International Journal of Human–Computer Interaction

Chen, Wenjuan; Tang, Congyun; Guo, Wenchen

doi: 10.1080/10447318.2026.2676758pmid: N/A

Abstract Creating comics with the assistance of generative systems is becoming increasingly accessible, but existing tools often lack intuitive ways for users to express and control their creative intent. We present CardComposer, a novel drag-and-drop interaction paradigm that enables users to construct comic scenes by visually assembling modular cards representing narrative elements such as composition, camera angle, character pose, and manga-specific symbols. Grounded in principles of visual storytelling and direct manipulation, our system enables users to create structured prompts by arranging and grouping cards spatially and semantically. This interaction model shifts the focus from parameter tuning and text prompting to expressive visual composition. Through a comparative user study with both novice and experienced comic creators (N = 21), we found that CardComposer supports greater perceived control, reduces cognitive load, and fosters creative exploration compared to conventional prompt-based workflows. We discuss the implications for designing visual interfaces that support both structured and flexible authoring in AI-assisted creative tools.

He, Hui; Xu, Ke; Ai, Di; Li, Xing; Zhang, Yi

doi: 10.1080/10447318.2026.2674825pmid: N/A

Abstract Immersive virtual reality (VR) has transformative potential for education, yet the principles for optimal learning environments remain nascent. A 2×2 between-subjects experiment with 118 college students examined the individual and interactive effects of temporal cues (with vs. without) and spatial cues (with vs. without) on cognitive load, perceived presence, and learning performance. Temporal cues reduced intrinsic cognitive load but did not directly enhance learning, whereas spatial cues reduced extraneous cognitive load, increased presence, and improved performance. Mediation analyses indicated that the effect of spatial cues on learning performance was fully mediated by extraneous cognitive load, but this benefit was diminished when temporal cues were concurrently provided. These findings suggest that cue interaction in immersive VR reflects a load-dependent reallocation of cognitive resources rather than simple attentional competition. This study highlights the importance of selectively deploying spatial and temporal cues to optimize cognitive load and learning in immersive VR environments.

Vuong, Quan-Hoang; Kianfar, Fatemeh; Tran, Thi Mai Anh; Sari, Ni Putu Wulan Purnama; Kumaladewi, Cresensia Dina Candra; La, Viet-Phuong; Nguyen, Minh-Hoang

doi: 10.1080/10447318.2026.2676185pmid: N/A

Abstract As digital worlds become increasingly immersive and ecologically sophisticated, they provide novel contexts for examining how human value systems, specifically human exceptionalism, are formed and transformed. This study investigates how virtual environment-modification behaviors and players’ sense of immersiveness jointly predict exceptionalism. Using Granular Interaction Thinking Theoryand the Bayesian Mindsponge Framework, we analyze five key activities—tree planting, flower planting, flower crossbreeding, terraforming, and creating conditions for bug respawn—based on a dataset of 640 Animal Crossing: New Horizons players from 29 countries. Results reveal two behavioral clusters distinguished by controllability. High-controllability behaviors (i.e., flower planting and terraforming) predict higher exceptionalism, whereas the flower-planting effect reverses among highly immersed players. Low-controllability behaviors (i.e., flower crossbreeding and manipulating bug spawning) predict lower exceptionalism, but these associations weaken or reverse under high immersiveness, respectively. These findings suggest leveraging virtual worlds to cultivate Nature Quotient (NQ), mitigate exceptionalist tendencies, and foster eco-surplus culture.

Attanasio, Carmine; Vilone, Giulia; Holzinger, Andreas; Longo, Luca

doi: 10.1080/10447318.2026.2675436pmid: N/A

Abstract One of the goals of Explainable Artificial Intelligence is to enhance users’ understanding of model function and inferential capabilities by providing human-understandable explanations. An Artificial Neural Network has been trained, and interpretable decision rules have been extracted through the C4.5 algorithm. These rules were integrated into a dynamic, interactive interface that allows users to visualise and understand the inferential mechanisms behind model predictions. To rigorously assess the explainability of these rules, this research introduces a user-centred and culturally adapted evaluation, via psychometrics, of two questionnaires for XAI: the System Causability Scale and a multi-dimensional XAI scale. Findings demonstrated acceptable reliability for both questionnaires and an acceptable level of construct validity. Beyond scale translation, this research contributes to knowledge by providing a rigorously validated Italian version of existing explainability and causability questionnaires, enabling reliable cross-cultural evaluation of XAI systems and facilitating comparable empirical studies across linguistic and cultural contexts.

Saxena, Prashant; Prahl, Andrew

doi: 10.1080/10447318.2026.2680242pmid: N/A

Abstract Authenticity is moving to the heart of communication theory as an audience-level judgment built from cues. Current frameworks assume a human communicator and predict that richer channels enhance authenticity, an assumption we test by contrasting human and synthetic authenticity. We define synthetic authenticity as the perception that an artificial agent is real enough to engage despite known ontological hollowness. Drawing on a bibliometric map of 2,685 articles (1966–2025) and a meta-synthesis of 111 AI persona studies (2017–2025), we argue that human authenticity relies on cue accumulation, where redundancy reinforces trust, while synthetic authenticity relies on cue minimalism, where redundancy triggers scrutiny unless sensory fidelity is calibrated to the task. We operationalize this divergence through the Threshold Model of Synthetic Authenticity, which posits sufficiency, script, and relational thresholds beyond which the suspension of disbelief collapses and human-AI interaction shifts from communication to forensic audit.

Abou Chaar, Dana; Bachkirov, Alexandre Anatolievich; Rezaur Razzak, Mohammad; Soliman, Mohammad

doi: 10.1080/10447318.2026.2678540pmid: N/A

Abstract This multi-country study investigates how online dark patterns influence consumer cognition, emotions, and behavior, ultimately shaping broader consumer culture on online platforms. Drawing on data from 698 consumers worldwide, the study employs PLS-SEM to reveal the links through which perceived deception influences symbolic incongruity, brand distrust, and brand hate, which in turn drive brand switching. This research identifies online complaining as a key mediating factor linking negative cognitive and emotional reactions to behavioral outcomes. The findings reveal a multi-stage process whereby dark patterns undermine brand trust and loyalty, thereby cultivating a culture of consumer skepticism. The present study contributes to a better understanding of the societal and cultural ramifications of dark-pattern practices that influence consumer behavior by clarifying the underlying cognitive and affective mechanisms. This highlights the importance of, and urgent need for, regulatory measures to safeguard integrity and trust.

Tian, Nana; Colombo, Giorgio; Schinazi, Victor R.

doi: 10.1080/10447318.2026.2674833pmid: N/A

Abstract Early detection of Cognitive Impairment (CI) is critical for timely intervention, preservation of independence, and reducing the burden of dementia. Yet, most screening tools remain lengthy, clinic-based, and poorly suited for large-scale unsupervised deployment. This article evaluates the test-retest reliability, validity, and usability of mini-SPACE, a short iPad-based serious game for detecting early signs of CI. Participants played mini-SPACE at home without supervision once a week for 3 weeks, with a more challenging version of the game in the final week. Mini-SPACE showed good test-retest reliability in unsupervised settings. Younger age was the primary predictor of performance, usability, and cognitive load. Importantly, the prediction of scores in the Montreal Cognitive Assessment (MoCA) improved with repeated measures. These findings support the screening potential of mini-SPACE with respect to MoCA-defined impairment in unsupervised settings, while clinical diagnostic accuracy remain to be established.

Zhang, Chenqi; Zhao, Hang; Feng, Zehui; Li, Junxuan; Zhu, Yuejia; Huang, Xianjing; Han, Ting

doi: 10.1080/10447318.2026.2676760pmid: N/A

Abstract Artificial intelligence (AI) has shown strong potential to assist product design, yet interaction modes and evaluation methods in human-AI co-creation remained unclear. This study systematically reviewed 50 empirical studies to: (1) categorize AI-supported design tools and their collaboration modes with designers; (2) analyze the design phases and domains supported by AI-supported design tools; and (3) synthesize assessment metrics for evaluating co-creation effectiveness. The review identified three interaction modes: human-centric, copilot, and AI-centric. It also showed that about 80% of studies focused on early ideation phases, while only 30% addressed evaluation and optimization. Assessment frameworks primarily targeted creative outcomes, system efficacy, and collaborative experiences. By mapping current research status and revealing moderation patterns among design phases, interaction modes, and user experience, this review provided actionable guidelines for designing future AI-assisted systems. Research findings highlighted AI’s potential to augment human creativity and provided support for the development of more effective, trustworthy co-creative AI-supported design tools.

Al-Sharafi, Mohammed A.; Helal, Mohamed Y. I.; Al-Zaeemi, Shehab Abdulhabib; Dwivedi, Yogesh K.; Almutairi, Samirah; Alwahaishi, Saleh Hussein; Chae, Inyoung

doi: 10.1080/10447318.2026.2677757pmid: N/A

Abstract The rapid growth of Buy Now, Pay Later (BNPL) services has expanded consumer access to credit, but it has also raised concerns about user satisfaction, trust, and long-term sustainability. This study develops and validates the Experience–Interaction–Trust model to explain and predict user disengagement in BNPL platforms. A multi-stage research design was employed to explore and test the experiential dimensions that shape consumer interactions with BNPL applications. In the exploratory phase, a bilingual corpus of approximately 39,000 authentic Arabic and English app reviews was analyzed. Text preprocessing, sentiment analysis, and topic modeling uncovered eleven core constructs: Harmony, Effortlessness, Favorability, Neglect, Breakdown, Automation, Boundedness, Disparity, Friction, Churn Risk, and NWOM. In the confirmatory phase, PLS regression and XGBoost were used to test and predict the relationships among constructs. The EXIT model demonstrates theoretical robustness and predictive utility, offering practical guidance for BNPL providers seeking to minimize churn, enhance user trust, and support sustainable digital financial innovation.

Zhang, Weiwei; Liu, Ting; Taclis Luo, Yiming; Pang, Patrick

doi: 10.1080/10447318.2026.2678541pmid: N/A

Abstract Amid global aging, older adults urgently need support for health self-management. Generative AI (GenAI) offers personalized health support, yet research lacks systematic analysis of human-computer interaction (HCI) and co-design in dynamic GenAI scenarios. This study used BERTopic-based dynamic topic modeling to trace the evolution of older adults’ needs across three iterative GenAI health video co-design workshops. Twenty community-dwelling older adults aged 60 to 84 participated, providing interview records, focus group transcripts, and user-GenAI prompt texts. Four core HCI themes emerged. Early rounds centered on basic operational demands such as interface clarity and guided prompts. Needs then shifted toward personalization, including adjustable voice speed and dialect support, and finally toward in-depth health value, such as tailored advice and real-time feedback. Concerns about information reliability and privacy persisted across all rounds. These findings guide phased HCI design that prioritizes simplicity, then personalization, with full-cycle privacy safeguards to support age-friendly GenAI health tools.