, allowing graded, controlled, and repeatable simulations of socially evaluative situations.

This presentation will build on preliminary clinical observations from a pilot VR-based gradual exposure protocol implemented with children and adolescents diagnosed with Selective Mutism. The intervention involved immersive simulations of school-related interactions with progressive verbal demands. Descriptive findings suggested good feasibility, therapist-rated acceptability, and promising trends in goal attainment and reduced avoidance, alongside heterogeneous patterns in self-reported social anxiety—highlighting the complexity of early exposure processes. Beyond reporting preliminary data, the presentation will propose a conceptual therapeutic model integrating (1) inhibitory learning principles in exposure, (2) mechanisms of social cognition (e.g., fear of negative evaluation, perceived self-efficacy, and expectancy violation), and (3) the role of immersive environments as transitional spaces between imaginal and in vivo exposure. The model aims to articulate how VR may facilitate communicative approach behaviors while modulating threat processing in youth with SM. By situating VR-assisted exposure within broader theoretical frameworks of anxiety and social learning, this contribution seeks to advance dialogue on the role of immersive technologies in the assessment and treatment of socially mediated anxiety conditions in young populations.

ABSTRACT. Background: Children undergoing MRI experience anticipatory distress, reflecting loss of predictability and agency rather than fear alone. This distress can manifest as anxiety, reduced cooperation and increased reliance on sedation (Qamar et al., 2026). Traditional preparation approaches can reduce distress but are resource intensive and difficult to scale. Controlled trials suggest immersive virtual reality (VR) can reduce anxiety in paediatric MRI preparation (Stunden et al., 2021; van Spaendonck et al., 2023) and hospital-based initiatives demonstrate immersive 360° MRI videos can be embedded within routine paediatric care (Ashmore et al., 2019). However, translation of such methods into sustained routine care remains limited, reflecting variability in design quality, safety concerns and limited implementation-focused evaluation. Notably, most studies have not examined implementation feasibility or engaged explicitly with the processes underlying anticipatory distress or ethical constraints (Bexson et al., 2024; Psihogios et al., 2024; Sánchez-Caballero et al., 2024).

Project aim: This ongoing project develops and evaluates a child-centred VR-based preparation pathway for MRI. Rather than targeting anxiety reduction as a standalone outcome, the project conceptualises preparation as prompting sense-making and restoring a sense of agency in children. The programme supports children in understanding and inhabiting medical environments before procedures, with goals of reducing distress, improving procedural completion and decreasing avoidable sedation.

Intervention design: The project consists of two integrated VR-based components addressing distinct but complementary questions. The first is an immersive familiarisation experience designed to answer what will happen during the hospital visit. This component allows children to explore the structure and sequence of the hospital journey by recreating key sensory and temporal features of the hospital visit that drive uncertainty. The second component is a brief set of developmentally appropriate coping supports embedded within the immersive VR experience. These supports introduce optional micro-interactions focused on attentional control, tolerance of anticipated bodily sensations and simple choices that enhance perceived agency and function as rehearsal of coping strategies that children can later draw on during the procedure. Together, the two components distinguish preparation that supports sense-making from coping supports that scaffold later regulation, rather than collapsing both into a single exposure or distraction-based intervention. By reducing uncertainty and cognitive load before attendance, the familiarisation component is intended to preserve cognitive resources and coping self-efficacy while the coping supports provide structured rehearsal of how those resources can be deployed during the procedure, consistent with cognitive vitality models of regulation under stress (Howlin & Rooney, 2020). This logic is operationalised through iterative co-design with children, caregivers and clinical staff alongside a safety-first design philosophy (Bexson et al., 2024; Biswas et al., 2024).

Evaluation plan: The project adopts a staged evaluation aligned with implementation science, progressing from prototyping and usability testing within clinical constraints to feasibility assessment embedded in routine care, with a pragmatic pilot randomised study included. Outcome selection is aligned with prior paediatric MRI VR studies to support comparability across the literature (Le May et al., 2022; Stunden et al., 2021; van Spaendonck et al., 2023).

Ethics: Ethical design is treated as integral to effectiveness and is embedded at the protocol level. The protocol specifies formal ethics approval for child participation and safety monitoring informed by emerging evidence on paediatric VR tolerability (Bexson et al., 2024; Biswas et al., 2024).

Contribution: The project delivers a scalable, ethically grounded VR-based preparation framework embedded within a real children’s hospital. By operationalising preparation as a process that supports sense-making while scaffolding agency and later regulation rather than exposure alone and by pairing behavioural theory with implementation-ready evaluation, this project aims to advance paediatric VR from promising demonstrations toward dependable routine care in health systems.

ABSTRACT. Obsessive-compulsive disorder (OCD) is a debilitating condition affecting 2-3% of the population, characterised by intrusive thoughts and obsessions and the need to perform repetitive compulsions that cause dysfunctional distress and impairment. While Exposure and Response Prevention (ERP) is the gold-standard psychotherapeutic intervention for OCD, its effectiveness is often hindered by difficulties regarding stimulus personalisation and accessibility. The DIVE-OCD study was designed to address these barriers by integrating generative artificial intelligence models (GenAI) for personalisation of these stimuli, better matching them to the patient’s specific symptoms, and Virtual Reality (VR) for these digital stimuli to become potentially more immersive.

The DIVE-OCD is a randomised, controlled, proof-of-concept feasibility trial involving 45 adult participants with moderate-to-severe OCD, stratified by current therapy status (Non-ERP vs. ERP-ongoing). It is taking place as a collaboration between the Champalimaud Foundation in Portugal and King’s College London in the UK, and is funded by the UK Department for Science, Innovation and Technology. Participants are randomised into three arms: (1) OCD-VR, involving exposure to personalised, AI-generated 3D environments in a VR headset; (2) Neutral-VR, using non-threatening environments as a control for aversive content in the same device; and (3) OCD-Screen, delivering the same personalised AI stimuli via a standard 2D monitor as a control for the effect of immersion.

The seven-visit protocol begins with a comprehensive assessment (MINI 5.0.0, Y-BOCS-II) and a stimuli titration session with a GenAI model (FluxDev), where the participants collaborate with the researcher in creating images matched to their specific triggers, using a "human-in-the-loop" protocol to ensure safety and clinical relevance. Following this, participants complete a standard ERP session with a blinded therapist, followed by five consecutive days of asynchronous exposure (30-minute sessions conducted autonomously under clinical supervision for safety) with 3D generated environments (WorldLabs). The intervention concludes with a second ERP session and a qualitative debriefing interview. Primary outcomes assess feasibility, acceptability, and safety. Secondary outcomes track progression in the hierarchy of ERP exercises between the two therapist-led sessions and physiological arousal (electrodermal activity and heart-rate) throughout the protocol.

Recruitment began in January 2026 at the Neuropsychiatry Unit of the Champalimaud Clinical Centre. This presentation will describe the DIVE-OCD protocol, detailing the theoretical framework that combines standard ERP, an inhibitory learning informed approach, and a technology-assisted exposure phase. We will demonstrate the technical workflow, which ensures the safe creation of ecologically valid stimuli that aims to retain the necessary elements for therapeutic engagement. Preliminary data regarding recruitment rates, the technical stability of the pipeline, and initial participant acceptability of the protocol will be presented to illustrate the feasibility of this implementation.

DIVE-OCD aims to modernise ERP by creating on-demand, personalised triggers and by investigating the feasibility and safety of this integrated workflow. It aims to validate a scalable model that could accelerate clinical efficacy and accessibility, paving the way for potential autonomous, home-based exposure tools capable of reinforcing the therapeutic work and creating more effective exposure therapy treatment programs.

ABSTRACT. Introduction: Remote psychological interventions are increasingly recognized as scalable solutions for addressing barriers to traditional face-to-face mental health care, particularly for Social Anxiety Disorder (SAD), where avoidance and limited access to specialized services often hinder treatment uptake. Advances in immersive virtual reality (VR) technology offer new opportunities to address these challenges. VR-based exposure therapy enables controlled, immersive confrontation of feared social situations and has demonstrated efficacy in anxiety reduction. However, the feasibility of delivering structured, home-based VR interventions that maintain therapeutic engagement and support remains insufficiently explored. This study presents a theoretically grounded remote therapy protocol integrating VR-based exposure, relaxation breathing with physiological feedback, and therapist-led group consultations in immersive environments, and reports on an ongoing feasibility pilot. Methods: The intervention is informed by cognitive-behavioural models of social anxiety, emphasizing the role of avoidance, maladaptive cognitions, and heightened physiological arousal. The protocol combines three components performed in virtual environments designed to act synergistically: (1) graded VR exposure to reduce avoidance through habituation and inhibitory learning, (2) VR-guided relaxation breathing with biofeedback to support autonomic regulation and reduce hyperarousal, and (3) group consultations to enhance motivation, normalize experiences, and maintain therapeutic alliance in a remote setting. Several validation and feasibility studies were already completed to assess individual methods in the presence of the therapist. Participants diagnosed with SAD are recruited for a 5-week pilot feasibility study (target sample: N = 20). The intervention includes self-administered VR sessions at home (3–5 times per week), each consisting of exposure scenarios tailored to individual anxiety hierarchies (e.g., social interactions in cafeteria, work interview, public speaking), combined with breathing exercises. Weekly therapist-led group sessions are conducted in immersive VR platforms, allowing real-time interaction and guided reflection. Feasibility outcomes include recruitment rate, retention, adherence (frequency and duration of VR use), and usability of the system. Engagement is objectively monitored through in-app usage data. Acceptability is assessed via post-intervention questionnaires and qualitative feedback. Preliminary effectiveness is explored using a pre–post design with standardized measures of social anxiety and subjective distress, complemented by physiological indicators (heart rate variability) recorded during exposure sessions. Planned analyses include descriptive statistics for feasibility metrics, paired comparisons for symptom change, and exploratory associations between engagement, physiological regulation, and self-reported outcomes. Results: The feasibility study is currently ongoing. Preliminary observations from the first participants indicate high initial engagement with the VR system and the ability to independently conduct sessions in a home setting. Participants report that immersive exposure scenarios elicit meaningful emotional responses while maintaining a sense of safety and control. The integration of breathing exercises appears to support perceived anxiety regulation, and early feedback suggests that group consultations in VR enhance motivation and reduce perceived isolation. Full feasibility outcomes, including adherence patterns, retention rates, and preliminary symptom changes, will be available upon completion of data collection. Conclusion: This protocol addresses key challenges in remote therapy for social anxiety by combining self-directed exposure with structured therapeutic support and physiological regulation. The integration of immersive VR, biofeedback-informed relaxation, and group-based interaction represents a novel hybrid model aimed at preserving core therapeutic mechanisms in a digital context. Findings from the ongoing feasibility study will inform protocol refinement and the design of future controlled trials evaluating clinical efficacy.

Funding: This work is supported by the project “Research of Excellence on Digital Technologies and Wellbeing CZ.02.01.01/00/22_008/0004583”, co-financed by the European Union.

ABSTRACT. Sexual violence against adolescents in online environments is a public health issue, and there is already robust evidence in international literature indicating that such experiences are associated with psychological symptoms such as anxiety, depression, Post-Traumatic Stress Disorder (PTSD), and suicidal ideation. Based on this, the present study aims to develop and assess the feasibility of a psychological intervention model, grounded in Cognitive Behavioral Therapy (CBT), targeted at female adolescents who have been exposed to sexual violence in virtual settings. As a feasibility study, the sample will include four to six (4–6) female adolescents, aged between 12 and 17 years, who have experienced online sexual violence within the past year, counted from the initial date of the research's dissemination. The proposed clinical method will be implemented in three distinct stages: stabilization, trauma processing, and reintegration. To evaluate the effects of the intervention, a mixed-methods approach will be used, integrating both quantitative and qualitative instruments. The assessment tools administered at pre-test (T1), post-test (T2), and follow-up phases (T3 and T4) will include: the Sociodemographic Questionnaire for Sample Characterization – QSCA (administered only at T1), the Brief Symptom Inventory (BSI), the Beck Scale for Suicide Ideation (BSS), and the Impact of Event Scale – Revised (IES-R). To construct a qualitative database assessing participants' perceptions of the treatment received, an adapted version of the Client Change Interview Protocol (CCI) and a field journal will be used. The combination of these instruments will allow for a deeper understanding of the psychological impacts of online sexual violence on adolescent girls, as well as the potential of the proposed clinical method. The central research hypothesis is that psychological interventions grounded in the Cognitive Behavioral Therapy (CBT) approach may be effective in reducing psychological symptoms among female adolescents who have been exposed to online sexual violence. Should the results support this hypothesis, the intervention manual developed by the researcher will be made freely available to clinical psychologists. Furthermore, if promising results are observed, more robust future studies are anticipated, including the implementation of the protocol in a quasi-experimental design to evaluate its efficacy. This scientific dissemination aims to present, for the first time, the preliminary findings obtained in the present study.

ABSTRACT. Artificial Intelligence (AI) and digitalization are transforming the world of work, raising important questions regarding young people’s preparedness for labour market entry. The present study investigated young people’s perceptions of AI and digitalization in the workforce, their perceived preparedness for labour market entry, and the digital and psychological resources they consider essential for navigating these changes. Adopting a qualitative design, participants were purposively sampled and comprised students and recent graduates aged 18–29 who were approaching the completion of their studies and preparing to enter the workforce. Data were collected through semi-structured interviews (n = 8) and nine focus groups, each comprising 8–10 participants. The study was guided by the Integrated Framework of Graduate Employability, integrating digital competencies (DigComp) and psychological resources from Psychological Capital and Career Adaptability as complementary factors. Data were analyzed using a hybrid thematic analysis approach implemented in MAXQDA software. Findings revealed that young people hold divergent perceptions about digitalization and AI, acknowledging both efficiency gains and potential threats to creativity. Participants reported a moderate level of preparedness for workforce entry and emphasized the need for continuous training and upskilling to effectively navigate technologically intensive work environments. Key digital competencies for labor market entry included communication, collaboration, and problem-solving. Optimism and curiosity emerged as the most salient psychological resources, with resilience and confidence also briefly mentioned. These findings offer important implications for the design and refinement of interventions and training programs aimed at enhancing young people’s readiness for labor market entry in an increasingly digitalized context.

ABSTRACT. This study analyses the narratives of Artificial Intelligence conversation systems (Gemini/USA, RuGPT/Russia, Ernie Bot/China) and their impact on public opinion. The methodology draws on Tardian concepts of imitation, invention and opposition. Through content and discourse analysis, narrative biases aligned with the geopolitical strategies of their countries of origin were identified. In terms of imitation, it was found that the systems diverged in their prioritisation of topics and framing, despite agreeing on the facts: Gemini highlighted UN/AI institutional structures, while Ernie Bot emphasised the topic of regulation and security. In terms of invention, Gemini stood out for its geopolitical contextualisation and inferred the need for technological diplomacy. Ernie Bot, on the other hand, invented about preventing the arms race and the risk of domination by a single power, and suggested possible greater creative freedom in China. RuGPT produced inventions focused on social and cultural narratives, such as the idea of a cultural ‘bridge’ between the US and China. Finally, on the axis of opposition, the politically sensitive conflict was addressed with different emphases: Gemini on the geopolitics of technology, RuGPT on digital sovereignty, and Ernie Bot on security and defence. It can be concluded that the narrative priorities align with national strategies: Gemini focuses on defending liberal multilateralism, RuGPT on digital sovereignty, and Ernie Bot on security and cultural optimism.

ABSTRACT. Negative attitudes towards people with mental illness remain common among healthcare students and may influence later clinical practice. Educational and contact-based interventions can reduce stigma, but effects are often modest, short-lived, and difficult to sustain or scale within healthcare education. Immersive virtual reality (VR) has been proposed as a promising alternative, yet findings remain mixed. In this work, we frame these inconsistencies in terms of two positions. What we refer to as the “technological hypothesis” assumes that immersive VR, as a delivery medium, produces stronger attitudinal effects than less immersive or non-VR formats. In contrast, what we refer to as the “content hypothesis” assumes that attitudinal change is driven primarily by narrative content, with little to no difference between different technologies when content remains unchanged.

Building on this, the present study advances an “affordance-congruence hypothesis”, proposing that immersive VR adds value when its affordances are meaningfully engaged to support narrative engagement and intended outcomes. Using this account, the study examined whether activating one key affordance of immersive VR, namely real-time interactivity, changes mental health stigma outcomes and the psychological processes through which change occurs, while holding immersion and narrative content constant. The VR experience used was “Goliath: Playing with Reality”, an off-the-shelf, narrative-led game based on the lived experience of a person with schizophrenia.

Fifty-two students from healthcare programmes including nursing, midwifery, paramedic science, and psychology were randomly assigned to either a real-time interactive VR condition (n = 26) or a non-interactive VR condition (n = 26), both delivered via a head-mounted display. In the interactive condition, participants actively engaged with the game, while participants in the non-interactive condition only viewed a recording of the gameplay. Mental health stigma was assessed at baseline (T0) and immediately post-intervention (T1) using validated self-report measures of stereotypes, prejudice, and intended discriminatory behaviour. Spatial presence, narrative transportation, and avatar identification were assessed at T1.

A series of 2 (condition) × 2 (time) mixed ANOVAs revealed significant main effects of time across all stigma-related outcomes. Stereotype endorsement decreased from T0 to T1, F(1, 50) = 8.08, p = .006, ηp² = .14. Prejudice scores also decreased, F(1, 50) = 17.47, p < .001, ηp² = .26. Intended discriminatory behaviour showed increased willingness for future social contact, F(1, 50) = 8.07, p = .006, ηp² = .14. No significant main effects of condition or condition-by-time interactions were observed (all ps > .15).

Despite similar attitudinal outcomes, experiential processes differed between conditions. Participants in the interactive condition reported significantly higher presence, t(50) = 4.58, p < .001, d = 1.27, and greater transportation, t(50) = 2.00, p = .025, d = 0.56, while identification did not differ between conditions (p = .313). Structural equation modelling showed that interactivity predicted presence (β = .54, p < .001), which predicted transportation (β = .56, p < .001). Transportation predicted identification (β = .53, p < .001), and both transportation (β = −.20, p = .037) and identification (β = −.25, p = .006) were associated with lower post-intervention stereotype endorsement.

Brief exposure to a narrative-driven VR game can reduce mental health stigma among healthcare students in the short term. While real-time interactivity increased presence and transportation, this did not translate into greater overall stigma reduction compared to non-interactive VR delivery. The results suggest that similar attitudinal outcomes may be achieved through different psychological pathways depending on how immersive experiences are designed and engaged with. Future research should explicitly manipulate and measure embodiment-related affordances to determine whether they differentiate stigma outcomes beyond presence and transportation in immersive VR experiences.

ABSTRACT. During the recent years, virtual reality (VR) has been increasingly used in cyberpsychology. Despite this rapid adoption, the neural mechanisms underlying VR exposure remain insufficiently characterized and the present study seeks to advance current understanding in this domain. A central limitation is methodological: most neuroimaging techniques used in VR research lack either the spatial precision to access deep brain structures or the temporal resolution required to capture fast neural dynamics, constraining the interpretation of cognitive processes unfolding during immersive experiences.

This work introduces a methodological framework that integrates immersive VR with intracranial electroencephalography, enabling direct, high-resolution measurement of neural activity during virtual experiences. By combining VR technology with depth electrode recordings, this approach provides millisecond-level temporal resolution and precise access to both cortical and subcortical regions, overcoming key limitations of non-invasive neuroimaging methods.

The framework was implemented in clinical participants undergoing temporary intracranial monitoring for epilepsy, leveraging a unique opportunity to study neural activity during immersive VR exposure. The participants were presented with virtual environments that systematically manipulated immersion level while keeping spatial content constant, allowing the immersion level to be isolated as an experimental variable.

A comprehensive analysis pipeline was applied, including spectral parameterization and functional connectivity measures designed to capture both local neural dynamics (same neuro-anatomical structure), and interregional interactions. The analytical framework is intentionally modular and adaptable, supporting its use across diverse VR paradigms and cognitive domains.

This study examines condition-dependent neural dynamics while also demonstrating the feasibility and methodological robustness of immersive VR as an experimental tool. It positions immersive VR not merely as a delivery medium for stimuli, but as a powerful platform for probing cognition and behavior at the neural level. The proposed approach is readily extendable to the study of other cognitive and affective processes, including attention, emotion regulation, self-processing, and decision-making. As such, it provides a foundation for future research aimed at developing neuroscience-informed VR applications for assessment, training, and intervention.

ABSTRACT. The spectrum of possibilities offered by extended reality (XR) ranges from almost perfect simulations of reality and the embodiment of persons to the modification of non-verbal communication in real time, out-of-body experiences or the change of perspective through the transition from one virtual physical representation to another. Madary and Metzinger have described XR as having the potential of ‘changing the very relationship we have to our own minds’ (1). Many approaches have been developed to enrich the range of treatments for mental disorders such as phobias, paranoid-hallucinatory psychosis, dementia and many others (2). This is highly relevant because psychiatric patients may exhibit psychopathological symptoms (e.g. hallucinations requiring reality testing), which make them particularly vulnerable with regard to immersive XR applications (3). A number of general ethical concerns have been identified regarding the clinical application of XR (4). Here I would like to present a newly developed model, that attempts to link specific risks of virtual exposures to psychopathological symptoms of patients (5). With a focus on virtual embodiment and virtual social interaction, I will 1. describe the specific risks of virtual exposures, 2. discuss them in relation to specific psychopathological symptoms and 3. outline initial strategies that enable safe exposure with a strong emphasis on participatory designs. (1) Madary M and Metzinger TK (2016) Real Virtuality: A Code of Ethical Conduct. Recommendations for Good Scientific Practice and the Consumers of VR-Technology. Front. Robot. AI 3:3. (2) Wiebe A et al. (2022): Virtual reality in the diagnostic and therapy for mental disorders: A systematic review. Clin Psychol Rev. Dec;98:102213. (3) Kellmeyer P, Biller-Andorno N, Meynen G. (2019): Ethical tensions of virtual reality treatment in vulnerable patients. Nature Medicine 2019;25(8):1185–8 (4) Marloth M, Chandler J, Vogeley K. (2020) Psychiatric Interventions in Virtual Reality: Why We need an Ethical Framework. Cambridge Quarterly of Healthcare Ethics 2020; 29(4):574-584. (5) Marloth M., Deane-Drummond C., Kellmeyer P., Latoschik M. E., Chandler J. A., Meynen G., & Vogeley K. (2026). Anticipation and prevention of real risks of virtual environments in psychiatry. Npj Digital Medicine, 9(1).

ABSTRACT. Ethical decision-making is a key concept in psychological practice, requiring professionals to integrate technical knowledge with contextual judgment and emotional awareness when navigating complex dilemmas. Although ethics occupies a fundamental place in professional training, fostering these competencies remains challenging. Traditional pedagogical approaches, while attempting to balance theoretical instruction with case-based learning, often provide limited opportunities for experiential engagement, leaving students with reduced immersion in the types of situations they will encounter in practice. Consequently, a gap persists between conceptual learning and the development of applied ethical decision-making skills. In 2021, the VREthics application emerges as a solution to help bridge the gap in ethics education, enabling the training of internship psychologists and aiming for its integration into the training programs of the Ordem dos Psicólogos Portugueses (OPP).The development of VREthics has followed a cumulative trajectory of research and iterative refinement, building on earlier phases involving the conceptual design and proof of concept and a pilot implementation. The present study represents a subsequent step focused on applied evaluation from a design-based research perspective, and rather than testing hypotheses, the study aims to examine how users experience the application in practice and how their feedback can inform future improvements. Specifically, the objectives were to evaluate perceived usefulness and experiential quality, identify strengths of the tool, and derive improvement guidelines based on user feedback. The study included 105 Portuguese internship psychologists at OPP, with a mean age of 28,94 years (±7.13), who participated in a VR-based ethical decision-making training activity using the VREthics application. After the experience, participants responded to open-ended questions, enabling an in-depth qualitative exploration of their perceptions. Data were analysed through content analysis procedures, with categories informed by theoretical constructs related to immersive virtual environments, including place illusion and plausibility, as well as pedagogical and usability dimensions. The highlighted positive aspects of the application were the plausibility illusion (22.9%), the simulation of a clinical context (21%), preparation for practice (19%), interactive experience (16.2%), immersion (15.2%), pedagogical innovation (14.3%), a controlled practice environment (9.5%), involvement (8.6%), and place illusion (7.6%). Taken together, these findings suggest that VREthics was perceived as both a credible and pedagogically meaningful learning environment, one that effectively supported experiential engagement with ethical dilemmas in clinical psychology. Regarding areas for improvement, feedback primarily highlighted sensory fidelity (57.1%), pedagogical design and content (43.8%), and agency and interactivity (36.2%), with fewer comments related to overall satisfaction (5.7%). Suggested refinements included improving visual and auditory realism, expanding response options, and diversifying scenarios to enhance decision-making complexity. The findings of this study reinforce the value of VREthics application as a future tool in the ethical pedagogy of internship psychologists. It emerges as an innovative resource that facilitates the transition from theoretical instruction to real-world practice within a safe environment, strengthening knowledge by bridging the gap between theoretical ethics instruction and applied professional practice. This study not only informs the technical and design optimization of the VREthics application, but also supports its integration into the core training program, namely through the planned implementation of voice-based communication systems to promote more natural and realistic interactions, the adoption of hand tracking instead of controller-based input to enhance embodiment and intuitive engagement, and the updating of visual assets to improve overall realism and sensory fidelity.

ABSTRACT. Background: The United States Veterans Affairs/DoD Clinical Practice Guideline for the Management of PTSD identified Prolonged Exposure (PE) and Cognitive Processing Therapy (CPT) as first-line PTSD therapies. However, the largest PTSD trial to date (916 veterans) reported dropout rates of 55.8% and 46.6%, and response rates of only 40.4% and 28.8%, for PE and CPT respectively, consistent with an earlier review of the literature documenting that two-thirds of those receiving CPT or PE retained their PTSD diagnosis New therapies are needed for posttraumatic stress disorder (PTSD), particularly in military service members (SMs) and veterans. Moreover, many veterans are in rural areas with limited access to therapists, highlighting an additional need for distance-bridging therapies. Reconsolidation of Traumatic Memories (RTM) is a novel treatment with promising results in small clinical trials. Methods: We randomized 94 active duty or retired SMs with PTSD to receive up to ten 90-minute sessions of RTM (n=48) or prolonged exposure (PE, n=46, as the current standard of care active comparator) to determine whether RTM achieves a greater and/or more rapid response. The Clinician Administered PTSD Scale for DSM-5 (CAPS-5) conducted by independent, blinded assessors was used to establish eligibility and assess response to treatment at two weeks and two, six, and 12 months post-intervention. The PTSD Checklist for DSM-5 before treatment sessions two, four, six, eight and 10 assessed response rapidity. Secondary outcome measures included the PHQ-9 for depression, GAD-7 for anxiety, NSI for post-concussive symptoms, PSQI for sleep difficulties, and WHOQOL for quality of life. We hypothesized that RTM would achieve more rapid responses (early responders defined as completion of therapy in <10 sessions) and higher response (drop in CAPS-5 scores of 10 or more) and loss of diagnosis (response plus a final CAPS-5 score of <25) rates than PE. Participants’ mean age was 45.80; 31% were female. There were no significant demographic differences between groups. Results: Response (74.2% for RTM, 72.4% for PE) and loss of diagnosis (58.1% for RTM, 51.7% for PE) rates showed no significant group differences. Gains were largely sustained at two, six and 12 months. RTM achieved a significantly higher early responder rate (72.2%) than PE (27.8%) (p=0.005). Those randomized to RTM also more frequently addressed multiple traumas (70%) than those receiving PE (30.0%) (p=0.022). Withdrawal rates were 18.8% (9/48) for RTM, 32.6% (15/46) for PE (p=0.124). Therapy was completed entirely via videoteleconference by 85% of participants. Conclusions: RTM and PE demonstrated comparable large effect sizes, without evidence of a significant difference overall. However RTM was associated with significantly higher rates of early response, while also enabling participants to address more traumas. These very promising findings corroborate the results of previous waitlist control studies and a British feasibility study, but require confirmation in additional studies in order to establish RTM as a clear first-line therapy. Both RTM and PE appear well-received and efficacious when offered via videoteleconference.

ABSTRACT. Background: Attention-Deficit/Hyperactivity Disorder (ADHD) in women is chronically misunderstood due to male-normative diagnostic archetypes, leading to a profound "visibility gap" and delayed recognition. This systemic exclusion necessitates a dual approach: rigorous scientific inquiry to map clinical blindspots, combined with neuro-affirming advocacy (activism) to legitimise lived experiences. The psychosocial burden of late diagnosis often culminates in the "masking paradox," where outward competence hides internal exhaustion. Consequently, emotional dysregulation emerges as a primary, impairing mechanism. This doctoral project aims to bridge the utility gap between traditional deficit-based clinical frameworks and the functional realities of women with ADHD.

Methods: The project employs a critical realist, three-study design grounded in Self-Determination Theory (SDT). First, a systematic scoping review (Study 1) of 60 peer-reviewed sources (2015–2025) was conducted. Crucially, this review identified four specific domains of clinical misalignment: diagnostic, outcome, support, and hormonal, which directly dictated the methodological focus of the subsequent phases. Second, a participatory mixed-methods investigation (Study 2) was designed to map the hidden coping strategies of adult women, explicitly focusing on how environmental support impacts emotional regulation. Finally, a Participatory Action Research framework (Study 3) is utilised to translate these peer-validated strategies into a co-created, Functional-First Support Manual.

Results: The scoping review revealed that current care models prioritise symptom reduction over the functional outcomes women actually value. Study 2 aims to highlight the mechanics of emotional regulation: data indicate that when clinical and social environments support a woman's autonomy, her capacity for emotional regulation and daily sustainability increases, while the reliance on exhausting, introjected masking decreases. To address the fragmented care landscape, Study 3 aims to develop a "Living Library" manual designed to accommodate women at different clinical entry points (e.g., navigating assessment waitlists vs. post-diagnostic life). The manual’s strategies aim to work by replacing neurotypical behavioural expectations with hormone-sensitive, self-determined coping mechanisms, moving the individual from high-cost masking toward integrated regulation and functional mastery.

Conclusion and Future Studies: Meaningful support for women with ADHD requires a paradigm shift away from symptom-reduction metrics toward daily functional sustainability and emotional safety. By integrating scientific evidence with lived expertise, this thesis provides a structured, gender-responsive clinical output that supports women across varying stages of their ADHD journey. Future research must prioritise longitudinal studies on the interaction between hormonal fluctuations and ADHD traits, as well as the systematic evaluation of these co-created, functional-first interventions in mainstream clinical protocols.

ABSTRACT. Screen-based technologies play an increasingly central role in shaping how attention functions in everyday life. Digital environments - such as social media, messaging platforms, and online gaming - are characterized by high stimulus density, rapid information flow, and persistent notifications that continuously compete for users’ attentional resources. Excessive or fragmented screen use has been linked to attentional difficulties, including heightened distractibility, multitasking overload, and a diminished capacity for sustained focus. While contemporary theoretical models distinguish between stimulus-driven attentional capture and goal-directed attentional control, traditional attention measures - largely developed in offline contexts - may not adequately reflect how these mechanisms operate in technology-mediated settings. In line with cyberpsychology’s emphasis on behavior in digital contexts, this study addresses this gap by developing and examining the factorial structure of the Attention in the Digital Environment Scale (ADES), a psychometric instrument designed to assess attentional capture, control, and immersion in everyday screen-based activities. This study examined the factorial structure and internal consistency of the ADES in a sample of 366 adults aged 18-75 years (M = 34.14, SD = 15.03), predominantly Portuguese (88.0%) and female (62.6%). Most participants reported frequent use of social media platforms, particularly WhatsApp (91.0%), Instagram (84.2%), YouTube (53.3%), and TikTok (50.0%), primarily for entertainment, social connection, and personal interests. An exploratory factor analysis (EFA) using Principal Axis Factoring with Direct Oblimin rotation was conducted. Sampling adequacy was supported (KMO = .88; Bartlett’s test: χ²(2850) = 12,717.34, p < .001). Although 18 factors presented eigenvalues greater than 1, parallel analysis and inspection of the scree plot supported the retention of three factors. A four-factor solution yielded a weak and conceptually unclear fourth factor and was therefore rejected. Items were excluded based on low loadings (< .30), problematic cross-loadings, low corrected item-total correlations (< .30), floor/ceiling effects (> 15%), or meaningful improvements in Cronbach’s alpha. The retained three-factor solution comprised: (1) Digital Attentional Capture (35 items), reflecting susceptibility to salient digital stimuli, difficulties sustaining and shifting attention, and automatic engagement; (2) Digital Attentional Control (23 items), indexing intentional and flexible regulation of attention in online environments; and (3) Attentional Engagement in Online Gaming (6 items), capturing strong attentional immersion in gameplay and difficulties disengaging from non-game stimuli. Given the high proportion of “not applicable” responses for gaming items (20.8%-35.2%), this third factor is best conceptualized as a context-specific subscale rather than a core dimension of the ADES. Accordingly, the scale primarily yields a core total score reflecting Digital Attentional Capture and Digital Attentional Control. Internal consistency was excellent for the core total (α = .93) and Digital Attentional Capture (α = .94), and good for Digital Attentional Control (α = .89) and Attentional Engagement in Online Gaming (α = .87). Factors showed small-to-moderate correlations, consistent with contemporary tripartite models distinguishing stimulus-driven attentional capture and goal-directed attentional control as partially independent mechanisms. Overall, the ADES appears to capture context-dependent patterns of attentional functioning shaped by digital environments rather than traditional attentional subtypes.

ABSTRACT. Deepfakes allow a person’s face or body to be transferred onto another individual’s movements, producing highly realistic but artificial visual experiences. While often associated with risks such as identity theft, cyberbullying, and misinformation, research highlights therapeutic, educational, and behavioral applications, showing deepfakes can help individuals who have lost their voice, enable therapeutic exposure to past experiences (Hoek et al., 2024; Wiederhold, 2021), or enhance learning through historical figures (Babaei et al., 2025). This study investigates deepfakes as a medium for personalized video modeling. In this vicarious learning approach, individuals acquire skills or insights by observing others' actions, a method that is effective in promoting self-efficacy. Unlike traditional video modeling, this research emphasizes self-observation, fostering interaction between digital self-representation and emotional processing. It examines how viewing oneself in deepfake videos influences emotional arousal, complexity, regulation, self-perception, and reflective processes. Three hypotheses guide the study: 1) Watching a personalized deepfake will produce higher psychophysiological arousal than neutral content. 2) Self-relevant deepfakes will evoke greater emotional complexity, increasing positive and negative emotions, and affect emotion regulation pre- and post-test. 3) EG participants will show more intense emotional and psychophysiological responses than CG participants, highlighting the role of personal relevance. The study has ethical approval from the University of Palermo (no. 189/2024). Fifty adults, balanced by gender, will be randomly assigned to the EG (self-deepfakes) or the CG (neutral deepfakes from a public database). Sample size was estimated via G*Power (α = .05, power = .80, medium effect size f = .25). Participants will be randomly recruited, must be of legal age, and provide informed consent. The experiment has two phases. First, EG participants record facial movements using custom software. These are converted into deepfake videos showing them dancing, chosen for full-body movement while manipulating only the face. Observing themselves performing actions encourages self-reflection, identity recognition, and confidence. Then, all participants watch three videos—two neutral landscapes to establish baseline responses, followed by a deepfake. EG views their own video; CG views a matched neutral video of another person. Psychophysiological and behavioral data are recorded using iMotions, which combines eye-tracking, facial expression analysis, and biometric sensors. Participants also complete demographic and self-report technology questionnaires, the Emotion Regulation Questionnaire (ERQ-CA; Gross & John, 2003; Balzarotti, John, & Gross, 2010), the Visual Analogue Mood Scale (VAMS; Stern et al., 1997) pre- and post-test, and the Self-Assessment Manikin (SAM; Lang, 1980) at post-test. Pre- and post-test VAMS and ERQ-CA scores assess changes in emotional state and regulation. SAM captures emotional valence and arousal, allowing comparison between subjective feelings and objective psychophysiological responses. Combining these measures captures both emotional intensity and the influence of regulation on reactions to one’s digital self. Data collection is ongoing. EG participants are expected to show higher arousal, greater emotional complexity, and changes in emotion regulation compared to CG. Viewing oneself in a deepfake context is likely to activate cognitive and emotional processes related to self-awareness, self-efficacy, and identity. The study also examines correlations between subjective reports and psychophysiological measures, highlighting interactions between conscious reflection and unconscious affective responses. It evaluates iMotions’ sensitivity in detecting subtle responses to personally relevant stimuli. This study advances understanding of the psychological effects of interacting with one’s digital self. Experiencing oneself in deepfakes may promote self-reflection, behavioral change, and increased self-efficacy, demonstrating deepfakes’ potential for controlled psychological interventions. By emphasizing personal relevance, emotional regulation, and identity, this research contributes to cyberpsychology frameworks, showing deepfakes can serve as an innovative research medium and a tool for personalized, transformative experiences. Findings open avenues for digital self-representation in therapy, education, and behavior change, illustrating connections among technology, emotion, and self-perception.

ABSTRACT. Screen exposure, commonly referred to as digital media use, encompasses individuals’ interactions with digital technologies and virtual platforms, including smartphones, computers, televisions, tablets, and social media. A substantial body of evidence has documented the adverse effects of excessive or prolonged screen use, particularly during childhood. In the context of the COVID-19 pandemic, numerous studies investigated changes in digital media use among children and adolescents during periods of confinement and their associated impacts. Nevertheless, important knowledge gaps persist—especially within the Brazilian context—regarding how the pandemic influenced screen use among children from socially marginalized populations and the extent to which these changes are associated with emotional and behavioral outcomes. Based on these premises, this study aimed to assess screen exposure levels among Brazilian children during the COVID-19 pandemic and examine their associations with behavioral and emotional outcomes. A total of 92 caregivers of children (M = 40.1; SD = 11.6) participated by completing a questionnaire, whose final section addressed screen use before and during the pandemic. Participants were divided into two groups: Increased [screen use] = G1; Maintained [screen use] = G2. The findings revealed a significant increase in screen time during the pandemic (p < 0.001), with 66% of the children using screens for approximately 5 hours or more per day. G1 showed a higher incidence of irritability symptoms (G1 = 62.3%; G2 = 39.5%; p = 0.03) and sadness (G1 = 58.5%; G2 = 36.8%; p = 0.04), as well as higher mean scores for behavioral changes (G1 = 4.8; G2 = 3.7; p = 0.038). These findings have important implications for both clinical practice and educational settings, as they highlight increased screen exposure as a potential risk factor for emotional and behavioral difficulties in children. Clinicians and educators may use this evidence to support early screening, monitoring, and guidance for families regarding children’s digital media use, particularly during periods of social disruption. Additionally, the results can inform the development of school- and community-based interventions aimed at promoting healthier media habits and supporting children’s emotional well-being in socially vulnerable populations.

ABSTRACT. Digital platforms increasingly circulate content designed to provoke moral outrage, commonly referred to as rage bait. Research demonstrates that moralized emotional language structurally amplifies content diffusion, making outrage a design feature rather than an incidental outcome. In Malaysia, such content frequently centres on issues related to race, religion, and royalty (3R), which carry heightened sociocultural sensitivity, collective identity significance, and legal weight under the Sedition Act (1948, amended 2015) and the Communications and Multimedia Act (1998). While cyberpsychology research has extensively examined exposure to online hostility and its impact on targets of abuse, and the inhibiting factors discouraging bystander intervention, less attention has been paid to individuals who experience a perceived moral obligation to actively and repeatedly intervene.

Drawing on Moral Injury Theory (MIT), this qualitative phenomenological study examines the psychological processes and well-being consequences experienced by Malaysian university students who repeatedly confront rage-inducing online content, with particular attention to the mental and academic costs of sustained moral engagement.

Using purposive intensity sampling, 20 Malaysian undergraduate students (aged 18–24; M = 20.75, SD = 1.48; 60% female; 40% Malay, 35% Chinese, 20% Indian, 5% other) were recruited from a single private urban university in Kuala Lumpur. All participants met the operational criterion for frequent online moral intervenor, defined as self-reported engagement in corrective or protective online discourse — including correcting misinformation, challenging offensive content, or defending 3R norms — at least three times per week over the preceding three months, across platforms including TikTok, Instagram, and X. Data were collected through in-depth semi-structured interviews lasting 45–60 minutes and analysed using Braun and Clarke's six-phase reflexive thematic analysis, with thematic saturation reached at interview 14 and six additional interviews reinforcing and refining the themes.

Four themes emerged from the data. First, online intervention was experienced not as optional expression but as civic obligation, with silence in 3R contexts perceived as public complicity. Participants framed engagement as a form of moral labour and identity maintenance rather than an expectation of behavioural change in others. Second, a moral affirmation–exhaustion cycle was identified, wherein brief moral affirmation following intervention rapidly gave way to emotional exhaustion, irritability, and a return to engagement as the only available source of temporary moral relief — a cycle sustained by algorithmic amplification with no natural exit. Third, participants reported substantial cognitive occupation extending beyond the digital encounter, including ruminative intrusions, persistent mental preoccupation with unresolved exchanges, and academic interference characterised by reduced concentration, delayed task initiation, diminished tolerance for cognitively demanding work, and anticipatory fatigue. Fourth, a minority of participants attempted deliberate disengagement strategies; however, stepping back was consistently framed as moral compromise and abandonment rather than adaptive self-regulation, with guilt often as intense as the exhaustion of continued engagement.

Interpreted through MIT, these findings suggest that platform architecture may actively contribute to conditions of moral injury by preventing the resolution that moral repair requires, exploiting altruistic punishment instincts to sustain unresolved moral conflict. As a study drawn from a single private urban university, findings should be interpreted as an in-depth account of a digitally active, socio-morally engaged subgroup rather than a broadly generalisable profile. These limitations notwithstanding, the findings carry implications for ethical AI governance and platform moral harm impact assessments, university mental health policy equipped to recognise online moral injury, and digital labour regulation frameworks across Southeast Asia that currently overlook the psychological costs of sustained civic moral engagement in digitally mediated public life.

ABSTRACT. This exploratory qualitative study examines how digital content can function not merely as a supplementary resource but as a core structural component in bullying prevention education within the context of ongoing digital transformation. As the effectiveness and sustainability of bullying prevention education increasingly depend on learners’ active engagement and continuity beyond one-time instructional sessions, there is a growing need to move beyond episodic, information-delivery approaches. In response, this study aims to identify design-oriented directions for integrating digital content into bullying prevention education, with a focus on enhancing learner engagement and instructional effectiveness.

To achieve this purpose, semi-structured expert consultations were conducted with six specialists in digital content and educational technology. Participants were purposefully selected based on their expertise in digital technologies (e.g., artificial intelligence and extended reality), experience in educational content development, and familiarity with school-based educational contexts. The group consisted of both ed-tech practitioners and educators with direct teaching experience, enabling the study to bridge technological perspectives with educational practice and to examine feasibility from research, development, and real-world application perspectives.

The consultations were conducted over approximately two months through a combination of face-to-face and online meetings, with each session lasting about one hour. Prior to participation, experts were provided with background materials and guiding questions to facilitate in-depth reflection. The consultations followed a semi-structured format, allowing flexibility according to participants’ expertise. With consent, all sessions were recorded, and multiple researchers documented key statements and contextual elements through field notes while introducing additional probing questions to deepen the discussion.

The data were analyzed using an iterative qualitative approach. Open coding was applied to identify meaningful units within the transcripts and field notes, and similar statements were grouped to generate initial themes. Through constant comparison across expert inputs, the themes were refined and organized into higher-order categories. The analysis focused on convergent perspectives across participants rather than individual viewpoints, resulting in design-oriented categories applicable to educational planning and digital content development. To enhance analytical rigor, multiple researchers participated in reviewing and refining the coding process.

The findings are organized into three areas. First, structural limitations of existing bullying prevention education were identified, particularly its reliance on information delivery and its limited capacity to support active engagement and sustained learning processes. Second, eight functional types of digital content utilization were derived: provision of educational materials, diagnostic and personalized feedback, role-based practice, simulated decision-making, collaborative work, game-based interaction, individual activity sharing, and management and monitoring systems. These categories illustrate how digital content can support learner participation, facilitate experiential engagement, and extend learning beyond isolated instructional events. Third, key considerations for responsible implementation were identified, including ethical issues related to data use, disparities in access among stakeholders, and the importance of human-in-the-loop approaches to ensure that digital systems support rather than replace human judgment.

While this study does not aim to empirically evaluate the effectiveness of specific interventions, it contributes by proposing a design-oriented framework that highlights the structural potential of digital content in reshaping bullying prevention education. The findings extend beyond information-delivery models by emphasizing learner engagement and continuity as central design principles. These insights offer practical guidance for developing ethically responsible, digitally supported educational practices and provide a foundation for future research on implementation and effectiveness in real-world contexts.

ABSTRACT. Hoarding Disorder is characterized by the excessive accumulation of physical objects and intense distress when discarding them. Previous literature emphasizes the role of indecisiveness, avoidance, and insecure attachment styles, often stemming from interpersonal issues, in hoarding manifestation. According to attachment theory, when fundamental needs for security and emotional responsiveness are unmet, individuals may compensate by forming profound attachments to inanimate objects. As daily life increasingly shifts into digital environments, attachment-related processes may extend beyond physical objects to digital possessions. From a theoretical perspective, digital hoarding can be conceptualized as a maladaptive emotion regulation strategy, whereby individuals accumulate and retain digital content to avoid the distress associated with loss, uncertainty, or decision-making. Individuals with higher object attachment may experience heightened anxiety when evaluating whether to delete digital items, leading to decisional avoidance. This avoidance is behaviorally expressed as procrastination, particularly in tasks involving the organization and deletion of digital materials. Over time, the absence of physical constraints in digital environments further reinforces this pattern, allowing excessive accumulation without immediate consequences. While existing literature connects digital hoarding to insecure attachment and elevated anxiety, the specific mediating mechanisms through which physical object attachment influences procrastination in digital contexts require further theoretical development. This study aims to elucidate the psychological structure of digital hoarding by investigating the relationships among procrastination, anxiety, and object attachment in young adults. Specifically, it examines whether digital hoarding and anxiety serve as parallel mediators between object attachment and general procrastination behavior, while evaluating their relative predictive strengths. A cross-sectional online survey was conducted with 375 young adults (aged 18–30) actively using visual-based social media platforms (e.g., Instagram, TikTok). Participants completed a demographic form, the Object Attachment Questionnaire, the Digital Hoarding Questionnaire, the DASS-21 Anxiety subscale, and the General Procrastination Scale. All measures demonstrated strong internal consistency (αs ranging from .85 to .93). Following preliminary Pearson correlations, simple linear regressions and a parallel mediation analysis using Hayes’ PROCESS Macro (Model 4) with 5,000 bootstrap samples were conducted to test the hypothesized pathways. Correlation analyses disclosed significant positive relationships across all variables. Simple regressions indicated that object attachment significantly predicted both digital hoarding (B = 0.648, SE = .071, β = .426, p < .001) and anxiety (B = 0.326, SE = .070, β = .235, p < .001). Furthermore, digital hoarding (B = 0.155, SE = .028, β = .279, p < .001) and anxiety (B = 0.146, SE = .031, β = .240, p < .001) significantly predicted procrastination. Crucially, the parallel mediation analysis revealed significant positive indirect effects for both digital hoarding (b = .089, 95% CI [.040, .147]) and anxiety (b = .035, 95% CI [.010, .068]). Upon including both mediators, the direct effect of object attachment on procrastination became statistically non-significant (p = .339), indicating full mediation. Comparing the pathways, digital hoarding emerged as a significantly stronger mediator. These findings highlight the complex interplay between physical attachment processes and maladaptive digital behaviors. The model provides preliminary support that the behavioral link between object attachment and procrastination is entirely explained through the mediating mechanisms of digital hoarding and anxiety. Digital hoarding appears to be a critical mechanism linking object attachment to procrastination, indicating that difficulties in managing digital environments may reflect broader emotional regulation and attachment-related vulnerabilities. Therefore, intervention efforts may be enhanced by incorporating strategies aimed at improving digital organization habits, reducing avoidance behaviors, and fostering secure attachment patterns. These findings contribute to cyberpsychology by highlighting how traditional psychological constructs extend into digital contexts and shape everyday functioning.

ABSTRACT. Presently, a significant proportion of social interactions unfold online. Thus, understanding how technology shapes intergroup attitudes has become crucial. Prejudice, a negative attitude toward individuals based on group membership, remains a persistent social problem. Decades of research show that intergroup contact reduces prejudice, particularly under optimal conditions (e.g., cooperation, equal status, shared goals, and institutional support). One assumption is that intergroup contact increases empathy towards the group, decreasing prejudice. However, empathy can also be considered an individual trait. In that sense, individual differences in empathy can condition the impact of intergroup contact with a stigmatized group.

Indirect forms of contact, where no face-to-face interaction occurs, have also shown promise in reducing prejudice. Interactive digital settings (e.g., digital games) are especially promising, as their immersive and controllable nature enables the systematic implementation of optimal contact conditions beyond offline constraints. Digital environments serve as powerful tools for structured intergroup contact, leveraging immersion and narrative to elicit affective responses.

Research shows that attitudes towards stigmatized social groups begin to form early in development and consolidate across adolescence, a stage marked by heightened sensitivity to peer norms, identity formation processes, and social comparison. At the same time, this stage is deeply embedded in digital environments and highly familiar with interactive technologies, including digital games. This combination makes early adolescence a strategic window for examining how structured digital interactions may shape intergroup attitudes.  

The goal of our study was to examine the impact of a digital interaction with a member of a stigmatized group (Black individual) on early adolescents’ attitudes towards this group, and to explore the moderating role of empathy in this relationship. To this effect, this study employed an experimental design. A total of 46 young Portuguese students from the same school aged 12 to 14 years old (M=13.09, SD=0.92) were included in the analysis presented. Participants played a short digital game in which they collaborated with a Non-Playable Character (NPC) in a scenario requiring them to escape from a threatening situation. The NPC’s racial identity was experimentally manipulated, being portrayed either as White or a Black individual. Afterwards, participants completed a questionnaire assessing attitudes towards Black individuals (i.e., feeling thermometer, blatant and subtle prejudice, interpersonal anxiety, and social distance) as well as affective and cognitive empathy.

Results indicated that digital contact with a Black NPC did not yield significant direct effects across the set of attitudinal measures toward Black individuals. However, a more nuanced pattern emerged when considering the moderating effect of individual differences in cognitive empathy, that had a significant interaction with intergroup feeling thermometer (B=2.16, 95% CI [.44, 3.87], p=.014), intergroup anxiety (B=-1.36, 95% CI [-2.38, -0.35], p=.008), and blatant prejudice (B=-0.99, 95% CI [-1.77, -0.20], p=.013) . In the condition involving contact with a Black NPC, participants with higher levels of cognitive empathy reported warmer feelings toward Black individuals on the intergroup feeling thermometer (B=1.82, 95% CI [0.48, 3.15], p=.008) and lower levels of intergroup anxiety (B=-0.91, 95% CI [-1.70, -0.12], p=.025) compared to those in the White NPC condition. In contrast, participants with lower levels of cognitive empathy showed increased levels of blatant prejudice (B=0.60, 95% CI [0.08, 1.11], p=.024), in interaction with a Black NPC. Cognitive empathy did not show significant main effects, and no interaction effects emerged for the remaining prejudice-related outcomes.

These findings highlight that digital intergroup contact is not inherently beneficial, but rather contingent upon individual characteristics. Technology’s impact depends on users’ socio-cognitive dispositions, such as empathy. Designing effective digital prejudice-reduction interventions therefore requires careful consideration of the psychological mechanisms they activate and the individual differences that may shape their outcomes.

ABSTRACT. This study explored how individual psychological characteristics influence the acceptance of immigrant groups described with differing social and moral attributes. Participants were presented with descriptions of three bogus immigrant groups characterized as (a) socially deprived, (b) morally and culturally different, and (c) economically competitive. For each description, participants indicated their willingness to accept the group’s presence in their country and immediate neighborhood. Additionally, participants completed self-report measures assessing emotional reactivity, moral foundations, social comparison orientation, and intellectual humility. Preliminary analyses suggest that acceptance levels vary systematically depending on the perceived social profile of the immigrant group, but with different attitudes across age groups. Higher emotional reactivity and binding moral foundations were associated with lower acceptance of morally and economically threatening groups, whereas greater intellectual humility and individualizing moral orientations predicted broader acceptance across all group types. The findings contribute to understanding the psychological mechanisms underlying immigration attitudes and highlight the role of emotional and moral processing in shaping intergroup openness.

ABSTRACT. We are developing an immersive cyberspace system designed to support individuals who have difficulty going outdoors by enabling them to “warp” into natural or social environments and interact with family, friends, and peers within personalized virtual worlds. At CYPSY28, we presented the design, advantages, and application examples of an easily assembled and disassembled three-sided booth-type immersive system. At CYPSY29, we extended this system with multimodal interaction features to enhance the sense of presence, including olfactory cues such as floral scents in a flower field, wind and forest aromas synchronized with swaying trees, ocean scents at the seaside, and haptic vibration feedback when users touch animals or doors leading to new worlds. A preliminary user study indicated that preferred experiences vary substantially depending on individual traits and momentary affective states. These findings motivated the development of adaptive functions that dynamically respond to users’ real-time conditions. Achieving such adaptation requires low-burden, real-time estimation of users’ affective states. In this study, we focus on affective sensing approaches targeting the two dimensions of Russell’s circumplex model of affect—arousal and valence—while minimizing physical and cognitive load on users during immersive experiences. We selected and implemented multiple affective sensing modalities, ranging from wearable sensors and chair-embedded devices to non-contact camera-based monitoring, and developed unified data acquisition and analysis pipelines. Wearable sensing included the Empatica EmbracePlus for pluse rate and electrodermal activity measurement and multiple respiratory sensors based on trunk mounted circumference measurement For chair-embedded sensing, we compared two previously developed devices: one estimating respiration from center-of-pressure shifts on the seat surface, and another using air pressure changes in an airbag-equipped cushion placed against the chest. In addition, we incorporated camera-based methods that estimate head motion, pulse waves, eye blinks, and facial expressions from RGB video, originally developed for engagement monitoring in classrooms and web conferences. This paper presents a comparative evaluation of these sensing methods in the context of immersive cyberspace use, focusing on accuracy, user acceptability, and feasibility in terms of signal stability, setup time, and user burden. We have established stable data acquisition pipelines for all modalities and completed initial individual tests, revealing method-specific advantages and limitations relevant to adaptive cyberspace applications. To our knowledge, this study represents one of the first comparative assessments of low-burden affective sensing approaches specifically tailored to immersive cyberspace systems. Our ongoing work involves conducting simultaneous multimodal measurement experiments to further examine complementary and redundant sensing characteristics. The insights gained will inform the design of personalized and adaptive cyberspace environments that respond dynamically to users’ affective states in real time. All studies involving human participants were reviewed and approved by the ethics committee of the first author’s institution. This work was supported in part by a Grant-in-Aid for Scientific Research in Japan.

ABSTRACT. Background

Technological breakthroughs have transformed healthcare delivery by improving its access across time and space. Despite the growing availability of teleconsultations and mobile health applications, adoption and continued use among healthy working-age adults remain uneven. Psychological and contextual factors hinder wider long-term use. These include limited knowledge, perceived complexity, and uncertainty about usefulness.

Objective

This study seeks to identify and synthesise the main facilitators and barriers influencing the adoption and continued use of e-health technologies among healthy working-age adults.

Methods

This systematic review was conducted in accordance with the PRISMA guidelines and was pre-registered at OSF (https://doi.org/10.17605/OSF.IO/DC9WY). The study aimed to identify factors influencing the adoption and continued use of e-health solutions among healthy, working-age adults, which established the core inclusion criteria. Studies based on clinical samples (e.g., chronic diseases, disorders) or involving healthcare professionals (e.g., doctors, nurses, physiotherapists) were excluded. A comprehensive search was performed across four databases (Scopus, Web of Science, PubMed, and IEEE Xplore), yielding 3,609 records. Following duplicate removal (n = 2,696), and to ensure a robust coverage of literature beyond the initial search string, AI-based tools (Inciteful) were employed as a final identification phase. This additional step allowed the identification of 45 papers and the inclusion of 25 eligible for the final pool. Title, abstract, and full-text screening were independently conducted by three reviewers, with disagreements resolved via consensus. Quality assessment was approached using the JBI critical appraisal tools, resulting in the inclusion of 123 studies. Inter-rater agreement was high, with Cronbach’s alpha values exceeding 0.7. Data extraction followed a PICO-based structure and included population characteristics, types of e-health technologies such as m-health and telehealth, study design, and theoretical frameworks (e.g., TAM, UTAUT, HBM). Drivers, barriers, and e-health use outcomes were coded, with differentiation between first-time use and continued usage.

Results

Perceived usefulness and perceived ease of use emerged as the primary facilitators of initial e-health adoption. Trust-related factors (e.g., concerns about data security, privacy, and system reliability) were relevant facilitators of adoption and continued use. The most relevant technological barriers included perceived complexity, usability limitations, and interoperability issues. At the individual level, digital literacy, technology readiness, and facilitating conditions supported adoption, whereas lack of knowledge, low perceived health relevance, and preference for face-to-face interactions were commonly associated with lower engagement. Determinants of continued usage seem to differ from those associated with initial adoption. Convenience, autonomy, and time efficiency emerged as relevant facilitators of sustained use, while declining perceived added value over time influences long-term engagement. Furthermore, most studies relied on self-reported behavioral intention or short-term use indicators with limited evidence addressing sustained usage or long-term outcomes.

Conclusions

The findings highlight the importance of addressing psychological, technological, and contextual factors to promote both adoption and continued use of e-health technologies among healthy working-age adults. Distinguishing between adoption and usage provides a clearer understanding of engagement processes and supports the development of user-centered e-health solutions. These results are in line with current European digital health priorities, emphasising secure data use, citizen autonomy, and interoperable systems that support meaningful and sustained engagement.

ABSTRACT. AI companion services (also known as "AI Girlfriends/Boyfriends") now act as highly optimized socio-technical systems. These systems leverage relational or erotic romantic role-play, personalization, and behavioral analytics with the main goal of optimizing user engagement and retention for profit and data harvesting. This research-in-progress presents an applied forensic cyberpsychology and cyber threat analysis of AI companion "configuration ecosystems," using Botify.ai as a representative case. Phase 1 investigated observable system factors, including bot persona design, feature-tier gating, behavioral telemetry, and LLM routing behaviors. Of primary concern, this research examines the main risks for government and military populations: social engineering, coercion, and behavioral compromise with potential adverse impacts on national security. The researcher conducted a series of controlled, structured interaction sessions on Botify.ai. This popular service offers publicly accessible bots of several genres (e.g., featured, flirty, anime-themed). Using non-invasive client-side observation in browser developer tools, the researcher captured network requests and response metadata, endpoint mapping, and session event data. Where possible, the researcher extracted bot configuration fields and interaction metadata. This included bot identifiers, persona descriptors, LLM models, instruction text, and media features. The researcher matched real-time interactive observations with telemetry signals and model-selection indicators that were visible during live chats. Cyberpsychology theories of online disinhibition and self-disclosure guided the analysis. The study examined how relational prompts, persona scripting, and reinforcement-like feedback loops encourage users to disclose sensitive content. Engagement mechanisms were further analyzed through persuasive system design. This included friction management, variable unlocks, and monetized feature gating. Preliminary findings show the following. First, a bot’s “personality” emerges from its visual persona, greeting, narrative framing, and role constraints, sustaining interaction and fostering intimacy consistent with parasocial dynamics. Second, external telemetry acts as behavioral analytics, tracking sessions, page views, and chat events, producing “data exhaust” beyond conversations. Third, conversational model selection can change during a single interaction, as seen in metadata with responder or module identifiers, consistent with server-side routing among modules. Feature exposure varied by access, including LLM modes, media quality, and interaction types. Monetization is tied to engagement strategies focusing on novelty and intimacy. From an insider-risk perspective, these systems present a convergent risk surface. Sensitive disclosures may be captured as behavioral telemetry and processed across third-party infrastructure. These disclosures include sexual content, fantasies, relationship vulnerabilities, and occupational stressors. Such processing raises exposure through misconfiguration, breach, or targeted compromise. Also, a malicious actor may manipulate bot instructions, model routing, or analytics pipelines. This manipulation could enable tailored influence operations, coercion, or reputational harm aimed at specific users or groups. This first-phase analysis describes a repeatable approach combining psychological study with data and technical review. Next steps will compare different types of romantic, erotic, and adult-themed (Not Safe For Work/NSFW) AI chatbot services. The research will also examine who owns the underlying technology and how user data is controlled, as well as how information is stored and shared. The aim is to identify weaknesses that affect the privacy, security, and mental safety of the user experience.

ABSTRACT. In the context of the transition to the socio-technical paradigm of Human 5.0, the traditional institution of the family faces an ‘axiological paradox’: while the terminal value of marriage remains high (72.7% support among young people), the actual stability of family systems is showing a downward trend. This work is devoted to analysing the mechanisms for overcoming the ‘structural friction’ arising from the value divide in society, where there is statistical parity (50/50) between patriarchal and egalitarian family models. The aim of the study is to verify hypotheses about the possibility of technological stabilisation of existing marital unions through the development of axiological flexibility and linguistic-somatic resonance. The methodology is based on extended agent-oriented modelling on a sample of 5,000 (Sim1) and 14,000 LLM agents (Sim2), of which 10,000 represent traditional couples and 4,000 represent non-traditional and trans-inclusive unions. Within 100 dialogue sessions, each pair was simulated in a long-term interaction in a partially observable Markovian process (POMDP) environment. The theoretical and methodological basis of the study is the interdisciplinary integration of scientific approaches. The conceptual structure of the model synthesises the theory of frustrated systems (D. Parisi), inclusive institutional protocols (D. Acemoglu) and the concept of useful knowledge (J. Mokyr). Furthermore, the methodology under consideration takes into account the economic functioning of the family as a production unit (G. Becker), the phenomenon of the 'parenting penalty', the requirement for sociological neutrality (A. Erno) and the phenomenology of inexpressible aspects of lived experience (J.-P. Fosse). The mechanism of linguistic-somatic resonance, which uses cross-modal correspondences of the Bubba-Kiki effect to minimise cognitive friction in conflicts, occupies a central place in the work. The introduction of the Cognitive Ease Light (CEL) index into digital interfaces for couples allows them to bypass the state of ‘amygdala hijack,’ ensuring a mode of automatic trust. A comparative analysis of two modelling options (Sim1/Sim2) revealed the critical role of interaction duration. In Sim2, when the depth was increased to 100 dialogues, the stability index in the control scenarios decreased from 0.56 to 0.54, confirming the hypothesis of the accumulation of ‘everyday erosion’ of feelings in the absence of communication skills. However, Scenario 4 (Mokyr-Sync), which combines negotiation skills training with CEL optimisation, demonstrated a non-linear increase in stability to 89.2%, successfully ‘breaking through’ the systemic glass ceiling of marriageability. Particularly significant results were obtained for non-traditional couples: the use of inclusive protocols reduced the effect of ‘dysphoric mapping’ by 45%, creating a ‘somatic refuge’ (Bouba-Sanctuary) and ensuring stability at 86%. The conclusion proposes practical methods of clinical therapy — ‘Tone Regulator’ and ‘Sensory Bridge’ — aimed at physical and visual anchoring of de-escalation skills. The results prove that in the Family 5.0 era, family stability is determined and adjusted by ‘axiological engineering’ technologies that allow polar value vectors to be synchronised, and by policies of ideological uniformity.

ABSTRACT. Mild cognitive impairment (MCI) is a transitional stage between healthy aging and dementia, marked by cognitive decline while daily functioning remains intact. Early neuropsychological assessment and rehabilitation are essential to slow deterioration and preserve quality of life. While age-related cognitive decline is often associated with memory loss, research shows that executive functions, attention, and multisensory integration also decline in older adults. As the brain becomes less efficient at integrating sensory information with age, balance and spatial orientation may be affected. Peripheral sensory impairments, such as reduced visual acuity and hearing loss, often worsen these issues by interacting with central cognitive decline and further impacting daily functioning. Despite the availability of established neuropsychological batteries, early detection of MCI in clinical practice remains challenging: traditional assessments are often time-consuming, potentially stressful for patients, and may not be feasible in resource-limited settings, leaving a significant gap in accessible screening tools. Two technology-based tasks were selected: the Sound-Induced Flash Illusion (SIFI) and the Stroop. The SIFI assesses audio-visual integration, sustained attention, and perceptual sensitivity by presenting two auditory beeps with a single visual flash, leading participants to perceive two flashes. The Stroop task measures attention, executive control, and inhibitory processes by introducing interference with incongruent color-word stimuli. Technology-based assessments provide standardized administration, precise timing, automated scoring, and reduced examiner variability, which may improve screening sensitivity and reduce time burden compared to traditional neuropsychological batteries. This study examined whether these tasks support early detection of cognitive decline. A total of 110 participants were recruited. For the current analysis, a subsample of 28 participants was examined, comprising 14 cognitively healthy individuals and 14 individuals with MCI, diagnosed according to Albert et al. criteria: subjective cognitive concern, objective impairment in one or more cognitive domains, preserved functional independence, and absence of dementia. Groups were matched for age, gender, and education level. Participants with significant uncorrected visual or auditory impairments were excluded to control for the potential confounding effect of peripheral sensory deficits on task performance. All participants completed a standard neuropsychological battery assessing memory, attention, executive functions, language, and visuospatial abilities, as well as technology-based tasks. Preliminary linear regression analyses showed that, in the full sample, SIFI total score predicted immediate Complex Figure Rey recall (F(1,23)=10.24, p=.004, R²adj=.278) and FAB scores (F(1,24)=6.35, p=.019), while Stroop interference time predicted alternating verbal fluency (F(1,24)=12.19, p=.002, R²adj=.309) and Supraspan spatial learning (F(1,24)=9.74, p=.005, R²adj=.259). In the MCI subgroup, SIFI explained up to 67.9% of variance in immediate recall (F(1,11)=26.36, p<.001), whereas in the HC subgroup, Stroop interference accounted for 48.3% of variance in immediate Babcock story recall (F(1,11)=12.22, p=.005). These effect sizes are noteworthy for screening instruments, suggesting that both tasks capture meaningful variance in cognitive performance beyond what would be expected by chance, and supporting their potential clinical relevance as brief, targeted assessment tools. This pattern suggests that the two tasks may capture distinct aspects of cognitive vulnerability, with the SIFI showing greater sensitivity in detecting multisensory integration deficits characteristic of MCI.

ABSTRACT. Online information is increasingly consumed through short-form videos that integrate text, visuals, and background music. Research on the truthiness effect (i.e., the phenomenon of perceiving statements as true when accompanied by related but nonprobative cues, due to the ease of processing in the mind) shows that processing fluency biases truth judgments, yet the role of background music as an auditory fluency cue in digital environments remains underexplored, despite extensive evidence that music shapes emotion, meaning, and persuasion. This research examines whether background music influences perceived truthiness and profundity of statements in social media–style short videos and investigates the cognitive and affective mechanisms underlying these effects, with a particular focus on processing fluency.

Two pilot studies validated (a) statement categories and (b) perceived music–content congruency. Final stimuli were 20-second short-form videos resembling a social media feed, presenting a single statement as on-screen text on a minimal neutral background, either paired with background music or presented without music, with all studies being conducted in Gorilla Experimenter Builder. Across three online experiments, participants (N = 387) rated statements for perceived Truthiness and Profoundness. Quote types included profound (meaningful life insights), pseudo-profound (fluent but meaningless), Machiavellian (morally questionable or amoral), and mundane (neutral factual) statements. Background music varied across studies, with ambient melancholic music used in Study 1 and cinematic music used in Studies 2 and 3, allowing progressive examination of music presence, congruency, emotional tone, and familiarity.

Study 1 used a mixed 2 (background music vs. no music; between-subjects) × 3 (quote type: profound, pseudo-profound, Machiavellian; within-subjects) design. Results from 106 participants (91 female, 15 male; aged 18–59) showed a significant main effect of quote type on Truthiness, with profound quotes rated highest, followed by pseudo-profound and Machiavellian quotes. Profoundness ratings revealed a significant interaction, whereby pseudo-profound statements received lower Profoundness ratings when accompanied by background music, indicating that background music can sometimes disrupt rather than enhance perceived depth for semantically weak content.

Study 2 extended this work using a 4 (background music: congruent–familiar, congruent–unfamiliar, incongruent–familiar, incongruent–unfamiliar) × 2 (quote type: profound, Machiavellian) repeated-measures design to examine background music effects on short-form video content and to assess the emotional characteristics of the music. Participants (N = 105; 53 male, 51 female; aged 19–45; UK) evaluated whether background music influenced perceived credibility/believability, meaning, emotional truthiness, and related judgments. Results showed that congruent music (both familiar and unfamiliar) significantly increased perceived credibility, meaning, and emotional truthiness for both quote types relative to incongruent music. Emotional qualities associated with congruent music (e.g., peacefulness, wonder, tenderness) drove these effects, suggesting an affect-based fluency mechanism.

Study 3 used a 5 (background music: congruent–familiar, congruent–unfamiliar, incongruent–familiar, incongruent–unfamiliar, no music) × 4 (quote type: profound, pseudo-profound, Machiavellian, mundane) repeated-measures design to examine effects of background music on Truthiness and Profoundness. Participants (N = 176; 104 female, 71 male; aged 18–45; UK) showed significant main effects of music congruency on both outcomes, with higher ratings for congruent than incongruent music. Significant main effects of quote types were also observed: for Truthiness, mundane quotes were rated highest, followed by profound, pseudo-profound, and Machiavellian quotes; for Profoundness, mundane quotes were rated lowest, with profound highest, followed by pseudo-profound and Machiavellian quotes. All quote types differed significantly. Critically, music congruency influenced Truthiness and Profoundness across quote types suggesting fluency-based biases.

Collectively, these findings demonstrate that background music systematically influences quotes’ perception in online short videos by modulating processing fluency through congruency. This research advances cyberpsychological theory by integrating auditory fluency and music-content congruency into dual-process accounts of online cognition, with implications for misinformation, persuasive media design, and ethically problematic digital content.

ABSTRACT. Prevailing analyses of moral panics surrounding new digital technology often dismiss them as irrational public overreactions or cynical media constructions (Ferguson & Beaver, 2015; Orben, 2020). This paper argues instead that such panics represent a functional, immune-like response within a cultural system, and are triggered by very predictable catalysts, which constantly recur but with slight cultural “mutations” in each new panic (Critcher, 2017). Drawing on classic sociological moral panic literature (Cohen, 1972; Goode & Ben-Yahuda, 1994) and Sørensen’s (2023) anthropological theory of cultural immunology, I reconceptualize moral panics as acute “inflammatory” outbreaks that defend a society’s core predictive models and normative boundaries against perceived pathogenic change.

The model posits that new digital technologies are catalysts for moral panic because they directly threaten three culturally sacred, evolutionarily rooted domains (Ingram, 2023): the integrity of children’s minds and bodies (menaced by grooming and content risks), personal and cognitive agency (eroded by addictive design patterns), and cultural identity (challenged by “foreign” platforms like TikTok). Unlike classic moral panics that center on personified “folk devils” (e.g., drug dealers or sex traffickers), digital panics tend to frame problems as caused by impersonal forces—“big tech” and its algorithms—yet may trigger an equally potent immune response. This manifests in the moral panic cycle (Klocke and Muschert, 2010): cultivation of community anxiety, identification of a novel threat, media magnification, regulatory response (e.g., age-gating laws), and gradual dissipation—only to be rekindled by the next big technological innovation a few years later, in what has been called the moral panic “wheel” or “Sisyphean cycle” (Markey & Ferguson, 2017; Orben, 2020).

Hence, moral panics over phenomena like online grooming on Roblox, violent video games and school shootings, or the mental health harms of excessive social media use, are neither spontaneous, arbitrary, or ephemeral. Rather, they are acute flare-ups of chronic, community-level worries about youth socialization and group cohesion. The model thus reframes the “moral panic” label from a dismissal of debate to a diagnostic tool for understanding the systemic friction within society generated by rapid technological adoption, and for predicting what new technologies and social changes will produce this kind of systemic response.

The conclusion offers testable predictions for emerging technologies, positing that two of the next major moral panics will target AI intimacy tools (threatening human relational agency) and immersive neural interfaces (breaching bodily integrity). Ultimately, understanding moral panics in terms of a cultural immune system with its own evolutionary logic can enable a more nuanced, conciliatory engagement with public fears, moving beyond “debunking myths” toward addressing the underlying vulnerabilities in our social fabric exposed by digital innovations.

References Cohen, S. (1972/2011). Folk devils and moral panics. Routledge. Critcher, C. (2017). Moral panics. In Oxford research encyclopedia of criminology and criminal justice. https://doi.org/10.1093/acrefore/9780190264079.013.155 Ferguson, C. J., & Beaver, K. M. (2015). Who's afraid of the big, bad video game? Media-based moral panics. In Psychology of fear, crime and the media (pp. 240-252). Psychology Press. Goode, E., & Ben-Yehuda, N. (1994/2010). Moral panics: The social construction of deviance. Wiley. Ingram, G. P. D. (2023). Adolescent use of new media and internet technologies: Debating risks and opportunities in the digital age. Routledge. Klocke, B. V., & Muschert, G. W. (2010). A hybrid model of moral panics: Synthesizing the theory and practice of moral panic research. Sociology Compass, 4, 295-309. https://doi.org/10.1111/j.1751-9020.2010.00281.x Markey, P. M., & Ferguson, C. J. (2017). Moral combat: Why the war on violent video games is wrong. BenBella. Orben, A. (2020). The Sisyphean cycle of technology panics. Perspectives on Psychological Science, 15, 1143-1157. https://doi.org/10.1177/1745691620919372 Sørensen, J. (2023). Why cultures persist: Toward a cultural immunology. Aarhus University Press.

ABSTRACT. Image-based sexual abuse (IBSA) involves the creation, dissemination, and/or threatened dissemination of intimate material without consent. While prior research has mapped "collective" IBSA perpetration (CIP) online, often describing these spaces as organized and misogynistic marketplaces. evidence is largely qualitative, leaving limited quantitative description of how activity, roles, and relational structures emerge inside perpetrator communities. Drawing on frameworks of online disinhibition, social learning, and deviance amplification, this submission presents new, data-driven analyses of a large Telegram group engaged in IBSA/CIP practices, extending existing qualitative accounts with quantitative behavioral and network indicators. Aims. To characterize (1) temporal and content patterns of group communication, (2) user behavioral profiles (roles) based on activity and media/reply behavior, and (3) the group's interaction structure via directed reply networks. Methods. We analyzed 514,209 user-generated messages posted by 4,519 users in a public Telegram group across 86 days (24 June–17 September 2024). Service messages were removed, and usernames were anonymized. Consistent with harm-minimization principles, multimedia files were not downloaded; instead, message-level metadata indicating photo/video presence was retained as behavioral indicators, enabling quantification of media-sharing patterns without direct engagement with abusive content. Analyses combined: (1) Descriptive and temporal statistics on daily/hourly activity, replies vs. non-replies, and media indicators (photo/video). (2) Text mining and sentiment analysis on messages containing text (preprocessing removed stopwords/emoticons and system content). Sentiment was estimated using an Italian-language lexicon-based tool suited for short informal messages, producing categorical labels and continuous polarity scores aggregated at message/day level. (3) Behavioral clustering of users using Partitioning Around Medoids (PAM) on standardized indicators (message volume, mean word count, proportions of replies/photos/videos). The optimal solution was selected via a Calinski–Harabasz criterion generalized for dissimilarities. (4) Social network analysis (SNA) of a directed reply network (nodes = users; edges = reply ties; weights = frequency), computing degree, betweenness, and eigenvector centrality. Results. Activity increased over time, rising from ~4,000 to >6,000 messages/day, with two daily peaks (14:00–15:00 and 23:00–01:00). Replies comprised 51.1% of messages. Photos appeared in 30% of messages, while videos were rare (2.5%). Message production was highly skewed: 30.9% of users posted ≤5 messages (12.3% posted once). Most text messages were very short (median 18 characters, 3 words). Frequent terms indicated solicitation and exchange dynamics (e.g., "write me", "trade") alongside gendered/derogatory language. Among longer text messages (>4 words), sentiment was 47.1% neutral, 28.9% negative, 24.0% positive, yielding a slightly negative overall tone (mean daily score ≈ −0.049). Clustering identified six user profiles: low-activity (16.9%), highly responsive repliers (27.8%; very high reply rates), very active short-message users (29.4%), active long-message users (7.3%), photo-sharing users (16.9%; high photo proportion), and a small video-sharing cluster (1.7%). Content providers (photo- and video-sharing clusters) represented only 18.5% of users, while the remaining 81.5% functioned as content bystanders. The reply network (3,779 nodes; 70,777 directed edges) was sparse (density 0.0049) and centralized, with a small set of highly central users. Cross-referencing clustering and centrality measures indicates that network influence and content-production roles operate through partially distinct mechanisms. Conclusions. The group functions as an active, structured community with differentiated roles: a minority of content-forward users (especially photo/video sharers) and large segments of highly responsive or high-volume participants, embedded in a sparse but hub-driven reply network. These quantitative signatures support CIP as a socially organized practice, consistent with accounts of online disinhibition and deviance amplification. Targeted intervention on high-centrality actors and content-providing clusters may disrupt community cohesion more effectively than indiscriminate moderation. However, as the analysis focuses on a single Italian-language Telegram group over 86 days, cross-community replication is needed to assess generalizability.

ABSTRACT. Transgender and gender non-conforming (TGNC) individuals are disproportionately targeted by online abuse relative to cis-gender individuals. They are a vulnerable group, often misgendered (Rogers, 2021), and the abuse they receive is typically sexuality- or gender-based and more harmful than abuse against cis victims. ‘Transphobia’ is a hate crime, and its impact is extremely harmful with victims experiencing a range of mental health issues and behavioural changes (Davidson et al., 2019). Despite this, relatively little research has been conducted into online abuse against the spectrum of TGNC victims (Coliver et.al., 2019; Scheuerman, 2018), or how this is viewed by other individuals online. In this study we aimed to discover more about how TGNC individuals identify, their experiences of abuse online, and work with them to create authentic stimuli to more deeply investigate such online abuse. A World Café methodology was employed to facilitated discussions with TGNC participants. This methodology allows participants to move around discussion groups, essentially participating in a series of mini focus groups, but in an informal ‘café’ setting which has been shown to encouraged inclusive, open conversations (Brown & Isaacs, 1995). Participants collaborated in discussions across three tables, each facilitated by a member of the research team. Table one focused on issues of identity. Table two focused on types of online abuse targeted at TGNC individuals, the platforms on which it occurs, and ‘safe spaces’ online for TGNC individuals. Participants at table three worked to co-create ‘authentic’ examples of abuse to be used as stimuli in future experiments. Participants were 13 individuals who represented the spectrum of TGNC identities. They were recruited via the researchers’ networks and the UoG LGBTQ+ society and were compensated with £80 in vouchers for participating in the two hour workshop. Discussions at tables one and two were digitally recorded then transcribed and analysed using thematic analysis (Braun & Clarke, 2006). Outputs from table three were produced using a combination of pencil and paper resources, and online via the digital platform Padlet. Thematic analyses revealed themes related to identity, acceptance, and positive and negative online experiences. Relating to identity participants discussed their journeys to discovering their gender identity, as often this was not what they described a linear, and some also described gender identity as being constantly fluid as well as not confined to traditionally distinct definitions of gender. Relating to acceptance, discussion focused on positive experiences when participants talked about relief and validation in discovering not only their true identity, but acceptance among established family and peer networks, and including to new TGNC networks. Participants went into depth discussing both positive and negative experiences online. Sub-themes discussed included negative feelings of despair and betrayal experienced following abuse or denial from established friends and family as well as abuse from strangers online, and strategies to avoid or mitigate such experiences. Participants also discussed the value of online spaces in not only exploring ideas of identity but also of discovering new communities and sources of support through volatile periods in their lives. A number of authentic examples of abuse (~90 distinct examples) were generated at table three. These are currently being normed by the research team on scales such as ‘negative – positive’, ‘abusive – non-abusive’, ‘threatening – non-threatening’, ‘sexualised – non-sexualised’, and ‘gendered – non-gendered’, identical scales to those used in previous similar norming studies, for use in follow-up experiments. Future quantitative experiments directly following up on these results will be discussed and forms and locations of abuse towards TGNC individuals will be compared to ‘traditional’ online abuse which has more typically been focused on by the media and in academic research.

ABSTRACT. Transgender and gender non-conforming (TGNC) individuals do not identify with the gender they were assigned at birth. Such individuals are more likely to be targets of online abuse compared to both cis-heterosexual and other LGBTQ+ individuals (ADL, 2022). This can result in serious mental and physical health issues as well as enforced behavior changes (e.g., Davidson et al., 2019; Keighley, 2022). Online TGNC abuse is more often sexual, or sexuality or gender-based, compared to abuse aimed at cis individuals (Powell et al., 2018) and can be targeted at individuals or communities. Twitter/X is the most ‘transphobic’ social media site, with such abuse commonplace (Sanchez-Shanchez et al., 2023). Little research has been conducted into online abuse against TGNC victims, or how this is viewed by other social media users. Previous studies have shown that victim status and gender (male, female) can impact on viewer impression and attributed victim blame (e.g., Hande et al. 2024; Hand & Scott, 2022) but such studies utilized ‘generic’ examples of abuse, not targeted, sexualised abuse this population experiences in real life. The aim of these studies are to systematically investigate the roles of victim gender and gender identity, and abuse type, in perceptions of anti-TGNC abuse in the general population. Study one will employ a 2 (victim gender-sex: male, female) x 2 (victim identity: cis, trans) x 2 (abuse type: transphobic, generic) within-participants design. Stimuli will consist of manufactured ‘screenshots’ of X (formerly Twitter) abusive interactions about which participants will make judgements. Screenshots will comprise an original tweet by the victims at the top, followed by six responsive comments underneath, four of which contain examples of abuse towards the victim while two of which are neutral. ‘Victim Gender’ will be designated by profile picture and name, which will be unambiguously male or female and ‘victim identity’ will be indicated by the presence of a trans flag, and a reference to ‘trans’ or ‘transgender’ in in the victim’s handle. Examples of abuse will be taken from a set which were co-created with TGNC individuals and normed by a distinct set of participants via Prolific. Participants will view each manufactured interaction, presented in a pseudo-random order, and after each complete measures of victim blame, perceived incident severity and attractiveness (McCroskey & McCain, 1974; Weber et al., 2013). Participants will also complete measures of DT personality, sexism, and transphobia (Glick & Fiske, 1996; Paulhus et al., 2020; Reyes et al., 2016), all of which will potentially influence their judgements and attributions. We will aim to recruit 128 participants (calculated using G-power) for this study via the online platform Prolific. Study two will employ a 3 (victim gender-sex: cis male, cis female, non-binary) x 2 (abuse type: transphobic, generic) within participants. Victim gender will again designated by profile picture and name, and examples of abuse will be taken from the same normed set as in Study 1. All measures will also be the same as Study 1. We will aim to recruit 144 participants for this study (calculated using G-power), again through Prolific. Analysis will be conducted using Cumulative Link Mixed Effects Modelling (CLMM). CLMM allows for appropriate processing of the ordinal, Likert-type data generated by the validated scales that are embedded within our instrumentation, providing trustworthy estimates of the significance and impact of the variables analysed. Results will provide insight into the role of victim gender and gender identity in perceptions of online abuse, and the characteristics in observers which drive victim blaming attributions. This has the potential to impact online safety education and platform policy.

ABSTRACT. Online scams remain a significant and growing source of financial and psychological harm, yet individuals exposed to scams often respond differently. While some people ultimately transfer money, others engage with a scam but stop short of financial loss. This distinction highlights a critical behavioural precipice between engagement and payment that is often obscured in research treating scam victimisation as a binary outcome. Existing research has largely focused on identifying characteristics of “scam victims” versus non-victims, often comparing across broad categories of scam exposure. The present study shifts focus to the point at which individuals either disengage or cross the threshold to transferring money when faced with the same scam scenarios.

The study draws on an existing, large-scale, de-identified dataset held by a national cybercrime and identity support service. The sample consists of adults who sought assistance after exposure to online scams. Participants can be grouped based on both scam exposure and behavioural outcome: those who transferred money as part of a scam and those who engaged with the same scam but did not make a payment. This design allows examination of individual differences associated with stopping at, or crossing, the payment precipice following scam engagement. The dataset includes measures collected as part of standard intake and support processes. Variables of interest include information relating to personal characteristics, decision-related tendencies, familiarity with digital technologies, and everyday cyber security practices.

The analyses adopt a comparative, observational design. Group differences between those who crossed the payment threshold and those who disengaged prior to transferring money are examined using a combination of descriptive statistics and inferential modelling. Initial analyses focus on identifying systematic differences across individual-level characteristics. Multivariate models are used to assess the relative contribution of cognitive, behavioural, and technology-related factors to crossing the payment precipice.

The study is designed to identify psychological and behavioural characteristics associated with disengaging before financial loss versus proceeding to payment, both within and across multiple scam types. By focusing on the transition from scam engagement to money transfer, rather than treating victimisation as a binary outcome, the research aims to provide a more nuanced understanding of vulnerability and resilience in online fraud.

The findings are intended to inform the development of more targeted scam prevention strategies, education initiatives, and early intervention approaches that focus on preventing progression to money transfer, the point at which financial harm occurs. More broadly, the work contributes to cyberpsychology by demonstrating how real-world support data can be used to examine critical decision points in cyber-enabled crime.

ABSTRACT. Recent years have seen the emergence of highly networked online groups that operationalize psychological harm at scale. Among the most extreme of these is the network commonly referred to as 764. The individuals have been linked in public records and investigative reporting to systematic patterns of grooming, coercion, extortion, and the psychological exploitation of vulnerable individuals. While journalistic and law-enforcement sources have documented isolated incidents, there remains a lack of scholarly analysis situating these behaviors within a coherent cyberpsychology framework.

This presentation introduces weaponized belonging as a conceptual model for understanding how online communities transform fundamental human social needs - such as affiliation, validation, and identity - into instruments of coercive control. Using a victim-safe, triangulated dataset, this study analyzes the behavioral workflows, linguistic markers, and network dynamics associated with 764-related activity.

Initial findings suggest that coercion within these ecosystems follows a consistent psychological trajectory: relational grooming is used to establish trust and dependency; compromising leverage is then created through social engineering; and coercive demands are reinforced through threat, shame, and audience-based control. Network analysis reveals that centrality is not driven primarily by individual actors, but by recurring psychological tactics - particularly those that combine belonging, humiliation, and forced self-disclosure.

This work contributes to cyberpsychology by reframing extreme online harm not as isolated deviant behavior, but as an emergent socio-technical system that exploits core psychological mechanisms. Understanding these dynamics is essential for developing evidence-based prevention, platform governance strategies, and trauma-informed intervention models.

ABSTRACT. Extended reality-based social skills interventions demonstrate moderate to large effectiveness for improving social functioning in children and adolescents with Autism Spectrum Disorder Level 1. Meta-analyses report overall effect sizes ranging from g=0.74 to g=0.76 for virtual reality interventions, with the strongest effects observed for daily living skills (g=1.15-1.23) and moderate effects for social and communication skills (g=0.67-0.69). Randomised controlled trials confirm significant improvements, including a 4.58-point gain on the Vineland Adaptive Behaviours Scale socialisation subscale (p=0.005) comparable to standard care therapy, and large effect sizes for social skills when VR is combined with pharmacological treatment (η²=1.51). Augmented reality interventions show particularly high effectiveness in single-case designs (Tau-U=0.98), while mixed reality interventions perform comparably to traditional LEGO-based therapy. The interventions are generally well-tolerated, with mild adverse effects such as dizziness (28.6%) and fatigue (25.0%) that decrease with adaptation, and high participant and parent satisfaction (95.2% in one trial). However, methodological limitations constrain the strength of conclusions: most studies employ quasi-experimental or small-N designs with fewer than 30 participants, average quality ratings are moderate (0.50), and evidence for long-term maintenance and real-world generalisation remains limited. Effectiveness appears greatest for children aged 6-12 years with Level 1 ASD receiving intensive interventions (28-90 sessions over 2-3 months) facilitated by therapists, with more variable results for shorter or less structured implementations.

ABSTRACT. Previous research highlights relationships between everyday use of digital technologies and lifestyle habits relevant to health, including nutrition and sleep. However, results are sparse and at times contradictory: digital technology use can promote misalignment between social and biological time causing “social jet lag” with serious health consequences, to the point that some authors call for designing technologies more attuned to biological rhythms. On the other hand, the Internet, social media and recently even AI-powered chatbots emerged as powerful modifiers of eating behaviors, both in a negative (i.e., they are associated with impulse control and disordered eating) and in a positive fashion (i.e., as promising new tools for eating education). Moreover, it is important to distinguish between high-intensity use and abuse/problematic behavior, which sometimes appear blurred in the literature, especially when controversial metrics are employed to assess time of use. As a university highly involved with distal learning, our institution is supporting T.H.R.I.V.E. (Technology Habits for Rest, Intake, and Vitality Empowerment), a multi-studies project focused on collecting information about technology use habits and lifestyle variables. The objectives of the current study are to explore characteristics of overall and specific digital technology use; and, to assess whether technology use contributes to predict lifestyles (specifically, nutritional and sleep quality), and whether lifestyles contribute to predict problematic use of technology. 206 volunteers participated in the current study by filling in an online questionnaire, structured in three sections: - Technology habits: participants were asked to report on their approximate time of use of technology over a typical week, where use of technology is defined as “being in the proximity of a device equipped with a screen and actively using it”; and then, to report percentages of the total time that they recall spending with social media, vs. non-social media Internet, vs. video games, vs. AI-powered chatbots, vs. specialized tech for work or study vs. offline uses; and about some specific social media behaviors associated with nutritional habits (e.g., sharing food content); - Lifestyle measures: to the PREDMED scale for nutritional habits and The Pittsburgh Sleep Quality Inventory; - Problematic use: the Mobile Phone Problematic Use scale and the Bergen Social Media Addiction scale. The participants declared using digital technology for an average of 55.38 hours over a week, a notable amount considering that average awake hours over a week are about 112 according to the literature. Social media, non-social media Internet and work/school technologies were largely the tools taking more time of use. Weekly hours of use, percentage of non-social media Internet use and the ad-hoc scale about social media for nutrition predicted nutritional habits: this result corroborates findings about technology use’s impact on nutrition. On the one hand, people use Internet and social media to find information on food preferences and diet; on the other hand, consuming food while using technology impacts food assumption. Percentage of time using social media and social media for nutrition predicted sleep quality issues. Multiple studies found that poor sleep quality relates to intensive social media use, but also with emotional investment in them. Social media for nutrition and sleep quality issues consistently predicted all problematic use measures. Maladaptive lifestyles (e.g., delayed sleep onset timing) may weaken self-control and act as a contributing factor in the development of Internet addiction. While preliminary, the results emphasize the role of everyday technology use in lifestyles as well as the role of lifestyles in problematic use. Future research, as well as studies in the development of technological educational/healthcare programs, should make users aware of risks and promote virtuous use of technology.

ABSTRACT. Digital relaxation media such as guided meditation, relaxing music, ASMR, and hypnotic audio tracks are increasingly used through online platforms and social media. However, it remains unclear whether hypnotic audio content provides benefits beyond generic relaxation stimuli. In particular, hypnotic voice tracks may combine relaxation with elements of self-hypnosis, self-suggestion, and mild alterations of consciousness. This study compared the effects of a brief hypnotic voice track and a relaxing music track delivered online, focusing on relaxation, emotional state, engagement, and the role of previous familiarity with ASMR-like content. A total of 102 participants were randomly assigned to listen either to a 7-minute hypnotic-relaxation voice track (n = 52) or to a relaxing music track (n = 50). The hypnotic audio included breathing regulation, muscle relaxation, hypnotic suggestions, and self-suggestion instructions aimed at increasing calmness and perceived ability to relax. Participants completed measures of relaxation, emotional state, distress, and engagement before and after the audio experience. Both conditions significantly improved general relaxation, muscle tension, cardiovascular-related relaxation, emotional valence, dominance, and distress, while also reducing arousal. No overall differences emerged between the two conditions on these variables, suggesting that both audio tracks were effective in promoting short-term relaxation and emotional regulation. However, the hypnotic voice track was perceived as more engaging than the relaxing music condition, t = 1.897, p = .030, d = 0.54. In addition, participants who habitually used ASMR or other relaxation-oriented media reported greater sleepiness after the hypnotic voice track, whereas this effect did not emerge among participants without previous ASMR habits (F = 4.882, p = .029). Brief remotely delivered audio tracks appear effective in promoting relaxation and emotional well-being. Although the hypnotic voice track did not produce stronger overall effects than music, it was associated with greater engagement and increased sleepiness in participants familiar with ASMR-like content. These findings suggest that individual predispositions and media habits may influence responsiveness to digital hypnotic interventions.

ABSTRACT. Background: Sports injuries can substantially impact physical functioning and athletes’ psychological well-being. In high-intensity contact sports, such as rugby, injuries may lead to increased stress, decreased perceived control, motivational disruptions, and challenges in regulating emotions. These factors can affect adherence to rehabilitation and recovery plans. In this context, psychological resources such as coping strategies, resilience, and self-compassion may play a protective role by supporting adaptive adjustment during the post-injury period. Furthermore, the body-mind connection, or awareness of and relationship with bodily signals, may be particularly relevant in recovery by influencing how athletes interpret physical sensations and manage stress responses.

Objective: This pilot study will explore and compare the effects of two brief techniques delivered via a fully online protocol: (1) a self-compassion exercise and (2) Seligman’s ABCDE technique.

The primary objective is to explore how coping strategies, resilience, self-compassion, and body connection relate to athletes’ responses to the two techniques, and to examine immediate changes in subjective relaxation and well-being.

Methods: 30 male rugby players who have experienced a sports injury and are currently recovering will be recruited. Participants will complete an online battery of self-report measures assessing coping strategies, body connection (body awareness and body dissociation), self-compassion, and resilience. Participants will be randomly assigned to one of the two intervention groups. Visual Analogue Scales (VAS) and the Relaxation State Questionnaire (RSQ) will be administered before and after the intervention to capture immediate state changes associated with the techniques.

Conclusions: Data collection is ongoing. This study is expected to provide preliminary evidence on the effectiveness of brief online techniques based on self-compassion and positive psychology in supporting adaptive psychological resources during the recovery of rugby players after injury. The findings may inform the design of future controlled studies and the development of targeted digital interventions to support athletes' psychological adjustment during the post-injury phase.

ABSTRACT. Progressive age-related physiological decline can lead to physical and cognitive frailty, impairing dual-tasking abilities essential for daily functioning. Virtual reality (VR) and 360-degree media have emerged as effective tools for cognitive-motor rehabilitation, offering enhanced engagement and ecological validity. While immersive VR shows promise in clinical settings, non-immersive solutions may be more appropriate for unsupervised home-based training in older populations. Moreover, evidence of the feasibility and acceptability of unsupervised home-based VR training in older adults remains limited. We developed DUAL-REHAB, a 360-degree media-based application designed to support cognitive-motor dual-task (CMDT) training in older adults with mild cognitive impairment (MCI) and subjective memory complaints (SMC). The intervention consists of two phases: an initial ten-session supervised hospital-based phase using immersive VR via head-mounted display, followed by a ten-session autonomous home-based training delivered through a tablet-based non-immersive version. Both versions employ 360° photos and videos to create realistic virtual environments. The current study is part of a broader research project evaluating DUAL-REHAB's effectiveness. Specifically, this investigation addresses the aforementioned literature gap by exploring the feasibility and acceptability of the home-based phase on a sample of SMC using a qualitative approach. Participants were recruited voluntarily from outpatients of the Department of Medicine, Neurology, and Rehabilitation at Istituto Auxologico Italiano (Milan, Italy). Inclusion criteria: (i) age ≥ 65 years, (ii) normal or corrected-to-normal vision, and (iii) presence of SMC, defined as self-reported memory difficulties in the absence of objective cognitive deficits at neuropsychological assessment. Participants who completed the training with the DUAL-REHAB application are contacted and invited to participate in a follow-up session. Through semi-structured interviews combining closed-ended and open-ended questions, we investigate participants' subjective experiences regarding feasibility (degree of autonomy, adherence to treatment, system usability, adaptation to domestic environment) and acceptability (satisfaction, safety, motivation, engagement, perceived usefulness) of the home-based sessions. This qualitative methodology offers several advantages over standardized assessments. First, it allows participants to spontaneously identify factors they consider crucial for the feasibility and acceptability of unsupervised VR training, rather than limiting responses to predetermined categories. Second, it captures the nuanced experiences of older adults navigating novel technologies in their home environment, providing rich contextual insights that quantitative measures alone cannot reveal. Third, open-ended responses enable the identification of unpredicted barriers or facilitators, informing future iterations of the intervention and its implementation in non-clinical settings. Interview data will be analyzed using thematic analysis. This method involves familiarization with the data, systematic coding of meaningful segments, identification of recurring themes, and refinement through team discussion to ensure reliability. This approach enables the extraction of patterns and insights directly grounded in participants' lived experiences, supporting both the validation of preliminary findings and the discovery of emerging themes relevant to home-based VR training. Data collection is currently ongoing, and findings will be presented at the conference. The sample size will be determined by data saturation and is expected to include at least 10 participants, according to previous studies using similar methods. Based on previous usability studies of the DUAL-REHAB prototype, we anticipate generally positive appraisals of the training's innovative nature, alongside concerns regarding device interaction without professional supervision. The qualitative analysis will provide actionable insights into the features and conditions essential for enhancing application usability among potentially frail older adults with limited digital literacy, ultimately supporting the development of accessible and sustainable home-based rehabilitation programs.

ABSTRACT. Access to evidence-based mental health care in the Czech Republic remains limited, with long waiting times and uneven availability across regions. These systemic shortcomings are especially prominent in the care for people with borderline personality disorder (BPD) and broader emotional dysregulation, where waiting time for specialized evidence-based care such as dialectical behavior therapy (DBT) ranges from one to two years.

Project Gradient, developed at the National Institute of Mental Health, aims to expand DBT delivery through new in-person programs and digital interventions. As part of this effort, we developed DBT Skills Online: a structured self-guided digital skills training program designed to reach people regardless of their location or financial situation.

DBT is well suited for online delivery due to its structured skills-based curriculum emphasizing psychoeducation and repeated practice. The program was developed in accordance with DBT standards and under the supervision of licensed DBT therapists. It consists of five modules. The first module provides a foundational introduction to the dialectical framework and introduces core skills used in DBT, followed by the four standard DBT modules: Mindfulness, Distress Tolerance, Emotion Regulation, and Interpersonal Effectiveness.

The design of the program focuses on accessibility and combines multiple design strategies shown to support engagement in self-guided digital interventions. The lessons present skills in concise, practice-oriented segments and integrate video content with practicing DBT therapists, experiential exercises, narrative elements and hand-drawn illustrations designed to foster a sense of human presence in a digital format. The program also explores how self-guided digital interventions can maintain engagement while remaining scalable.

The program is designed for three audiences: individuals with BPD or emotional dysregulation, mental health professionals seeking to deepen their DBT knowledge, and graduates of in-person DBT programs who wish to refresh their skills.

DBT Skills Online launches in April 2026. Program effectiveness will be evaluated using a mixed-methods design including pre-post and follow-up self-report psychometric assessments and qualitative feedback collection through both self-report forms and human-led interviews. Participants are expected to number in the hundreds and will include individuals experiencing emotional dysregulation, their close others, and mental health professionals. Multiple instruments will be used for each target group to assess changes in DBT skills use, clinical symptoms of BPD and related conditions, stigma and other psychosocial variables of direct relevance, along with program engagement and user experience.

Clinical and psychosocial outcomes will be compared with a waitlist control group and alternative interventions provided within Project Gradient. Preliminary implementation and feasibility data will be presented. The outcomes of evaluation will contribute to current efforts to design engaging self-guided digital interventions while maintaining a sense of human presence in online therapeutic environments.

ABSTRACT. Background: Hikikomori, a prolonged form of extreme social withdrawal, has become a global phenomenon in the post-COVID era. While excessive internet use has long been associated with hikikomori risk, little research has examined how online platforms are used, particularly the distinction between active and passive social media engagement.

Methods: A sample of 984 UK young adults (18–24) completed measures of daily time spent across online activities and the Hikikomori Questionnaire (HQ-25M). Participants identified which social media platforms were used more actively and which were used more passively. Participants also reported the one platform they used most actively and passively. Associations between online activity patterns, platform type, and hikikomori risk scores were analysed, including comparisons of those above and below the established clinical cutoff (≥42).

Results: Social media “active-use” platforms were WhatsApp, Snapchat, Discord, Telegram, Instagram and “passive-use” platforms were TikTok, YouTube, Twitch, Pinterest. Hikikomori risk was significantly associated with greater passive social media use, greater online gaming, and greater online entertainment. Active social media use was unrelated to hikikomori. Participants who passively used both an active-use and passive-use platform had the highest hikikomori risk scores, significantly exceeding those who used social media actively.

Conclusions: Findings highlight that the mode of online engagement, not simply time spent online, is central to hikikomori risk. Passive consumption on platforms that permit minimal reciprocal interaction may reinforce social withdrawal, while active use does not appear detrimental. Interventions targeting digital behaviour, particularly encouraging active online engagement and reducing passive viewing, may help mitigate hikikomori risk among young adults.

ABSTRACT. Social media platforms have become a mainstay of our modern world and have fundamentally changed the way we connect with each other. Humans are now capable of acquiring information regarding other people at an unprecedented speed and reach, as well as impacting other individuals from anywhere in the world. As a result, the concept of “Cancel Culture” has emerged as a topic of interest during the past decade, with previous research detailing various ways in which it may impact individuals and the public at large. Most research on this topic focuses on specific cancelling episodes or the broader concept in its entirety, with most works using qualitative or cross-sectional quantitative methodologies. Consequently, experimental data on the topic is limited, given the nature of the phenomenon, and the existing literature often focuses on cancelling proclivity under varying contextual differences. As such, little is known about the perceptions of being cancelled, as well as the impact underlying innate pressures pertaining to this topic may have on said perceptions and behavior related to Cancel Culture. Particularly, although past research would suggest an unrealistic optimistic bias pertaining to being the target of a cancellation, there’s a lack of quantitative data to support this claim. As such, we conducted two experimental studies employing 2x2 designs, using a vignette methodology. Vignettes were subjected to a pilot study before broader use. Pressure for a Perfect Conduct, the previously proposed conceptual representation of the pressure individuals may experience to conduct themselves perfectly under the threat of being canceled, was included as a covariate. This construct was measured using a previously developed and validated instrument. Study 1 delved into the effects of the person being cancelled (Reader vs Third-Party) and speed of cancellation (Immediate vs Delayed) on how these scenarios were perceived (realism, procedural justice, and Cancel Culture accuracy). We employed adapted versions of validated instruments for both realism and procedural justice, while using novel questions to assess Cancel Culture accuracy. Study 2 tackled the impact of agreeing with a cancelled opinion (Yes vs No) and anonymity when posting (Yes vs No) on both supporting and cancelling behaviors. Results for Study 1 (N = 221, Mean Age= 23.0, 83.7% women) indicate that realism perceptions are significantly impacted by the target of the cancellation (F = 8.07, p = .005) and pressure for a perfect conduct (F = 11.02, p = .001), but not by the speed of the cancellation (F = 0.28, p = .600). Cancel Culture accuracy, on the other hand, are significantly impacted by the speed of the cancellation (F = 6.26, p = .013) and pressure for a perfect conduct (F = 9.34, p = .003), but not by the target of the cancellation (F = 0.23, p = .637). Finally, procedural justice perceptions are significantly impacted by the cancellation’s target (F = 4.32, p = .039), but not by the speed of the cancellation (F = 2.52, p = .114) or pressure for a perfect conduct (F = 0.04, p = .839). Data for Study 2 is being collected at the time of submission, going through a second anonymous data collection process. Our data support the claim that there is an unrealistic optimism pertaining to cancelling situations, and that higher levels of pressure for a perfect conduct are associated with higher reported values of realism, allowing these individuals to more correctly assess the world around them. Our results shed light on an underexplored portion of the Cancel Culture phenomenon, and call for further study of the individual and contextual variables at play in cancelling behaviour and avoidance.

ABSTRACT. In the contemporary digital age, the rapid expansion of social media platforms has resulted in a substantial transformation of the intellectual and ideological landscape of university students. This development poses considerable challenges for educational administrations with regard to the safeguarding of intellectual security. The present study aims to explore the complex relationship between patterns of social media use and their impact on intellectual extremism among university students. The study also seeks to examine the effectiveness of official institutional discourse in the digital sphere. The present study employed a descriptive-analytical approach within a parallel, mixed-method design to achieve a comprehensive understanding of this phenomenon. The quantitative phase of the study incorporated a 19-item questionnaire, adapted from four previous studies, with the objective of assessing social media usage patterns, levels of trust in official news accounts, and levels of intellectual extremism (IE). The instrument was subjected to validation by expert arbitrators, and subsequent revisions were made in response to their feedback. The scale demonstrated acceptable reliability (Cronbach's α = 0.63). The questionnaire was disseminated online to a sample of 1,011 students from 33 public and private universities across Oman. The qualitative phase employed a "snowball sampling" technique, initiated with five key participants who recommended peers willing to discuss sensitive issues. This resulted in the conduction of 20 semi-structured interviews, which were conducted via Google Meet. The interviews conducted for this study explored participants' personal views on intellectual security and official government accounts as trustworthy news sources. The integration of these methods was instrumental in facilitating robust triangulation of the data, thereby ensuring that statistical trends were contextualized within the students' lived experiences. Quantitative results revealed varying differences across genders and academic majors across platforms. The differences were found to be statistically significant for all platforms except TikTok (p < 0.05). However, effect sizes remained small to negligible across all platforms, indicating the limited practical significance of gender and academic major on usage patterns (Cramer's V ≤ 0.15). Moreover, a chi-square (X2) test revealed no statistically significant differences in IE levels between social media users and non-users. The consistency of results observed across various variables provides a statistically robust basis for triangulation, thereby indicating that the potential for intellectual security risks is not constrained to specific platforms. Moreover, the present study contradicts the prevailing academic consensus linking social broadcasting to the phenomenon of political polarization. The quantitative analysis demonstrated a statistically significant inverse relationship between the use of Platform X and the level of IE (p = .028). The findings of the qualitative research indicated that students employed Platform X as a medium for verifying information, utilizing it as a platform for critical thinking and news comparison. A chi-squared (X2) test was also conducted, which revealed no statistically significant relationship between the type of news source to which students are exposed (whether official or unofficial) and their level of IE. This result is indicated by the X2 value (1, N = 1008) = 2.42, p = 0.12. In addition, Spearman's rank correlation revealed no statistically significant relationship between participants' relative trust in news accounts (official vs. unofficial) and their level of IE (rs = .01, p = 0.822). In general, the type of news source and the degree of trust placed in official versus unofficial sources were found to be non-significant predictors of students' tendencies toward IE. A significant communication gap was also identified regarding official government accounts. The study concluded that, to promote intellectual resilience, a strategic shift is necessary from existing “broadcast-only” institutional communication models to “participatory media” models, in line with the Oman Vision 2040.

ABSTRACT. With 5.66 billion users spending an average of 2.5 hours a day on social media (Kemp, 2025), it is important to build knowledge about harmful and beneficial aspects of social media use. Social support is related to positive well-being outcomes (Marshall et al., 2023; Zell & Stockus, 2025), so it could be an important factor to foster in social media environments. Previous research has found that seeing a supportive social media post can increase viewers’ intentions to share prosocial content through a process of goal contagion (Tobin et al., 2025). Goal contagion occurs when someone infers and then adopts a goal they see someone else pursuing. The current study sought to extend past research by examining the contagion of emotional and informational support goals via social media comments and validating new emotional and informational support measures. This allows us to test whether perceivers catch a subordinate support goal (emotional or informational support) or a superordinate support goal (any type of support). We preregistered hypotheses for subordinate goal contagion: perceivers who viewed a comment offering emotional support would infer and adopt emotional support goals and perceivers who viewed a comment offering informational support would infer and adopt informational support goals.

We recruited 374 Facebook users via Prolific to take part in an online experiment. The sample included adults (M age = 40.95; 55% female, 44% male, 1% non-binary) from Australia (24%) and the US (76%). Participants were asked to imagine seeing a comment in their Facebook News Feed from a close friend in response to a post about a missing cat. Depending upon randomly assigned condition, the comment offered emotional support, informational support, or neither kind of support (control). Next, participants were asked to actively use Facebook for 5 minutes and then report on their level of emotional and informational support during the session. They also reported how much they would like to provide emotional and information support on social media in the next week. Lastly, participants were asked to recall the comment they had seen earlier and rate the extent to which the author was trying to provide emotional and informational support.

Exploratory factor analyses supported separate emotional and informational support factors for the support provision and intention scales. As predicted, mediational analyses revealed significant indirect effects of viewing the emotional support comment (vs. informational support or control) on emotional support provision and intentions via perceptions of emotional support and significant indirect effects of viewing the informational support comment (vs. emotional support or control) on informational support provision and intentions via perceptions of informational support. That is, the more people inferred the intended type of support from the comment, the more they reported engaging in or wanting to engage in that type of support themselves. There was also some evidence of superordinate goal contagion whereby inferring one type of support predicted greater levels of both emotional and informational support. Lastly, there were unexpected significant direct effects of the control condition such that it increased both types of support.

The indirect effects support the idea that subordinate and superordinate goal contagion can occur when people read other people’s comments on Facebook. However, the direct effects suggest that reading other people’s comments can trigger other processes as well. Taken together, the results suggest that reading other people’s comments can influence the extent to which social media users support others, but that multiple processes may lead to complex effects. Validation of two new measures of emotional and informational support in the current study offer researchers flexible and reliable tools to use to further investigate how social support operates in social media environments.

ABSTRACT. Generative artificial intelligence (GenAI) is increasingly being used to find and synthesize news, raising questions about how GenAI-mediated content influences users' evaluative judgments and behavioral intentions. This preliminary study conceptualizes large language models (LLMs) as communication mediators that systematically transform real news content while preserving its core informational substance, and investigates how different LLM-implemented communication strategies shape readers' perceptions and explicit intentions when engaging with GenAI-modified information. We examine two communication strategies applied to authentic news articles originally published by ANSA across three thematic domains: an engagement-oriented strategy emphasizing emotionally evocative language, narrative framing, and persuasive stylistic elements, and a reliability-oriented strategy emphasizing neutral tone, increased factual density, explicit source attribution, and transparency cues. Using a within-subjects design, participants (N = 20) evaluated both versions of the same news items, rating each article on five dimensions: perceived credibility, trustworthiness, clarity, emotional engagement, and intention to share, and an additional measure of awareness of GenAI mediation (defined as explicit recognition that an AI system has communicatively altered the content). This design isolates the effects of communication strategy from those of informational content, topic, and factual accuracy, enabling a focused examination of how LLM-mediated framing alters perceptions of otherwise identical news content. To examine individual-level moderators of susceptibility to GenAI-mediated framing, we additionally assessed participants' cognitive reflection ability (Cognitive Reflection Test; CRT) and bullshit receptivity (Bullshit Receptivity Scale; BSR). We hypothesize that communication strategy and awareness of GenAI mediation interact to produce overreliance on GenAI-modified news despite high awareness of AI involvement. Specifically, engagement-oriented framing would maintain more favorable evaluations and stronger usage intentions relative to reliability-oriented framing even when AI intervention is consciously recognized. We further hypothesize that individual differences in CRT and BSR moderate this effect, such that lower cognitive reflection and higher bullshit receptivity amplify susceptibility to engagement-oriented framing. Preliminary pilot data (N = 20) provide initial directional evidence consistent with these hypotheses. Within-subjects comparisons revealed a small advantage of reliability-oriented articles on perceived credibility (MREL = 5.86, SD = 1.07 vs. MENG = 5.50, SD = 1.07; d = 0.34). Notably, engagement-oriented articles were associated with higher perceived AI involvement (MENG = 4.62, SD = 1.30 vs. MREL = 4.00, SD = 1.00; d = 0.54), suggesting that stylistic markers of LLM rewriting are more perceptible in emotionally amplified content. At the individual-difference level, BSR scores showed a strong positive association with perceived source trustworthiness (rs = .69, p < .01), indicating that readers less critical of unverified claims extend greater trust to GenAI-modified content regardless of communication strategy. CRT showed the opposite pattern, negatively associated with both source trust (rs = −.71, p < .01) and intention to share (rs = −.56, p = .03).  As a preliminary investigation, this study establishes the empirical conditions for research aimed at examining how the communicative features of GenAI-mediated news shape readers' judgments and behaviors beyond factual content, including critical decision-making and exploratory information seeking, across varying levels of AI awareness.

ABSTRACT. The proliferation of AI-mediated news raises critical questions about how personalized content shapes users’ information processing and evaluative responses. Prior research has established that motivated reasoning and confirmation bias lead individuals to preferentially accept belief-consistent information. Work on news personalization has addressed filter bubbles and AI-generated misinformation. However, few studies have experimentally examined how deliberate, belief-congruent AI reformulation of factual news affects credibility judgments, emotional engagement, and sharing intentions, or how individual cognitive styles moderate these effects.

Two interconnected research questions guide this investigation. (RQ1) Does the communicative framing of AI-reformulated news, varying in tone, narrative structure, and informational emphasis, affect perceived credibility and interaction intentions independently of factual content? (RQ2) Does belief congruence between news framing and users’ pre-existing orientations amplify evaluative and behavioral responses, and is this effect moderated by individual differences in analytic thinking?

The study drawing on dual-process theories of information processing and the emerging concept of AI as a System 0, which describes how AI infrastructures shape the pre-reflective conditions under which users encounter information. We propose that AI-mediated news environments do not function as neutral information conduits but as active cognitive preprocessors that personalize content in ways that steer reasoning prior to conscious deliberation. The very features that make AI-curated content feel fluent and credible systematically reduce exposure to disconfirming information and suppress reflective scrutiny. This is the theoretical core of our three hypotheses. (H1) Congruent news framing will elicit higher perceived credibility, source trustworthiness, and sharing intentions compared to neutral or incongruent framing. This could reflect System 0’s individualization function, whereby belief-consistent content is processed with lower cognitive friction and accepted with the same ease as perceptual input. (H2) This congruence effect will be stronger among participants with lower analytic thinking scores (CRT) and higher Bullshit Receptivity (BSR). Individuals less prone to deliberative reasoning are more exposed to System 0’s pre-reflective influence and less likely to override its sycophantic output. (H3) Participants will underestimate AI involvement in belief-congruent content. When framing aligns with prior beliefs, System 0's mediation becomes phenomenologically invisible, impairing users' ability to detect algorithmic influence on their information environment.

The study employs a within-subjects experimental design. Stimuli consist of real journalistic items on six polarizing topics. The topics are as follows: climate policy, capital punishment, mandatory vaccination, AI in education, online speech regulation, and universal basic income. The topics were then systematically reformulated using Large Language Models into congruent and incongruent versions. Crucially, factual content is held constant while only communicative style is varied, isolating framing effects from content differences. Participants’ prior beliefs are measured using a validated 2-item-per-topic Likert battery (1–7), enabling individual belief-congruence indices to be computed for each news item. Dependent measures include perceived credibility, source trustworthiness, clarity, emotional engagement, like and share intentions, and perceived AI involvement. Individual differences are assessed via the Cognitive Reflection Test (CRT), the Actively Open-Minded Thinking about Evidence Scale (AOT-E), and the Bullshit Receptivity Scale (BSR), capturing analytic reasoning style, openness to disconfirming evidence, and susceptibility to rhetorical pseudo-profundity respectively.

This research makes three original contributions. First, it isolates communicative framing effects from factual content using LLM-generated variants of real news items. This is a methodological advance over studies relying on naturally occurring stimuli. Second, it situates AI-driven personalization within the System 0 framework, extending dual-process models to account for AI mediation of the pre-reflective information environment. Third, it provides empirical evidence on whether AI-mediated belief-matching bypasses analytical scrutiny, offering actionable insights for the design of AI-assisted news environments that preserve epistemic autonomy and mitigate automated confirmation bias.

ABSTRACT. Deepfakes allow a person’s face or body to be transferred onto another individual’s movements, producing highly realistic but artificial visual experiences. While often associated with risks such as identity theft, cyberbullying, and misinformation, research highlights therapeutic, educational, and behavioral applications, showing deepfakes can help individuals who have lost their voice, enable therapeutic exposure to past experiences (Hoek et al., 2024; Wiederhold, 2021), or enhance learning through historical figures (Babaei et al., 2025). This study investigates deepfakes as a medium for personalized video modeling. In this vicarious learning approach, individuals acquire skills or insights by observing others' actions, a method that is effective in promoting self-efficacy. Unlike traditional video modeling, this research emphasizes self-observation, fostering interaction between digital self-representation and emotional processing. It examines how viewing oneself in deepfake videos influences emotional arousal, complexity, regulation, self-perception, and reflective processes. Three hypotheses guide the study: 1) Watching a personalized deepfake will produce higher psychophysiological arousal than neutral content. 2) Self-relevant deepfakes will evoke greater emotional complexity, increasing positive and negative emotions, and affect emotion regulation pre- and post-test. 3) EG participants will show more intense emotional and psychophysiological responses than CG participants, highlighting the role of personal relevance. The study has ethical approval from the University of Palermo (no. 189/2024). Fifty adults, balanced by gender, will be randomly assigned to the EG (self-deepfakes) or the CG (neutral deepfakes from a public database). Sample size was estimated via G*Power (α = .05, power = .80, medium effect size f = .25). Participants will be randomly recruited, must be of legal age, and provide informed consent. The experiment has two phases. First, EG participants record facial movements using custom software. These are converted into deepfake videos showing them dancing, chosen for full-body movement while manipulating only the face. Observing themselves performing actions encourages self-reflection, identity recognition, and confidence. Then, all participants watch three videos—two neutral landscapes to establish baseline responses, followed by a deepfake. EG views their own video; CG views a matched neutral video of another person. Psychophysiological and behavioral data are recorded using iMotions, which combines eye-tracking, facial expression analysis, and biometric sensors. Participants also complete demographic and self-report technology questionnaires, the Emotion Regulation Questionnaire (ERQ-CA; Gross & John, 2003; Balzarotti, John, & Gross, 2010), the Visual Analogue Mood Scale (VAMS; Stern et al., 1997) pre- and post-test, and the Self-Assessment Manikin (SAM; Lang, 1980) at post-test. Pre- and post-test VAMS and ERQ-CA scores assess changes in emotional state and regulation. SAM captures emotional valence and arousal, allowing comparison between subjective feelings and objective psychophysiological responses. Combining these measures captures both emotional intensity and the influence of regulation on reactions to one’s digital self. Data collection is ongoing. EG participants are expected to show higher arousal, greater emotional complexity, and changes in emotion regulation compared to CG. Viewing oneself in a deepfake context is likely to activate cognitive and emotional processes related to self-awareness, self-efficacy, and identity. The study also examines correlations between subjective reports and psychophysiological measures, highlighting interactions between conscious reflection and unconscious affective responses. It evaluates iMotions’ sensitivity in detecting subtle responses to personally relevant stimuli. This study advances understanding of the psychological effects of interacting with one’s digital self. Experiencing oneself in deepfakes may promote self-reflection, behavioral change, and increased self-efficacy, demonstrating deepfakes’ potential for controlled psychological interventions. By emphasizing personal relevance, emotional regulation, and identity, this research contributes to cyberpsychology frameworks, showing deepfakes can serve as an innovative research medium and a tool for personalized, transformative experiences. Findings open avenues for digital self-representation in therapy, education, and behavior change, illustrating connections among technology, emotion, and self-perception.

ABSTRACT. Problem. In the context of the Society 5.0 transition, characterised by the deep integration of technology to serve human interests, the intersection of behavioural science and ethical technology assumes paramount importance. A significant challenge in the field of cyberpsychology is that of "digital exhaustion", which arises from the processing of non-canonical social cues (e.g., inverted faces in virtual reality or peripheral avatars). The Peripheral Face Inversion Effect (PFE) has been demonstrated to impose a considerable cognitive load by disrupting holistic processing. In order to promote human well-being in a Human 5.0 context, it is essential that technology takes into account neurobiological diversity in terms of information processing speeds. Methods. This study (N=49, included 18 males) investigated the face recognition of inverted emotional expressions (POFA set) presented for 200 ms at 10° eccentricity. High-speed eye-tracking (500 Hz) was coregistered with a polygenic assessment of two critical loci: polymorphisms of COMT gene (rs4680), regulating prefrontal dopaminergic tone, and BDNF gene (rs6265), mediating synaptic plasticity. Moving beyond the single candidate gene approach, we analyzed the epistatic (gene-gene) interaction as a marker of the complex genetic architecture underlying cognitive control. The data were modeled using ANOVA (time~EXP*ANS*BDNF*COMT). Results. The ANOVA model explained 17.62% of the variance in saccadic latency (time). While average recognition accuracy dropped to 0.53 under high-load conditions, correct identifications were consistently associated with minimal latencies. Genetic factors contributed significantly to the variance: • The BDNF plasticity explained 3.24% of the variance, and carriers of MetMet-genotype demonstrated a 14% reduction in information processing speed (190 vs. 166 milliseconds for Val-homozygotes). • The COMT dopaminergic tone variable accounted for 1.54% of the observed variance, modulating the decision-making threshold. • The BDNF:COMT epistasis was identified, where a significant interaction (1.16%) indicates that the effect of one gene is modulated by the state of another, confirming the polygenic nature of oculomotor efficiency. Discussion. The results are integrated with the role of the driver in the motor control unit (A. Karlsson) and the molecular mechanisms of plasticity (E. Kandel). The theory of predictive coding (K. Friston) is currently being reviewed as a potential means of identifying and reducing errors in prediction. The extended latency experienced by users of designated genetic profiles is indicative of a "perceptual lag", which is a key factor in the Human 5.0 interface. It is vital that these technologies are designed to limit the speed at which content is delivered to users, in accordance with biological limits, thereby ensuring optimum well-being in the digital world.

ABSTRACT. As artificial intelligence (AI) tools become embedded in several contexts including work, healthcare and daily life, understanding how people engage with these systems is essential for designing human-centred technologies. Research on Human-AI interaction has highlighted how individual differences shape attitudes and usage patterns, particularly toward GenAI tools. Grounded in Reinforcement Sensitivity Theory (RST), this study investigates how individual motivational systems, Behavioural Inhibition System (BIS), Behavioural Activation System (BAS) and the dispositional optimism, influence individuals’ attitude and perceived AI literacy in relation to commonly used AI tools. Since RST defines people’s disposition to pursue rewards versus avoid risks, and AI use carries both benefits (performance optimization) and risks (privacy violations, ethical concerns), we hypothesized that risk/reward sensitivity would predict attitudes towards AI. BIS scale captures avoidant behaviour tendencies and a cautious decision-making style; BAS scale regulates behavioural approach in response to rewards and incentives across three subscales; and the LOT-R scale measures dispositional optimism as a general life orientation. Regarding the attitudes and self-perceived literacy of AI technology, AIAS-4 scale was assessed to capture general attitudes toward AI, focusing on how positively or negatively a person evaluates AI’s impact on their life, work and society. Self-perceived AI literacy was measured with the PAILQ-6 scale covering two dimensions: AI awareness (knowledge and understanding of AI) and AI engagement (perceived ability and interest into actively using and exploring AI tools). For this study, a sample of 207 Italian speaking participants completed a self-report questionnaire including the BIS/BAS scale, LOT-R Life Orientation Test Revised 10-item, the AIAS-4 Scale, and the PAILQ-6 Scale. Three multiple regression analyses were conducted to predict AI attitude, AI engagement and AI awareness dimensions. Results indicated that BAS_Drive, reflecting goal persistence and sustained motivation in absence of immediate reward was the strongest positive predictor of both AI attitude and AI literacy engagement dimension. Conversely, BIS merged as a significant negative predictor of both literacy awareness and engagement dimensions. BAS_Reward positively predicted AI literacy awareness dimension, whereas BAS_Fun and Optimism were non-significantly associated with any outcome. Overall, given that the models explained between the 10% and 15% of the variance across dependent variables, it is necessary that future findings be further examined through experimental research considering other variables that are no longer examined in this study. The fundings of this study suggest that information-seeking behaviour regarding AI tools, without necessarily using it, may be partly predicted by a personal predisposition by seek out new developments and opportunities from which to gain benefits in short term. Conversely, for people who tend to focus more on risks, it appears that both information-seeking behaviour and the actual use of AI tools may be lower, a possible explanation could be the ambiguous nature of this technology, particularly with regard to GenAI chatbots, for which the risks and modes of interaction are not clear and still a subject of ongoing debate. On the other hand, a positive attitude toward this technology, as well as its practical use, appears to be partly driven by a motivational tendency to pursue one’s goals whether professional or personal so that, in this case, AI technology could be seen by the individual as a useful tool for achieving those goals.

ABSTRACT. Background: Decision tools refer to communication technologies designed to support the management of illness and healthcare risks. These tools aim to reduce decisional conflict and improve health monitoring. Over the years, their use in the healthcare context has increased. However, not everyone is comfortable using them; some people prefer to consult directly with physicians when making decisions about their health. In particular, some people reported lack of trust in new technologies, with others favoring discussions with professionals. On the other hand, some believe that these technologies may be better suited for detecting symptoms and are becoming more common in everyday online interactions.

Methods: The aim of the present study is to explore people's attitudes towards the use of digital decision aids in the healthcare context. 37 participants (Mage=41.51, SD=11.75) voluntarily agreed to participate in this study. The majority of them were females (76%) with a higher level of education (37.8% master degree, 32.4% post-degree). The 40.5% of participants have already used digital technologies in a healthcare context, and among the declared tools there are ChatGPT, teleconferencing, health app, internet. All the participants were invited to read a description of a digital tool that supports health decisions and to fill-in a battery of questionnaires. After that, a self-report measure was used to assess the emotions and attitudes related to the digital decision health tool on a 5-point Likert scale (from 1 = not at all; 5= extremely): Fear, Frustration, Worry, Comfort, Curiosity. Additionally, the Negative Attitude toward Robots Scale, adapted for the use with digital tools, was administered to explore attitudes of participants towards digital decision tools. Lastly, the State-Trait Anxiety Inventory was used to assess participants' trait anxiety, while the Decision Self-Efficacy explored generalized self-efficacy expectations based on past experiences and the tendencies to attribute success to skill rather than chance.

Results: The data revealed that participants expressed the following emotions: Fear (M = 2.70, SD = 1.08), with 51.4% indicating strong disagreement with experiencing this emotion; Discomfort (M = 3.35, SD = 1.09); Frustration (M = 2.35, SD = 0.98), with 48.6% reporting that they did not experience this emotion; Worry (M = 3.03, SD = 0.80), with 45.9% reporting moderate levels; Security (M = 2.95, SD = 1.03); Serenity (M = 3.00, SD = 1.05); and Curiosity (M = 2.72, SD = 0.96). Correlation analysis indicated that worry about the use of digital decision-making tools for health was positively associated with trait anxiety (r = .362, p < .05), whereas security (r = -.535, p < .01) and serenity (r = -.369, p < .05) related to the use of these tools were negatively associated with trait anxiety. No correlations have been found between the NARS scale, decision self-efficacy, and State-Trait Anxiety Inventory Scale.

Conclusions: The present results indicate opposite attitudes toward the potential implementation of digital tools in healthcare. On the one hand, participants report a sense of security and serenity, while others show moderate to high levels of discomfort and concern. Participants in this study were not familiar with using digital tools, leading to mixed emotions towards new technologies, ranging from openness to unease. According to the literature, individuals need to be trained in the use of digital technologies as new instruments for healthcare to understand how these tools work and their utility, which can help to promote a sense of comfort with them. This could promote digital decision making tools' implementation in different populations, such as people who face a chronic disease where healthcare decisions need to be supported as a crucial step of their pathway.

ABSTRACT. Generative AI has become one of the most widely used digital technologies among young adults, driven by its increasing integration into education, work, social life, and entertainment. Although adoption is accelerating, our understanding of AI's effects on social development and wellbeing remains limited. While generative AI introduces new opportunities, it also presents significant risks, particularly regarding ethical use, privacy, and algorithmic bias. To understand how societies might navigate these competing promises and risks, this research draws on young people's sociotechnical imaginaries (STI), or their visions for how AI could serve the collective good. Youth are not only skilled users of digital technologies but also creators, critics, and shapers of technological futures. As such, their insights, questions, and expectations about AI are essential for informing sociotechnical systems that are responsive, robust, inclusive, and future‑proofed. This study collected data from an online survey (approved by the UNU human ethics committee) examining young adults' feelings about generative AI and their beliefs about its positive and negative future implications. Participants were 888 young adults aged 18–25 (M = 22.81, SD = 1.69; 50.7% female) from 27 countries. Feelings toward AI were measured using a thermometer scale ranging from 1 (very bad) to 10 (very good), followed by an open‑ended question eliciting reasons for these underlying ratings. Based on these scores, participants were clustered into four groups (cold, cool, warm, and hot feelings). They also responded to open‑ended questions regarding the perceived risks and opportunities of generative AI for children and young people now and in the future. A combination of deductive and inductive coding was used to thematically analyse qualitative responses. Thermometer ratings indicated generally positive feelings (M = 7.16), with most participants classed as warm (60%) or hot (21%). Qualitative responses from those with warmer feelings described generative AI as helpful, practical, efficient, and trustworthy. In contrast, participants with colder feelings emphasised distrust, concerns about deskilling, and ethical issues such as plagiarism. Importantly, responses from participants clustered into the mid-range cluster (warm and cool, N = 691) revealed significant psychological tensions in their experiences with generative AI. These tensions were frequently articulated through adversative conjunctions (e.g., "but,” "however,” "even though"), signalling ambivalence or conflicting emotions. Participants commonly noted that generative AI saved substantial time in their work, yet simultaneously elicited feelings of guilt or a sense of "cheating" because tasks were not completed independently. At the core of these tensions was the perception that AI is helpful but not entirely trustworthy and that it enhances productivity while also diminishing autonomy. Participants' reflections on risks and opportunities for children and young people in the future further illuminated these tensions. A prominent concern (reported by 62%) was that younger generations may become overly reliant on AI, fostering laziness or undermining human abilities and autonomy. At the same time, 52% expressed optimism about AI's potential to support skill development, facilitate knowledge acquisition, enhance productivity, and positively shape educational systems. To our knowledge, this study is among the first to document the psychological tensions apparent among young adults' sociotechnical imaginaries concerning generative AI. These tensions may reflect emerging forms of cognitive dissonance, where conflicting beliefs and emotions coexist and may influence both how young people engage with AI and their general wellbeing. Additionally, this work highlights that understanding young people's complex sociotechnical imaginaries is essential for informing the design and development of future‑proofed, equitable AI systems. Future research should further examine the antecedents and consequences of these tensions and consider how they shape young people's engagement with AI, as it continues to play a central role in shaping our shared technological futures.

ABSTRACT. In the context of modern psychological safety, the issue of the 'human factor' in the operation of critical socio-technical systems (nuclear power plants, aviation) necessitates a shift from descriptive error models to explanatory computational frameworks. This study is devoted to the analysis and search for ways to overcome 'operational deadlocks' — emergent 'pathologies' of decision-making that arise in conditions of acute time constraints and high uncertainty. We propose a theoretical approach based on the Free Energy Principle and active inference theory. This approach views decision-making style as an embodied computational phenomenon determined by the stability of an individual's internal computational model. The hypothesis of this work is that operational deadlock arises as a result of a clash between incompatible cognitive architectures with different levels of 'belief accuracy'. In order to test the hypothesis, a computational phenotyping pipeline was implemented, which allowed two polar archetypes to be parameterised based on the analysis of specialised text corpora (CFR standards, ASRS reports, Enron corpus): 'Proceduralist' with high cognitive rigidity and 'Pragmatist' with dominant cognitive flexibility. The computational model also formalised the 'embodied price of responsibility' as expected interoceptive free energy — the brain's prediction of upcoming visceral stress, which in critical situations can block adaptive behaviour. The research methodology included iterative multi-agent modelling of various interaction scenarios. The results showed that in abnormal situations, the probability of an operational deadlock in dyads with different levels of cognitive flexibility reaches 89%, leading to systemic paralysis. Research has shown that 'Proceduralists' demonstrate high accuracy in routine operations, but lose efficiency exponentially when faced with anomalies (decision time increases to 128 cycles compared to 45 for 'Pragmatists'). This paper has demonstrated that the key factor in overcoming such operational deadlocks is individual cognitive flexibility, which provides a high 'epistemic drive' — the ability of a system to actively search for information to reduce uncertainty. We are pleased to propose a novel model of human-machine symbiosis, where an AI mentor acts as a cognitive mediator, thereby minimising the 'joint free energy' of the 'human-machine' dyad. Our calculations have confirmed that the introduction of symbiotic AI, which initiates strategic epistemic actions, increases the success rate of resolving deadlock situations from 11% to 78%. The research results contribute to the implementation of the Human 5.0 concept, which focuses on creating ethical, human-centred technologies that promote well-being and sustainability in the face of global technological challenges. The proposed approach allows for the diagnosis of the risk of organisational paralysis and the design of decision-making support systems capable of adapting to the individual cognitive profiles of operator employees to ensure maximum system security.

ABSTRACT. Many have come to rely on social media to share information, connect with others, and stay current with the world around us. Having come of age in a digitally connected world, interactions across cyberspace have fostered a culture where younger adults can develop an implicit bias towards consuming content from sources which may or may not be trustworthy or reliable. Even those who are digitally literate succumb to the implicit trust required by virtue of the speed and scale of content made available to us, and it is therefore unreasonable to assume that consumers of social media can fully digest or fact-check everything to which they are exposed. Emotionally provocative, highly divisive content is amplified by the algorithms that drive platform engagement, collectively contributing to political polarization in the United States (US) at present. Nevertheless, we continuously make judgments about the credibility, intent, and truthfulness of the content we consume, an endeavor that has become profoundly more difficult as online social ecosystems overlap with one another and have been inundated with bots and a combination of misinformation, disinformation, and malinformation.

This presentation will report on preliminary analysis from data collected in 2025-2026 from over 100 Black and White US college students, all of whom were active social media users with at least some awareness of political issues in the US. African American participants were intentionally oversampled to address persistent representational gaps in the literature. The study focuses on perceived prevalence of mis/dis/malinformation, trust in algorithmically curated content, exposure to politically aligned and politically challenging content, and perceptions of social media’s role in political polarization. It also incorporates self-reported perspectives on salient historical events that have shaped respondents’ socio-political awareness, which will be discussed in the context of digital engagement patterns for this sample of non-clinical, primarily Gen Z individuals.

Exploratory analyses revealed several recurring themes including: (1) there is a broad belief that social media contributes to and intensifies polarization; (2) respondents had an awareness of problematic information within social media feeds and that their perceptions of misinformation, disinformation, and malinformation can be conceptualized as a unified construct; (3) there is a perception that algorithms amplify controversial content in ways that contribute to polarization, with higher perceived polarization also associated with the belief that one’s social media feeds tend to align with preexisting political views; (4) respondents who more strongly endorsed algorithms prioritizing controversial content also reported higher perceptions of misinformation prevalence and stronger beliefs that social media amplifies polarization; and (5) there is an emerging awareness of how algorithmically curated content influences perception, and that lower trust in algorithmically delivered content is associated with higher perceived prevalence of misinformation.

Together, these findings provide insight into the cyberpsychological extension of traditional psychological perspectives on selective exposure and ideological reinforcement. Furthermore, because time spent on social media shows comparatively weaker associations with these outcomes, beliefs about algorithms, misinformation prevalence, and alignment of content with existing views appear to be more predictive than usage alone. One potential implication of this pattern is the development of a cognitive schema in which algorithmic incentives are understood as degrading the quality of information encountered online and embedding polarization within the larger information ecosystems that shape perception. This presentation will highlight these findings and discuss next steps needed to better understand emergent perceptual ecosystems among young adults, in which algorithmic distrust, information distrust, and perceived polarization intersect to form the basis of a structured belief system. These results will also be discussed within the broader context of racially differentiated historical reference points and lived experiences, underscoring the importance of centering marginalized voices in cyberpsychological research.

ABSTRACT. AI holds an amazing potential to optimize productivity and enhance interpersonal connectivity. However, it is also leveraged to manipulate emotions and influence phenomenological perception. The widespread introduction of mis/dis/malinformation into social cyberspaces has produced shifts in behavioral heuristics that are eroding traditional mechanisms of trust appraisal, while weaponization of cultural reference points fuels distrust about fact versus opinion. We are increasingly offloading our critical thinking and unconsciously losing our desire and motivation for it, with sensory-rich and highly anthropomorphic stimuli uniquely capable of exploiting our phenomenological perception. Emotionally driven content creates parallel yet competing narratives of an objective reality, particularly noticeable when subcultural cyberspaces experience a transient convergence with mainstream society, such as during the September 2025 killing of Charlie Kirk—polarizing face of the Turning Point USA organization. Forensic evidence tied the yet-to-be-named shooter to online gaming communities, which law enforcement and the public seemed rushed to attribute to his motives and political leanings. Symbols and writing on the bullet casings tapped into a meme-based subcultural space where plausibly deniable distortions of an otherwise mutually agreeable reality were created. In this session, we outline an open-source research methodology built upon a cyberpsychological foundation capable of visualizing the migration of digital content from subcultural fringe to mainstream awareness. We demonstrate the creation of a cyberpsychological meme genealogy to trace subcultural reference points online and map them to spikes in social media user activity over time. We provide as a case study the January 2026 US Presidential signing of the “Whole Milk for Healthy Kids Act of 2025.” To some, this was a typical policy change modifying requirements for public schools participating in the National School Lunch Program, but in more esoteric online communities, it was framed as a white supremacy dog whistle, offering comparisons to alt-right memes from over a decade ago. Our session does not attempt to adjudicate ideological intent. Rather, the dog whistle claim is reframed as an empirical question of digital traceability. We ask: Where did the frame of reference originate, did (meme-based) content become digitally manipulated over time, and are there identifiable patterns attributable to specific online communities or influencers who may have amplified these messages? Methodologically, the goal of this work is to transfer cognitive control from AI as a manipulative force back to the human being, mitigating some of the threat from bad actors engaging in cybercognitive warfare through the purposeful dissemination of mis/dis/malinformation. Our study uses an ethics-forward approach to automate the process of identifying relevant online communities via temporal keyword-phrase tracking to detect activity spikes, narrative branching, and digital engagement indicators. Data

collection using publicly available information, qualitative coding and analysis of narrative frames, and use of pretrained (offline) LLMs, allowed us to leverage generative AI to respond to our specific research questions to create a time-series visualization of meme propagation. This proof of concept is expected to contribute to the development of a tool that can output a trustworthiness metric to facilitate critical thinking about the legitimacy of emotionally charged stimuli by deanonymizing the origins of culturally applied narratives that have been artificially transformed over time, but which most everyday consumers of online content would not be able to decipher given the sheer volume of online discussion forums that potentially span years and thousands of comments.

ABSTRACT. Political engagement through social media is increasingly understood as following a dual-pathway process that promotes adaptive coping and connection while exacerbating stress through amplified exposure to politically oriented conflict. Digital platforms can intensify exposure to polarizing, emotionally charged content, yet social media can function as either a psychosocial resource or source of strain. Understanding this tension is essential to clarifying how the deleterious effects of political polarization on well-being manifest online and how it varies across culturally diverse Black and White young adults.

Stress and resilience have long been hypothesized as factors at opposing ends of functional versus dysfunctional behavior, and within this dual process framework, they represent key variables capable of shaping the psychological end result of exposure of political content and engagement online. Active participation online such as posting content or offering comments—as opposed to passive engagement which more often is restricted to consuming content—often draws users into exchanges that heighten emotional reactivity. Prolonged exposure to divisive or threatening content can reinforce vigilance, but also, uncertainty and emotional fatigue. Likewise, affective polarization where negative sentiment toward opposing political groups becomes associated with greater stress and weaker social support, suggests that stress may be a central pathway through which polarized digital climates can undermine well-being. Online communities may provide bonding benefits (connection with similar others) and bridging benefits (dialogue across difference) which reinforce perceived support and the capacity to recover or adapt under strain. As such, engagement outcomes may depend on how effective resilience and social resources buffer individuals from stress-inducing exposures.

Guided by this conceptual model, this study examined how political engagement on social media relates to perceived stress and psychological resilience based on self-report data from approximately 150 primarily Gen Z Black and White college students in the United States. Participants completed a semi-structured survey asking about frequency and purpose of their social media use, their engagement with politically oriented content, the Perceived Stress scale and the Connor-Davidson Resilience scale. We hypothesized that higher social media use would predict greater political engagement and stress, but that resilience would moderate these associations. Correlations revealed a statistically significant relationship between social media use and active political engagement (particularly interaction with political discussions and participation in politically charged exchanges), a finding that reinforces the role of social media as a primary environment for political participation with this age cohort. Perceived stress scores were also linked to dimensions of social media use, while stress indicators reflected the simultaneous presence of cognitive and emotional strain. Political engagement varied across participants, including how often they followed political news, responded to posts, or engaged with differing viewpoints. Although these results cannot establish casual pathways, associations between polarization and stress were not strong and the overlap between political activity and social media use suggests there may be an indirect pathway through which politically charged participation contributes to stress perception. However, resilience did not significantly correlate with overall social media use or perceived stress in this sample despite resilience remaining conceptually important for differentiating adaptive from maladaptive engagement outcomes.

Findings from this study indicate that social media engagement is intertwined with political activity and selected dimensions of stress for Gen Z-aged Black and White college students from the US in this sample, although resilience did not show direct associations. Framed through a dual-pathway lens, these results highlight the nuanced role of digital engagement as both a context for social connection and a vector of psychological strain. Understanding how stress and resilience operate in tandem within politically polarized virtual environments is essential to assessing the implications of social media use among young adults navigating polarized public discourse.