📚 Research Digest

Finding: An AI-driven framework for adaptive extended reality (XR) environments that uses physiological signals (HRV, GSR) to detect users' affective states and dynamically adjust the simulation, showing improved presence, flow, and reduced cognitive workload in a 60-participant study.

Why it matters: Directly relevant to HCI, user engagement mechanisms, and the design of adaptive systems that maintain flow states—a key concept in gamification, learning environments, and persuasive technology.

Method: Closed-loop architecture combining reinforcement learning, affective state detection via biometrics, and adaptive narrative generation—tested with physiological and self-report measures across 60 participants.

Jargon: *Presence* = subjective sense of "being there" in a virtual environment; *Flow* = optimal engagement state where challenge matches skill; *HRV/GSR* = heart rate variability/galvanic skin response, physiological stress and arousal indicators; *XR* = extended reality (umbrella term for VR/AR/MR); *Ambient intelligence* = environments that sense and adapt to user context automatically.

Finding: The RESET Model proposes a five-phase behavioral framework for sustainable digital self-regulation, synthesizing Western behavioral theories (SDT, Habit Formation, Choice Architecture, CLT, Attention Restoration) with Indian philosophical constructs, arguing that abstinence-based digital detox fails because it removes external stimuli without addressing internal compulsion.

Why it matters: Directly relevant to behavior change design, persuasive technology, habit formation, and addiction-by-design concerns — particularly the structural critique of detox interventions and the choice architecture synthesis.

Method: Systematic model-development procedure validated against conceptual contribution criteria; introduces the Digital Detox Effectiveness Index (DDEI) as a proposed multidimensional outcome measure.

Jargon: *Pratyahara* = withdrawal of senses (yoga concept); *Triguna* = three fundamental qualities of mind (tamas/rajas/sattva); *Sakshi Bhav* = witness consciousness/metacognitive awareness; *DDEI* = proposed composite measure for evaluating digital detox outcomes.

Finding: Diegetic navigation cues (embedded in the game world) trend toward higher immersion and flow than HUD-based guidance, though results were non-significant likely due to small sample (n=10); qualitatively, players found diegetic systems more immersive while HUDs felt clearer and task-focused.

Why it matters: Directly informs game design and gamification decisions about how UI feedback mechanisms affect player experience, immersion, and exploration behavior—relevant to both game design teaching and persuasive/engagement system design.

Method: Within-subjects study using Game Experience Questionnaire (GEQ) plus semi-structured interviews; custom-built exploration environment testing both guidance types.

Jargon: *Diegetic guidance* = navigation cues that exist within the game world itself (e.g., wind effects, birds, environmental signs); *Non-diegetic* = overlaid UI elements like a HUD (Heads-Up Display) that exist outside the game world; *Flow* = state of optimal engagement/challenge balance.

Finding: Shifted imitation learning in the game Hades creates a "Dark Zagreus" boss that mimics the player's previous successful run, achieving personalized dynamic difficulty adjustment (DDA) without scripted behavior.

Why it matters: Directly relevant to game AI design, adaptive difficulty, and player engagement mechanics—useful for teaching gamification and games design, particularly how AI can personalize challenge to sustain motivation and engagement.

Method: Two-week user study (n=20) comparing scripted vs. imitation-based AI boss behaviors, with both quantitative (player experience metrics) and qualitative measures.

Jargon: *Imitation learning* — AI trained to replicate observed behavior rather than optimized via reward signals; *Dynamic difficulty adjustment (DDA)* — automatically tuning game challenge to match player skill in real time; *Roguelike* — a game genre with procedurally generated levels and permadeath.

Finding: Video-based prompting interventions in blended learning improved self-regulated learning (SRL) skills and knowledge outcomes, but only for part-time students with lower entrance scores—not for other learners despite identical interventions.

Why it matters: Directly relevant to learning sciences and instructional design, particularly how technology-mediated scaffolding (prompting via video) supports goal-setting, time management, and self-evaluation—core SRL domains the professor teaches.

Method: Quasi-experimental design with 215 undergraduates; COPES model (Conditions, Operations, Products, Evaluations, Standards) used to structure prompt videos in a persuasion course.

Jargon: *Self-regulated learning (SRL)*: learners' ability to monitor and control their own learning processes. *COPES model*: a framework describing how learners set conditions, apply operations, produce products, evaluate outcomes against standards. *Blended learning*: mix of in-person and online/technology-mediated instruction.

Finding: Bibliometric analysis of 2,324 gamification-in-education papers (2000–2025) identifies four research clusters: student engagement/instructional design, medical education, immersive tech (AR/VR), and corporate/workforce development, with the last cluster showing rapid recent growth.

Why it matters: Provides a landscape map of gamification research relevant to teaching about games, gamification, motivation, and learning sciences, and flags corporate/workforce gamification as an emerging area worth tracking.

Method: Bibliometric analysis using VOSviewer on SCOPUS data — useful for identifying research trends but produces no new empirical findings about gamification effectiveness.

Jargon: *Bibliometric analysis* = quantitative mapping of publication patterns, citations, and co-authorship; *VOSviewer* = software for visualizing bibliometric networks; *SDG 4* = UN Sustainable Development Goal 4 (Quality Education).

Finding: Players at high risk of extreme social withdrawal (hikikomori) prefer cooperative over competitive gameplay, dislike negative social interactions, and favor story-driven and strategic game elements, suggesting specific design features for game-based interventions.

Why it matters: Directly informs serious game design and gamification for vulnerable populations, with implications for behavior change design and persuasive technology targeting social anxiety and isolation.

Method: Survey (N=529) with inductive thematic analysis of open-ended responses, with participants categorized by a validated hikikomori risk scale.

Jargon: *Hikikomori* — Japanese term for extreme social withdrawal/isolation, often lasting months or years; *NHR scale* — NEET-Hikikomori Risk scale, a validated instrument measuring risk of this condition; *NEET* — Not in Education, Employment, or Training.

Finding: A multimodal instructional design integrating prominent sign language visuals and Universal Design for Learning principles significantly improved digital media literacy acquisition and short-term retention in deaf/hard-of-hearing learners by reducing split-attention effects.

Why it matters: Directly applies Cognitive Load Theory and UDL to instructional design, relevant to learning sciences and HCI accessibility, with implications for designing cognitively equitable digital learning environments.

Method: One-group quasi-experimental pre/post design (n=30) with semi-structured interviews; no control group limits causal claims.

Jargon: *Split-attention effect* — cognitive load increase when learners must mentally integrate spatially/temporally separated information sources; *UDL (Universal Design for Learning)* — framework for flexible instructional design accommodating diverse learners; *Extraneous cognitive load* — mental effort caused by poor instructional design rather than the material itself.

Finding: Proposes "cognitive algovigilance" as a framework for monitoring how clinical AI systems affect human decision-making, distinguishing cognitive augmentation from cognitive toxicity (e.g., automation bias, deskilling, alert fatigue).

Why it matters: The concepts—automation bias, deskilling, trust miscalibration, and cognitive dependence—are directly relevant to AI's impact on human collaboration and decision-making, but the clinical framing limits applicability to the professor's core teaching areas.

Jargon: *Cognitive algovigilance*: systematic surveillance of cognitive effects from human-AI interaction; *Cognitive Adverse Events*: AI-mediated changes in decision-making that increase error probability; *verification decay*: gradual reduction in humans' tendency to check AI outputs over time.

Finding: A distributed IoT system was developed to study how humans (particularly aging populations) discover and learn multimodal interactions with sensor-actuator "widgets" that have no predefined function or shape.

Why it matters: Tangentially relevant to HCI and exploratory interaction design, but the focus is narrow engineering/behavioral measurement infrastructure rather than insights about cognition, motivation, learning, or design.

Method: IoT widgets connected via MQTT protocol with interaction sequences modeled as Markov chains (stochastic state-transition matrices governing sensor-actuator reward relationships).

Jargon: MQTT = lightweight messaging protocol for IoT devices; Markov chains = mathematical models where next state depends only on current state, used here to create deterministic or random interaction sequences.