CHCI@VT Research Shines at 2025 HFES International Annual Meeting

A PRISMA-Based Review of Empathic Intervention in Driving Contexts

Mungyeong Choe, Myounghoon Jeon

Driving is an emotionally charged activity where both positive and negative emotions can impact safety and performance. Empathic interventions, particularly those utilizing in-vehicle agents, have emerged as a promising approach to regulate drivers’ emotions and enhance road safety. Utilizing the PRISMA framework, we systematically reviewed studies from various databases and additional sources, ultimately including 16 papers for in-depth analysis. Brief highlights from these works underscore the range of methods employed—from voice-based dialogues to non-verbal cues—and the importance of aligning intervention timing and modality with driver affect. This synthesis provides a foundation for understanding current practices and identifies gaps for further exploration. By illuminating how empathic techniques can be integrated into automotive systems, this research contributes to the design of safer, more supportive in-vehicle technologies and lays the groundwork for advancing driver-centered innovation in the field.

An Investigation of Look and Pick Cues for Guidance in Dense Operating Environments

Jacob Belga, Ryan Ghamandi, Nayan Chawla, Ryan McKendrick, Ryan P. McMahan, Joseph J. LaViola

Interaction cues, which inform users about potential actions to take, are common to many types of extended reality applications. While numerous studies have compared individual interaction cues, few studies have investigated combinations of interaction cues for complex tasks, particularly in contexts involving high counts of similar actions. In this paper, we present a within-subject study (n = 48) investigating the effects of interaction cue combinations for guiding users through in-cockpit procedures for a UH-60 Black Hawk helicopter. The results of our study indicate significant effects for different interaction cue combinations for perceived mental effort and discomfort. Furthermore, the completion time results in our study contradict previous results pertaining to the effectiveness of specific interaction cues, which we believe are due to the context and complexity of the real-world application underlying our study. These results imply that Look Arrow and Pick Arrow are the optimal interaction cue combination for scenarios that involve dense operating environments. Our results also reveal that more research on interaction cues needs to be conducted in a broader diversity of contexts and applications to better understand effective interaction cues and combinations in different scenarios.

Beyond the Table: Anchored Positional Responding Approach for Identifying and Measuring Inherent Issues in Individual Decision-Making during Focus Groups

Saeid Alimoradi, Rafael N. C. Patrick, Deborah E. Dickerson

Focus Groups are often prone to adverse group effects, such as conformity and dominance, which can negatively impact data validity and trustworthiness. In this study, we introduce an alternative approach – Anchored Positional Responding (APR) – that utilizes the Anchoring Effect to identify adverse group effects. In the APR approach, participants are first required to physically position themselves within a room-sized replicated Likert Scale (as external anchors) in response to a given question prompt, then proceed to respond verbally.

Changes in Visual Strategies during Psychomotor Skill Training in Laparoscopic Surgery: A Longitudinal Eye Tracking Study

Shiyu Deng, Jinwoo Oh, Nathan Lau, Sarah Parker

This study investigates the evolution of gaze behaviors during the acquisition of psychomotor skills required for laparoscopic surgery, with a focus on both how the gaze behaviors change over time and how they differ between individuals who achieve proficiency early and those who do not, using scene-dependent eye metrics. We conducted a longitudinal study where participants with no prior laparoscopic experience practiced the peg transfer task in the Fundamentals of Laparoscopic Surgery (FLS) curriculum until they either met FLS proficiency criteria or completed a maximum of six one-hour training sessions. The hypothesis is that as performance improves, gaze behavior will gradually shift from focusing primarily on the object currently being manipulated (i.e., confirmatory or feedback gaze behavior) to the future target location (i.e., anticipatory or feedforward gaze behavior). Further, individuals who achieve proficiency earlier are expected to adopt this feedforward visual strategy more quickly than those requiring more training sessions or those unable to reach proficiency.

Classifying Diverse Manual Material Handling Tasks Using Vision Transformers and Recurrent Neural Networks

Mohammad Sadra Rajabi, Aanuoluwapo Ojelade, Sunwook Kim, Maury Nussbaum

Frequent or prolonged manual material handling (MMH) is a major risk factor for work-related musculoskeletal disorders, which cause considerable health and economic burdens. Assessing physical exposures is essential for identifying high-risk tasks and implementing targeted ergonomic interventions. However, variability in MMH task performance across individuals and work settings complicates physical exposure assessments. Further, conventional tools often suffer from limitations such as bias, discomfort, behavioral interference, and high costs. Non-contact (ambient) methods and automated data collection and analysis present promising alternatives for assessing physical exposure. We investigated the use of vision transformers and recurrent neural networks for non-contact MMH task classification using RGB video for eight simulated MMH tasks. Spatial features were extracted using the Contrastive Language-Image Pre-training vision transformer, then classified by a Bidirectional Long Short-Term Memory model to capture temporal dependencies between video frames. Our model achieved a mean accuracy of 88% in classifying MMH tasks, demonstrating comparable performance to methods using depth cameras or wearable sensors, while potentially offering better scalability and feasibility for real environments. Future work includes improving temporal modeling, integrating task-adapted feature extraction, and validating across more diverse workers and occupational environments.

Comparison of Musculoskeletal Model Estimates of Exoskeleton Effects on Spine Loads During Lifting Using OpenSim vs. the AnyBody Modeling System

Mohamad Behjati Ashtiani, Mohammadhossein Akhavanfar, Lingyu Li, Sunwook Kim, Maury Nussbaum

Back-support exoskeletons (BSEs) are designed to reduce physical stress during lifting by supplying assistive torques, however influences on spinal loading remains unclear. We used two widely applied musculoskeletal modeling platforms—OpenSim and the AnyBody Modeling System (AMS)—to estimate intervertebral joint forces (IJFs) during symmetric and asymmetric lifting tasks performed with and without BSEs. We analyzed data from 18 participants who completed repetitive lifting/lowering across three task conditions and four intervention conditions (three BSEs and a control without a BSE). Simulations were conducted using both platforms to estimate axial compression and anteroposterior shear forces at the L4/L5 level, with peak values (95th percentile) extracted for comparison. OpenSim produced consistently higher estimates of both axial and shear forces compared to AMS. While both models indicated that BSE use reduced spine loading, the magnitude of reduction varied, with OpenSim generally showing larger effects. Axial compression estimates were strongly correlated between models (r > 0.95), but shear force estimates showed weaker and sometimes negative correlations, particularly during asymmetric tasks. These discrepancies likely stem from differences in modeling assumptions and the treatment of passive structures. Future work should incorporate refined interaction models and validate IJF estimates with in vivo data such as electromyography.

Decoding Driving Words: A Holistic Exploration of Older Drivers’ Early-stage ADAS Adaptation Using Structural Topic Modeling on Naturalistic Driving Data

Dan Liang, Nathan Lau, Jon Antin

Advanced Driver Assistance Systems (ADAS), such as Adaptive Cruise Control (ACC) and Lane Keeping Assist (LKA), offer potential safety benefits for older drivers. The early adaptation phase is crucial for both long-term adoption and appropriate use of ADAS. This study applied Symbolic Aggregate Approximation and Structural Topic Modeling (STM) to naturalistic driving data to examine behavioral and perceptual adaptation. STM assessed how topic prevalence related to weeks of ADAS exposure and weekly median trust and satisfaction ratings for ACC and LKA.

The results revealed (1) increasing prevalence of active steering inputs on highway is associated with continued exposure to ADAS presence in weeks, (2) decreasing prevalence of active speed adjustments on highway is associated with higher ACC trust and satisfaction, (3) increasing prevalence of active steering inputs in mixed speed environments is associated with higher ACC trust and satisfaction rating, (4) decreasing prevalence of active speed adjustment on highway is associated with higher LKA trust.

These findings offer insights for designing policies and training to help older novice drivers integrate ADAS features effectively, foster appropriate system use, and develop accurate mental models of ADAS capabilities and limits.

Effects of Two Passive Back-Support Exoskeletons on Physical Demands, Discomfort, Perceived Exertion, and Usability in Simulated Manual Mining Tasks

Feyisayo Akinwande, Sunwook Kim, Maury A. Nussbaum

Work-related musculoskeletal disorders (WMSDs) remain a major health concern in the mining industry, with the back being the area most commonly affected. Given that back-support exoskeletons (BSEs) have the potential to prevent back WMSDs, we investigated the effectiveness of two passive BSEs (soft vs. rigid) during mining-relevant tasks compared to a control condition (no BSE). Participants (n=18) performed cable hanging/installation, core box lifting, and overhead wire mesh installation tasks under different task conditions. Using both BSEs significantly reduced activity in the left trunk extensor muscles during core box lifting and overhead wire mesh installation tasks. The rigid BSE increased discomfort in the waist and thighs during the cable hanging/installation task, increased discomfort in the chest and waist during the core box lifting task, and increased discomfort in the waist during the overhead wire mesh installation task. In contrast, the soft BSE reduced perceived exertion in the upper back during all tasks. Moreover, participants rated the soft BSE more favorably in terms of usability and perceived balance. While both BSEs showed promising results for potential WMSD prevention, future work is recommended to assess these designs during more prolonged use among mining workers, especially given the discomfort reported with the rigid BSE.

Evaluating the Accuracy of AI-Powered Ergonomic Assessments Using a Commercial Computer Vision System

Saman Jamshid Nezhad Zahabi, Sunwook Kim, Maury A. Nussbaum, Sol Lim

Workers engaged in material handling tasks face a high risk of work-related musculoskeletal disorders (WMSDs). While recent AI-driven computer vision systems claim to assess ergonomic risk factors with minimal manual input, their accuracy remains largely unverified. We evaluated the accuracy of a commercial AI system in estimating key parameters of the Revised NIOSH Lifting Equation (RNLE) by comparing its outputs to those obtained from marker-based motion capture data (ground truth). Ten sex-balanced participants performed various lifting tasks, during which their movements were recorded using three cameras positioned at different angles. Simultaneously, 3D motion capture data were collected for comparison. The video recordings were uploaded to the commercial AI software, which analyzed the footage and extracted RNLE parameters. These outputs were then compared to the motion capture data to assess the AI system’s accuracy. Results revealed notable inaccuracies in the commercial system’s estimates, particularly for horizontal and vertical distances. These errors led to overestimated Recommended Weight Limits (RWL) and underestimated Lifting Index (LI) values. Among the three cameras, the side view yielded the most accurate results, while the moving camera produced the least reliable estimates. These findings suggest that current commercial AI-based ergonomic tools require substantial improvement before reliable workplace use.

Evaluating Rigid and Soft Back-Support Exoskeletons for Roofing: Biomechanical Benefits and Trade-Offs

Jiwon Choi, Sunwook Kim, Ahmad Raza Usmani, Alan Barr, Carisa Harris-Adamson, Maury A. Nussbaum

Roofing is among the most physically demanding construction trades, associated with high rates of low back injuries. Passive back-support exoskeletons (BSEs) have been proposed as a potential intervention to reduce physical demands, yet little is known about the effectiveness of different BSE types (rigid vs. soft) for roofing work. Eighteen participants performed lab-based simulations of roofing tasks under 12 different conditions. These conditions included two types of BSEs (rigid and soft) and a control (No BSE), task locations (middle row and bottom row of a mock roof), and roof slopes (18° and 26°). Outcome measures included lumbar muscle activation, lumbar kinematics, and ratings of perceived exertion and discomfort. Using the rigid BSE significantly reduced lumbar muscle activity (11–17%), lateral bending (~18%), and axial rotation (~15%), whereas using the soft BSE significantly decreased lumbar flexion (~9%) and axial rotation (~16%) compared to No BSE. Both BSEs reduced perceived exertion and discomfort in the low back (~16%); however, the rigid BSE increased leg discomfort (~26%), and the soft BSE increased shoulder exertion (~19%). Our results suggest that using BSEs can be beneficial in roofing tasks, but also highlight the importance of balancing biomechanical benefits with device-specific trade-offs to ensure effective application.

Humanizing In-Vehicle Virtual Agents: A Systematic Review of Anthropomorphism and Its Impacts

Gayoung Ban, Myounghoon Jeon

As cars become smarter, many now include in-vehicle virtual agents (IVAs) designed to help drivers stay informed and engaged. These agents often have humanlike features—a concept known as anthropomorphism. But does giving a car a “personality” actually improve the driving experience? This study systematically reviews research on how anthropomorphic features (like faces, voices, and emotions) influence trust, engagement, and driver performance across different driving environments. By examining existing studies, we reveal that while humanlike agents can improve trust and user satisfaction—especially in highly automated vehicles—there are also risks, such as overreliance or reduced attention. The findings highlight the need for more consistent definitions, better evaluation methods, and careful design to ensure these technologies enhance safety. This review provides a roadmap for future research and more intuitive in-vehicle technologies.

Investigating Sensitivity to Vibrotactile Stimulation in Isometric and Dynamic Tasks: Implications for Ergonomic Intervention

Mahdis Tajdari, Xiang Yang, Md Shafiqul Islam, Sol Lim

Work-related musculoskeletal disorders (WMSDs) and safety hazards are persistent concerns in physically demanding environments such as construction. While traditional ergonomic interventions often rely on visual or auditory feedback, these modalities can be compromised by environmental noise or visual obstructions. Vibrotactile feedback systems offer a promising alternative for both posture correction and real-time hazard awareness. This study evaluated vibrotactile sensitivity and spatial acuity during static and dynamic tasks to guide effective actuator placement. Participants received vibrations on the upper arm, lower back, upper back, and thigh while exerting force or simulating common construction tasks. Findings indicate that the lower back is a poor site for tactile feedback due to low sensitivity and acuity. Stronger vibrations are needed on the thigh during force exertion to ensure detectability, while the upper arm supports high perceptual accuracy, even during dynamic tasks. These insights support anatomically-informed, task-adaptive vibrotactile designs for improving workplace safety and injury prevention.

The Quality of Model-Estimated Shoulder Muscle Activity During Overhead Work Varies with Task Demands and Exoskeleton Use

Lingyu Li, Mohamad Behjati Ashtiani, Sunwook Kim, Maury Nussbaum

Passive arm-support exoskeletons (ASEs) can alleviate shoulder stress during overhead tasks. However, the effects of ASEs vary by device and task, and existing evaluation protocols remain time-consuming and resource-intensive. Musculoskeletal models could reduce dependence on electromyography (EMG) measures by predicting muscle activation. However, it is unknown whether model performance is sufficient or consistent when an ASE is used across different task conditions. We evaluated model-estimated shoulder muscle activity from a commercial biomechanical model during dynamic overhead push tasks at three over-shoulder heights, in both the forward and upward directions, both with and without an ASE. Kinematics and hand-force data were input into the AnyBody Modeling System to estimate shoulder muscle activity. Model outputs were then compared to normalized EMG using pattern-similarity and magnitude-difference metrics. Overall, model performance closely matched normalized EMG patterns at low arm-elevation angles, but performance decreased at larger elevation angles and declined further when an ASE was included. These outcomes suggest that model-estimated shoulder muscle activity is reasonably accurate under certain conditions. However, model improvements are needed to ensure adequate performance across a broader range of tasks.

Understanding Cybersecurity Skill Levels through Psychological Measures: Clustering Hackers with Traits Questionnaires

Jinwoo Oh, Po-Yu Chen, Hsiang-Wen Hsing, Nathan Lau, Peggy Wu, Kunal Srivastava, Nikolos Gurney, Kylie Molinaro, Stoney Trent

In the evolving field of cybersecurity, red-team or attack exercises provide invaluable insight into attackers' skills and behaviors. These insights can extend beyond technical ability, including cognitive behaviors, problem-solving approaches, decision-making under pressure, and adaptability. However, participant recruitment for large representative samples of hackers remains a challenge, limiting the discoveries of the psychological traits associated with red-team proficiency. This paper presents an exploratory study on psychological profiles of cybersecurity professionals and hobbyists recruited for a representative study in a Kali Linux environment by clustering results from psychometric questionnaires to correlate traits with basic hacking skills.

Details of Posters

Assessing the Effectiveness of Vulnerable Road User Alerts via Personal Listening Devices in an Immersive Crosswalk Environment

Esha Mahendran, Rafael N. C. Patrick

The study explores the use of personal listening devices (PLDs) as navigational aids to deliver safety alerts to vulnerable road users (VRUs) in a virtual reality environment. With the rise of electric vehicles (EVs), pedestrians face increased risk as EVs emit significantly less noise than conventional vehicles, reducing the auditory cues typically used to judge safe crossing opportunities. Thirty-one participants completed 24 simulated crosswalk trials in a virtual reality environment created with Unreal Engine 4.26. The scenario replicated a campus street with two vehicles and a safe crossing gap. Participants received verbal and dissonant auditory alerts through both open-ear and closed-ear headphones. An experimental platform, the Tesseract, generated realistic street noise and integrated it with the VR scene, while Cycling ’74 Max software controlled auditory delivery. Visual conditions alternated between day (with cues) and night (without cues), altering visibility of the crosswalk. Findings demonstrate how auditory alerts, delivered via air and bone conduction headphones, influenced pedestrian reaction and crossing behavior. This research provides insight into designing effective PLD-based V2P systems, supporting urban planners, safety regulators, and developers in enhancing pedestrian safety in EV-dominated environments.

Collaboration Dynamics in Neurodiverse Teams During Simulated Remote Assembly: An Exploratory Study

Jacqueline Elise Bruen, Manhua Wang, Megan Fok, Lili Guan, Thomas J. Shaw, Caroline Byrd Hornburg, Angela Scarpa, Sunwook Kim, Myounghoon Jeon

Employment is vital for adult independence, yet autistic individuals face substantial employment barriers. As collaboration is critical in the workplace, this study examined how pairs of individuals (one autistic, one non-autistic) worked together remotely. They performed remote LEGO™ assembly tasks, with assigned roles that created different levels of information access. We recruited 17 pairs; in this preliminary analysis of five pairs, we examined their interaction using behavioral coding across seven teamwork-centric domains. Initial results showed higher collaboration (e.g., flow, mutual understanding, knowledge exchange) when information was restricted to one partner ('Engineer' role). Problem-solving communication (coded as 'Argument') also increased under these conditions, especially when the autistic partner held the restricted information. Collaboration patterns were otherwise similar regardless of role assignment. These findings suggest that task structure influences interaction, offering insights into designing tasks to leverage neurodiversity strengths in the workplace.

Driving with Empathy: Workshop Report on AI-driven In-vehicle Empathic Agent Design for Automated Vehicles

Mungyeong Choe, Jiayuan Dong, Esther Bosch, Ignacio Alvarez, Michael Oehl, Christophe Jallais, Areen Alsaid, Myounghoon Jeon

The rapid development of automated vehicles offers promising avenues to enhance user experience and safety by integrating empathic in-vehicle interfaces. The report from the workshop [Hidden for review] series, held during the [Hidden for review], investigates the potential of employing generative artificial intelligence (AI) to develop these empathic interfaces. Our workshop emphasized the importance of emotion recognition and regulation for passengers in driving contexts and the role of future in-vehicle empathic interfaces. Through collaborative design sessions, participants developed personalized in-vehicle agents to address drivers' emotional states to improve safety and enjoyment. We explored insights by investigating the potential and possibilities of using generative AI in the design process. This report discusses the significance of empathic displays in facilitating the adoption of automated vehicles and explores the potential and limitations of using generative AI in the design phases.

Estimating 3D Dynamic External Hand Forces From Markerless Motion Capture Using Attention-Based Recurrent Neural Networks

Aanuoluwapo Ojelade, Mohammad Sadra Rajabi, Sunwook Kim, Maury A Nussbaum

Physical exposure assessment are essential for identifying high-risk tasks, guiding the development and assessment of ergonomic interventions, and enhancing our understanding of exposure-risk relationships. Yet, accurately assessing three-dimensional, dynamic hand forces for physical exposure is challenging due to the need for specialized equipment. We explored using data from markerless motion capture systems to predict hand forces during two-handed manual material handling tasks, using attention-based deep learning architecture. Model architectures included an input layer, recurrent neural network layers, an attention layer, dropout functions, and output layers.

Our results were encouraging overall, but predictions were less accurate in both the proximal-distal and anterior-posterior directions during box pushing and pulling tasks. Overall, our findings indicate that the proposed approach has the potential to predict dynamic external hand forces, without direct measurement (e.g., load cell, instrumented gloves), offering a balance of simplicity and non-intrusiveness in quantifying physical exposure. Future work is needed, though, to improve the performance of the model in predicting hand forces during push and pull task conditions.

Factors Affecting the Integration of Lift Assists in Automotive Assembly: Operator Perspectives

Ahmad Raza Usmani, Sunwook Kim, Marty Smets, Maury Nussbaum

Automotive assembly workers face high risks of low back pain (LBP) due to frequent material handling (MH) tasks. Lift assists (LAs) potentially reduce these risks by offsetting gravitational forces while allowing operator control. However, integrating LAs remains challenging, and operator perspectives remain largely unexplored. We interviewed 16 automotive assembly operators to identify challenges, limitations, and preferences related to LA use. Content analysis revealed seven themes, grouped into two categories: Usability Factors and Concerns. Usability Factors included Hand Controls, Operational Functions, Physical Design, Workflow Efficiency, and Stakeholder Integration. Operators highlighted the need for fewer buttons with ergonomic controls, efficient maneuverability, and improved part manipulation. Training was critical for developing proficiency. Bulky and heavy LAs were criticized for increased physical demands, and poor maintenance disrupted workflows. Concerns included safety risks—such as falling payloads, pinch points, and obstruction hazards—and physical exertion in the back, shoulder, and lower limbs. Our findings suggest that improving LA usability requires early collaboration among engineers, ergonomists, designers, and operators, with emphasis on simple hand controls, smooth maneuverability, and light-weight designs. Comprehensive training and regular maintenance are also essential. Future research will quantify the effects of different LAs and task factors on physical demands and workflow efficiency.

How Does an In-Vehicle Agent Accent Influence Driver Behavior and Perception?

Mungyeong Choe, Nicholas Kilp, Abby Walker, Koeun Choi, Myounghoon Jeon

Accents convey subtle social and cultural cues, raising the question of whether they also influence human–machine interactions in critical settings such as driving. In this study, we employed a 3 × 3 mixed design, with linguistic background (Southern U.S., non-Southern U.S., international) as a between-subjects factor and accent condition (Standard U.S. English, Southern U.S. English, Navigation-Only) as a within-subjects factor. A total of 36 participants completed a series of driving scenarios in a motion-based driving simulator while exposed to in-vehicle agents with varied accents. Objective measures indicated that the Southern accent condition yielded more stable driving performance. Subjective evaluations were also more favorable in the Southern accent condition compared to the two other conditions. Overall, these findings underscore the potential benefits of incorporating regionally distinctive voice accents into in-vehicle agents to enhance user engagement, improve driving stability, and ultimately promote safer driving behavior.

How to Design VR Training for Learning Basketball Tactics-Perspective From Users With Team Basketball Experience

Hsiang-Wen Hsing, Nathan Lau, Po-Yu Chen

This study explored how organized team basketball players can potentially learn basketball tactics in Virtual Reality (VR). We conducted semi-structured interviews with 13 participants, who had prior organized team basketball experience, to gather data on how they currently learned basketball tactics and feedback on how they like learning a pick-and-roll scenario with a VR training prototype under different user perspectives and graphical fidelity levels. Thematic analysis of the interview transcripts revealed several considerations in designing VR environments for learning basketball tactics. Offense is free-flowing and less reliant on knowing defensive strategies, while defense requires a deep understanding of offensive tactics. Players perceive egocentric (first-person) and exocentric (third-person) perspectives to have different benefits. Egocentric view enhances immersion whereas exocentric view enhances spatial awareness. Simplistic VR avatar of players help focused attention, while realistic VR avatars supported decision-making. Clear understanding of player roles is crucial for team coordination while confusion and spatial awareness hinder performance. Participants noted that VR could complement current training methods but adoption may be limited due to familiarity. These findings suggest that design of VR basketball training should support multiple user perspectives and graphical fidelities to learn various player roles effectively.

Impact of Negative Emotion, and Social Influence on Pre-evacuation Decision-making

Johnson Adetooto, Seonho Woo, Myounghoon Jeon, Young-Jun Son, Behzad Esmaeili

Pre-evacuation delays emerge when fear, cognitive anchoring, and social cues disrupt attention to hazard signals, creating a gap between threat awareness and action. This laboratory study investigates how these factors influence evacuation behavior by combining fear induction (written recall of past experiences), eye-tracking measurement, and controlled social influence. Participants (N = 22) engaged in an incentivized task before an unannounced fire alarm, while confederates either evacuated or stayed. Results revealed: (1) social influence strongly shaped decisions, with participants less likely to evacuate when confederates stayed; (2) task anchoring delayed alarm recognition; and (3) although fear levels rose significantly after induction, fear did not significantly predict evacuation behavior. Eye-tracking analysis showed that evacuators had fewer fixations and saccades, suggesting quicker visual disengagement from the task and faster attention to safety-relevant cues. These findings indicate that evacuation is shaped more by attentional dynamics and social context than by emotional arousal alone. Practically, the study supports using dynamic alert systems to interrupt attentional tunneling and incorporating training that addresses the influence of social cues during emergencies.

Surveying Perinatal Mental Health Challenges: Insights for a Remote Intervention System

Sakshi Taori, Sunwook Kim, Sol Lim

Perinatal mood and anxiety disorders (PMADs) adversely affect both maternal health and child’s psychological development. These challenges are particularly pronounced among racial/ethnic minorities, who often face disparities receiving diagnosis and care. Remote intervention systems have gained traction in providing mental health support, however, current systems fail to address women’s changing needs throughout the entire perinatal period. This study aimed to inform the development of a personalized remote intervention system by surveying 31 Black or African American women in different stages of the perinatal period. Participants reported the highest PMAD symptoms during the second and third trimesters, citing stressors such as hormonal changes, financial burden, and lack of family support. Remote support was most preferred during pregnancy due to accessibility, flexibility, and reduced costs. Key features for the intervention system included communication with healthcare providers, therapy access, positive affirmations, symptom monitoring, and time management tools. Barriers to in-person care included stigma, time constraints, and distrust in healthcare providers. Findings highlight the need for a user-centered and flexible remote intervention platform tailored to the evolving needs of women throughout the perinatal period. Future work will involve prototyping such a system to enhance mental health outcomes and reduce disparities in PMAD care.

Understanding the Context of Sit-Stand Desk Usage for Promoting Healthy and Productive Behaviors

Junghoon Chung, Sol Lim, Donghan Hu, Daniel Vargas Diaz, Sang Won Lee

Understanding the context of sit-stand desk usage is crucial for fostering healthier and more productive workplace behaviors. We conducted a 15-day study involving ten graduate students engaged in desk-based knowledge work, collecting comprehensive physical, work-related, and personal context data using desk height logs, screen activity tracking, and self-reported intention metrics. Our analysis revealed no significant differences in standing ratios across productive, neutral, and distracting work states, suggesting personal routines strongly influence posture choices. Notifications significantly increased standing durations without compromising productivity, demonstrating the effectiveness of timely prompts. Utilizing an XGBoost regression model, we successfully imputed sparse intention data with high reliability, highlighting physical and personal contexts as key predictors. This research provides valuable insights and methodologies for developing intelligent, context-aware posture interventions. By integrating quantitative findings with qualitative feedback, we offer a framework for future ergonomic systems aimed at optimizing health benefits and workplace productivity.

Details of Panels

Exploiting Cognitive Biases in Cyber Adversaries: Insights, Challenges, and Future Directions for Human-Centred Cyber Defenses

Palvi Aggarwal, Cleotilde Gonzalez, Nathan Lau, Robert Gutzwiller, Scott Brown, Prashanth Rajivan, Kimberly Ferguson-Walter

Cyberattacks are increasing at an unprecedented rate, becoming more sophisticated and human driven. For decades, cybersecurity has been predominantly techno-centric, relying on automated solutions to detect and prevent threats. These methods typically focus on securing systems, identifying vulnerabilities, and recognizing known attack patterns. However, attackers continuously innovate to bypass these defenses, exploiting even the smallest security gaps.

Beyond technological exploits, adversaries increasingly manipulate human factors to bypass security measures. Instead of relying solely on technical vulnerabilities, they exploit cognitive weaknesses, social engineering tactics, and psychological manipulation to gain access to systems. Over the past decade, the Human Factors community focused towards understanding the role of human cognition in cybersecurity. Research has explored situational awareness, individual and team cognition, and more recently, psychological defenses against cyber adversaries. More recent studies delve into leveraging cognitive biases—turning attackers' own mental shortcuts against them to degrade their decision-making and operational effectiveness. As theories, frameworks, and methodologies evolve, so do the challenges in implementing these innovative defense techniques in practice. This expert panel brings together leading scholars to discuss their perspectives on the opportunities and challenges in triggering and measuring cognitive vulnerabilities in attackers as effective methods for cyberdefense.

Details of Lightning Lessons

Traveling from Fiction to Future : Ethical Design Principles for AI-Integrated Extended Reality Workplaces

Esha Mahendran

This lightning lesson examines how popular media can inform the ethical design of AI-integrated extended reality (XR) workplaces. Insights are drawn from the anime Psycho-Pass, the video game Detroit: Become Human, and the television series Black Mirror, each illustrating emotionally responsive and adaptive technologies. These portrayals are translated into design strategies for context adaptivity, multimodal feedback, and embodied cognitive allies, reframing dystopian warnings into constructive ethical guidelines. The resulting framework envisions XR workplaces that are adaptive, inclusive, and trustworthy, ensuring that technology supports human agency, cognition, and well-being.